Editor's note: This story has been updated.
First: The facility's main concrete spillway suffers serious damage, resulting in erosion of the rock beneath it -- and potentially threatening the safety of the dam itself.
Second: Water fills Lake Oroville, the gigantic reservoir behind the dam, and begins surging down a steep unpaved hillside that's meant to serve as an emergency spillway. The slope suffers serious erosion, again potentially threatening the dam's safety.
These scenes will sound familiar to anyone who followed the crisis at Oroville, which began to unfold six months ago this week. In fact, you may think you watched them happen.
That finding, apparently based on inaccurate or incomplete information from the California Department of Water Resources about the condition of the dam's spillways and the strength of the rock they were built on, has led some dam safety experts and DWR critics to question both the evidence the safety board reviewed and whether the review process itself is effective.
The study -- called a potential failure mode analysis, or PFMA -- was performed in 2014 as part of what's called a Part 12D safety review, a title that refers to a Federal Energy Regulatory Commission regulation that requires independent safety inspections for most large dams every five years.
The 2014 PFMA declined further study of 31 "candidate" possible failure modes, including the two that outlined possible spillway failures.
Although the details of the evidence analyzed for the spillway failure scenarios is not publicly available, Department of Water Resources documents KQED obtained under the California Public Records Act and other recently released records give a clear indication of why the experts decided to shelve them.
In dismissing the possibility that the main spillway might be lost, a PFMA summary said the 3,000-foot-long concrete chute "is in good condition and the underlying rock is very competent. ... It is seen as highly unlikely that the concrete lining will fail and highly unlikely that significant erosion of the rock will occur during one spilling event."
The summary noted the spillway chute had undergone "minor repairs" in 2009.
The experts declined further study of the emergency spillway scenario -- which contemplated an overflow as much as 25 times greater than the one that actually occurred in February -- because of "the rock between the Feather River and emergency spillway is very competent and resistant to erosion."
"Competent" rock refers to rock that is only lightly weathered and is structurally sound and resistant to erosion. But despite the declarations in the PFMA summary, the February near-disaster showed that extensive areas of the rock underlying the concrete spillway and the hillside serving as the emergency spillway were incompetent and highly erodible.
The PFMA's assumptions about the main spillway, too, are called into question by records that show DWR was well aware the chute's concrete surface was prone to chronic cracking and that water flowed through numerous flaws in the chute floor and undermined parts of the structure.
The records show the spillway underwent top-to-bottom repairs on at least three occasions, including:
- 1977, when it was less than a decade old.
- 1997, after record releases caused what DWR described as "distress throughout the length of the spillway" that required issuance of an emergency contract for repair.
- 2009, when what the PFMA summary describes as "minor repairs" involved repairing cracks, exposed rebar, eroded concrete, broken subsurface drains and voids that had opened under the chute's slab.
The four outside consultants hired to conduct the 2014 Part 12D safety review, including the PFMA process, did not respond to emailed requests for comment.
Department of Water Resources spokeswoman Erin Mellon said in an email Monday the evidence presented for the potential failure mode analysis included a site inspection and historical documents including past geologic studies of the area.
While not offering specifics about the data considered, Mellon said "the information gathered showed the spillway was in good condition and the rock beneath it was extremely competent."
J. David Rogers, a professor of geologic engineering at the Missouri University of Science and Technology who has visited the Oroville facility since the February crisis, said he was startled by the description of the rock under the spillways as being sound.
"The principal issue with performance of the service and emergency spillways has been the poor quality of the underlying 'rock,' " Rogers said in an email. He characterized the material as "deeply weathered ... essentially with the consistency of stiff soil, not rock. The depth of weathering varies from 20 to 80 vertical feet, based on simple visual inspection."
Rogers said the review panel would not have had "any easy means" of determining the true character of the material underlying the spillways -- except perhaps for the appearance of "apparently competent" bedrock exposed at the top of the spillway adjacent to the massive control structure that houses a set of eight huge floodgates.
But the appearance of that apparently sound rock might have been misleading, Rogers said, because crews had to remove 70 feet of material to reach it. Most of the rest of the spillway, he said, was excavated no more than 20 or 25 feet deep.
That left masses of weaker material in place under the spillway. When the concrete chute breached during moderate flows in February, he said, "the rushing water was able to pluck crushed rock and soil" from beneath the spillway, ultimatelhy destroying it.
