Independent forensic investigations - reflections on Oroville Dam4 October 2018
John W France was the Independent Forensic Team Leader for the investigations into the 2017 Oroville Dam spillway incident. Here he gives a fascinating insight into the pressures and responsibility of such an important task.
In January 2017, I was beginning what I thought would be my first year of semi-retirement with a plan of working about half time. After 42 years of consulting engineering practice, it was time. Two months later, my plans for the year changed drastically when I accepted the assignment to lead the team charged with investigating the Oroville Dam spillway incident.
On 7 February 2017, I, like most of the dam safety community, began to follow the events at Oroville Dam, the tallest dam in the US at 235m. After a section of the service spillway chute failed that day, our community and the public watched the news of the following several days. This culminated in an evacuation order for almost 190,000 people on 12 February because of fears that the emergency spillway structure might fail, releasing the upper 9m or more of the stored reservoir water and flooding the downstream area.
On 13 February, one day after the evacuation order, the Federal Energy Regulatory Commission (FERC), one of the dam’s two regulators, ordered the dam’s owner, the California Department of Water Resources to “perform a forensic analysis aimed at determining the cause of the chute failure and ascertaining if the failure mode could occur again.” FERC required that the investigation be performed by individuals with no previous involvement in assessing the spillway structure.
DWR asked the Association of State Dam Safety Officials (ASDSO) and the United States Society on Dams (USSD) to recommend forensic team members. In March 2017, an ASDSO/USSD task force asked me to serve as team leader for the forensic investigation. With some trepidation because of the importance of this event to our professional community, I agreed and began working with the task force to assemble the team. At that time, details of the event were not clear and there was much speculation and misinformation circulating about. From what the task force and I knew, we judged that the team would need expertise in a range of disciplines:
- Hydraulic engineering.
- Spillway design and construction.
- Civil engineering.
- Engineering geology.
- Geotechnical engineering.
- Dam safety programme management.
- Human factors evaluations.
To provide the needed expertise, the task force recommended Irfan Alvi, Peter Dickson, Henry Falvey, Stephen Rigbey, and John Trojanowski as team members. The team was accepted by DWR and FERC, and we began our work in the middle of April 2017.
Although an ASDSO committee has been working on developing guidance for dam failure and incident investigations, no established and proven protocols exist for this type of investigation. One of the team’s first tasks was to decide how we were going to conduct the investigation. Initially, the team developed the following statement of purpose to guide our activities:
To complete a thorough review of available information to develop findings and opinions on the chain of conditions, actions, and inactions that caused the damage to the service spillway and emergency spillway, and why opportunities for intervention in the chain of conditions, actions, or inactions may not have been realised. Evaluations of actions, inactions, and decisions for the various stages of the project (pre-design, design, construction, operations, and maintenance) will consider the states of practice applicable to the various time periods involved.
In developing this statement of purpose, we recognised that we needed to be very careful to not apply hindsight bias to our evaluations. We needed to put ourselves in the place of the individuals involved at the times of various actions or inactions. I believe that throughout the investigation we stayed true to that intent.
As we began the work, the magnitude of the undertaking became apparent to the team. DWR provided us with electronic copies of a very large amount of information from its files, and, as we compiled the project history and assembled organisation charts, it became clear that a very large number of people and numerous organisations had touched the project in its nearly 50-year history. Even though we recognised that a large effort would be involved, we initially thought that the investigation could be completed by the end of the summer or early fall, and not, as it ultimately turned out, extending to early January 2018.
The identification of the physics of the spillway chute failure and the emergency spillway damage and the likely physical causes of those events became clear relatively quickly. Uplift pressures developed under areas of the concrete spillway chute, resulting in sudden loss of a section of the chute, followed by erosion of weathered rock underlying the chute in that area. The physical conditions that allowed this to occur were the result of the original design and construction and inadequate chute slab repairs over the years, leading to gradual deterioration of the condition of the slab and increased uplift forces. After water flowed over the emergency spillway during the incident, for the first time ever, the flowing water attacked and eroded areas of strongly weathered rock oriented such that the erosion moved toward the crest structure. With the erosion moving rapidly toward the emergency spillway crest structure, the evacuation order was issued. The team understood these essentials of the physical factors by the end of June 2017, and that understanding was only refined somewhat between then and the final report on 5 January 2018.
The team next turned to evaluating the human factors – the actions and inactions of individuals, groups, and organisations throughout the history of the project. The purpose of the human factors investigation was to answer a variety of questions such as:
- Why was the constructed spillway incompatible with its foundations?
- Why did the spillway design include underdrain pipes protruding into the slab in a manner that caused extensive cracking of the slab and flow of water through the slab?
- Why was the significance of the large service spillway underdrain system flows not recognised as a threat to the spillway chute?
- Why was the bedrock at the locations of the spillways believed to be generally non-erodible, despite design and construction geologic investigation results that showed areas of deep, strongly weathered rock?
- Why were service spillway chute repairs limited to repeated programs of generally surficial repairs?
Answering these questions required scrutiny of extensive historic records and interviews with many people involved with the project over the years. Despite being extensive, the historic records were found to lack some key elements, such as original spillway design calculations, records of some periods of construction, records of some of the repairs, etc. This required the team to make certain inferences from the available information.
The team interviewed more than 75 individuals, with each interview lasting at least an hour and some interviews lasting much longer. Some individuals were interviewed more than once. All interviews were confidential and conducted in a conversational style. Initially, the interviews were not highly structured, but later the team found that it was useful to prepare a customised list of questions for each interview. The interviews were supplemented by confidential surveys of DWR personnel. As might be expected, the interviews and surveys produced some inconsistent information, which the team had to weigh and assess in developing its findings.
Thorough completion of the human factors evaluation took more than four months, ending in November 2017. Preparation of the final report followed, with the report issued 5 January 2018. During the investigation, the team felt the pressure of the anticipation of our report, but we believed that it was most important that our investigation and report be thorough and complete, even if this took longer than expected. Throughout the investigation, all team members were involved in all aspects of the investigation and also in preparation of the entire report, not just in those aspects related to their specialisation. This collaborative team approach was very beneficial in challenging and vetting all of the findings before inclusion in the report.
After completion of the human factors investigation, the team concluded:
The Oroville Dam spillway incident was caused by a long-term systemic failure of the California Department of Water Resources (DWR), regulatory, and general industry practices to recognise and address inherent spillway design and construction weaknesses, poor bedrock quality, and deteriorated service spillway chute conditions. The incident cannot reasonably be “blamed” mainly on any one individual, group, or organisation.
This conclusion did not satisfy some individuals in the public and the press, but the team judged it to be the correct conclusion. Although dam safety practice in the US has improved significantly since the failures of Teton Dam and Kelly Barnes Dam in the 1970s, the February 2017 incident at Oroville Dam provides several lessons to be learned to further improve our practice. The lessons are detailed in the forensic report, but three of the more important ones are:
- Physical visual inspections alone, though important, may not identify latent vulnerabilities that can threaten a dam or its appurtenant structures.
- For older structures, comprehensive reviews of all available information need to be completed to evaluate the design, construction, and performance of a dam and its appurtenant structures against the current state of practice and knowledge.
- Failures of components of dams and appurtenant structures (e.g. a spillway chute) can compromise the operation of the facility and create a serious incident without release of stored reservoir water. Addressing these lessons will require the dam safety community to change some of the ways it does its work.
Although the investigation required thousands of hours of work and consumed our lives for nine months, all of the team members have no regrets about taking on the assignment. We are pleased that our report appears to have been well received, and we hope the lessons to be learned will lead to improved best practices in dam safety.