Emergency Spillway: Theory vs. Reality
The main spillway failure in February came just as a series of powerful, warm storms dumped heavy rain over the Feather River watershed, partially melting a thick blanket of snow that had fallen earlier in the season and triggering a surge of runoff into Lake Oroville. Dam managers responded by ramping up releases down the spillway to about 50,000 cubic feet per second, less than 20 percent of its design capacity and about one-third of its record flow rate in January 1997.
After a large hole appeared in the spillway on Feb. 7, DWR temporarily stopped releasing water down the chute, then resumed limited flows as large sections of the rock beneath the concrete structure rapidly eroded -- contrary to the confident assertions that the material was "very competent."
Meantime, the lake rose toward its rim. On Feb. 11, for the first time in the dam's history, water began pouring over a 1,700-foot-long weir adjacent to the concrete chute and down the hill designated as the dam's emergency spillway. Less than 30 hours later, on the afternoon of Feb. 12, uphill erosion of the type dismissed by the 2014 PFMA had scoured channels as much as 50 feet deep and appeared to be threatening the stability of the weir. The fear of a collapse, which would unleash a catastrophic flood down the Feather River, prompted the emergency evacuation of Oroville and other riverside communities.
That uncontrolled flow in February peaked at about 12,500 cubic feet per second, just a tiny fraction of the 350,000 cubic feet -- about 2.6 million gallons -- per second it was supposed to handle. (The regulated flow over Niagara Falls, by comparison, is 100,000 cfs during "tourist" hours.)
That theoretical maximum flow over Oroville's emergency weir represented what might happen in a worst-case flood scenario, when both spillways would have a single purpose: preventing water from flowing over the top of Oroville Dam, an event that could lead to the structure's collapse.
The Department of Water Resources and FERC had insisted for more than a decade prior to the February emergency -- and through three separate potential failure mode analyses -- that even under the duress of extremely high flows racing down the tree- and brush-covered slope, erosion would be limited to just 1 to 4 feet of surface material.
Erin Mellon, the DWR spokeswoman, on Monday repeated earlier department statements that it's up to an independent forensics team -- appointed in March at FERC's direction -- to analyze the causes of the spillway failure. She didn't respond to a question about why DWR's understanding of the conditions underlying the spillways was so different from actual conditions.
Federal Energy Regulatory Commission spokeswoman Celeste Miller said that a "panel of expert consultants" is reviewing the history and performance of the agency's safety program, including past potential failure analyses, for Oroville Dam.
A Need for Re-evaluation?
Martin McCann, an adjunct professor at Stanford’s College of Civil and Environmental Engineering and director of the college’s National Performance of Dams Program, says the Oroville episode shows a re-evaluation of the way potential failure mode analyses are done may be in order.
"In light of the two regulatory bodies and a responsible owner being involved in the PFMA process, we've got to ask ourselves, 'Are we getting it right?'" McCann said in an interview.
"This stuff is not easy -- it really is quite difficult," he said. "It takes expertise, it takes experience, it takes interpretation and evaluation on the part of experts, and that's very hard. Perhaps the process needs to be re-examined."
David Rogers, the Missouri geoengineering expert, said Oroville could shake up future safety dam reviews.
"Unexpected failures have a way of waking people up and providing a much needed 'reality check' for both the engineers and the politicians," Rogers said.
One of those who agrees is Assemblyman James Gallagher, who represents Oroville. He's proposed legislation, AB 1270, that would require more frequent and tougher dam safety inspections and subject them to periodic review by UC Berkeley's Center for Catastrophic Risk Management.
Gallagher said in an interview last week that the 2014 PFMA decision to decline further study of the possibility of spillway failures "goes to one of the fundamental issues" raised by the Oroville crisis.
"We have independent oversight," Gallagher said. "But the question is, are they asking the right questions? Do they have the right data to properly provide that oversight? ... When something catastrophic happens, do we have enough information and the right processes and procedures in place to handle those situations? And the answer right now is clearly, no, we don't."
This story has been updated to incorporate information from newly released documents on the 2014 Potential Failure Mode Analysis involving Oroville Dam's main and emergency spillways and to include a response from the Federal Energy Regulatory Commission. The story has also been edited to add links to some of the historical documents referenced in the text.