Journal of Dam Safety - Winter 2023 / 21.1

PROPOSED LEGISLATION FOR THE INVESTIGATION OF DAM SAFETY INCIDENTS AND FAILURES

A NUMERICAL STUDY OF HISTORICAL DAM FAILURES USING DSS-WISE LITE WEB-BASED SYSTEM

SCREENING TOOL FOR PREDICTING DROWNING POTENTIAL AT LOW-HEAD DAMS

Innovative approach to mitigate seismic risks and increase resiliency

The Priest Rapids Right Embankment Improvement Project improves the dam’s seismic resiliency while reducing the risks to the dam and downstream communities.

Looking for a smart, creative, and inspired team that works on some of the best projects for some of the best clients in the industry? Scan to learn more about the opportunities Stantec has for dam and levee professionals around the world.

ASDSO DAM SAFETY JOURNAL

The Journal of Dam Safety welcomes articles from any writer, but the opinions expressed in those articles do not necessarily reflect the opinions and policies of ASDSO.

EDITORIAL COMMITTEE

Gregory Paxson, Chair (Schnabel Engineering)

Peter Baril (GZA GeoEnvironmental Inc.)

Sushil K. Chaudhary (NM Office of the State Engineer)

Khaled Chowdhury (US Army Corps of Engineers)

Keith A. Ferguson (HDR Engineering, Inc.)

R. Craig Findlay (Findlay Engineering, Inc)

John W. France (JWF Consulting)

James W. Gallagher (NH Department of Environmental Services, Retired)

Michael J. Johnson (GENTERRA Consultants, Inc.)

Ian Maki (CA Department of Water Resources)

Lee Mauney (HDR)

Arthur C. Miller (AECOM)

Abbas Mokhtar-zadeh (Stone & Webster Engineering Corp)

Moumita Mukherjee (IN Department of Natural Resources)

Alan Rauch (Stantec Consulting Services)

Bryant Robbins (US Army Corps of Engineers)

John Roche (MD Department of the Environment)

Robrecht Schmitz

Nathan J. Snorteland (US Army Corps of Engineers)

William Sturtevant (Colorado Springs Utilities)

Blake P. Tullis (Utah State University)

James H. Weldon (Jim Weldon and Associates, LLC)

ASDSO REVIEW TEAM & CONTRIBUTORS

Lori C. Spragens, Executive Director: lspragens@damsafety.org

Katelyn Riley, Communications Director: kriley@damsafety.org

Brittany Lewis, Executive Office Manager/ Programs Support : blewis@damsafety.org

The Journal of Dam Safety is compiled, written and edited by the Association of State Dam Safety Officials

239 S. Limestone Street, Lexington, KY 40508

Tel: (859) 550-2788

Fax: (859) 550-2795

Email: info@damsafety.org

www.damsafety.org

This issue of The Journal of Dam Safety includes three articles with diverse topics, but hopefully, topics that you will enjoy and be able to apply to your own experiences.

Nathaniel Gee has completed doctoral research related to other industries’ guidelines on investigating safety incidents; his article with Dr. Rollin Hotchkiss proposes legislation for similar investigations related to dam safety incidents and failures.

For our second article, the authors revisit the use of hydraulic tools to model dam breaches and resulting inundation by comparing results with historic dam failures. This time, the web-based DSS-WISE ™ tool is used to model the 2005 Taum Sauk Dam failure and the 2020 Edenville and Sanford Dam failures. Validating our hydraulic models is important and also makes for interesting research.

The third article is focused on public safety at low-head dams and includes the development of a screening tool for evaluating potential drownings based on the geometry and hydraulics of these structures. Over 50 incidents that resulted in fatalities at low head dams were tested using the tool to estimate whether dangerous conditions develop, characterized by the type of hydraulic jump that occurs. Tools like this should help us prioritize removal or modification of the most dangerous structures.

This issue also includes a review of one of the most well-known books on floods and civil engineering, Rising Tide: The Great Mississippi Flood of 1927 and How it Changed America. Check out the review whether or not you have read the book.

Don’t miss the Volunteer Spotlight on ASDSO’s Journal Editorial Committee chair Greg Paxson and our Northeast Regional Spotlight on the collaborative response to the 2023 Vermont floods. We round out the issue with ASDSO news.

The Journal of Dam Safety is a quarterly publication dedicated to sharing technical content to benefit engineers, owners, operators, and others involved in dam and levee safety. Topics are presented from various geographic regions, relate to all types of dams, and represent different perspectives. Articles are selected to share important information, lessons to be learned, and to promote new technologies that can benefit the dam safety community. The journal is also a valuable source for industry news, organizational updates, and upcoming events.

We cannot stress enough that the Technical Journal Committee is always looking for articles of interest to our community. If you have an exciting project or topic to share with your peers, please contact Greg Paxson or others on the Technical Journal Committee to begin the process with a short abstract. Articles must be original work and appropriate for the readership. Please feel free to email gpaxson@schnabel-eng.com for more information on authorship or to provide feedback on recent articles.

ASDSO 2023-2024 BOARD OF DIRECTORS

Sharon K. Tapia, P.E. President California

Mia Kannik, P.E. President-Elect Ohio

Everett Taylor, P.E. Treasurer Utah

John Roche, P.E. Secretary Maryland

David M. Griffin, P.E. Past President Georgia

Jon Eggan Indiana

Keith Conrad, P.E. Nevada

Tim Gokie, P.E. Nebraska

William McKercher, P.E. Mississippi

Terry Medley, P.E. Kansas

William Salomaa Massachusetts

Ryan Stack, P.E. Missouri

William Vinson, P.E., CPM North Carolina

Ben Wagner, P.E. Alaska

Anthony Nokovich, P.E. Advisory Committee Chair American Water

Jennifer Williams, P.E. Advisory Committee Vice-Chair AECOM

ASDSO Disclaimer for Journal of Dam Safety

The material presented in this ASDSO publication has been prepared in accordance with generally recognized engineering principles and practices, and is for general information only. This information should not be used without first securing competent advice with respect to its suitability for any general or specific application. The contents of this publication are not intended to be and should not be construed to be a standard of the Association of State Dam Safety Officials (ASDSO) and are not intended for use as a reference in purchase specifications, contracts, regulations, statutes, or any other legal document.

No reference made in this publication to any specific method, product, process or service constitutes or implies an endorsement, recommendation, or warranty thereof by ASDSO.

ASDSO makes no representation or warranty of any kind, whether express or implied, concerning the accuracy, completeness, suitability or utility of any information, apparatus, product, or process discussed in this publication, and assumes no liability therefore. Anyone utilizing this information assumes all liability arising from such use, including but not limited to infringement of any patent or patents.

Each

AP Photo / Charlie Neibergall

ASDSO TRAINING CALENDAR

View additional information including agenda and registration details at www.damsafety.org/training

May 14

٣ Preventing Fatalities in Confined Spaces at Dams Webinar

May 21-23

٣ HEC-RAS 2D Classroom Seminar - Greenwood Village, CO

June 11

٣ Construction Quality Control / Information Data Systems for Dam Rehabilitation Webinar

June 18-21

٣ Seepage through Earth Dams Virtual Seminar

ASDSO provides a comprehensive library of over 130 On-Demand Webinars for dam safety professionals.

On-Demand webinars allow the registrant to access the webinar on-demand (any time 24/7) for up to one year following the date of purchase. On-Demand webinars allow registered participants to complete the quiz at the end and receive PDH credit.

In addition, all live webinars are added to the On-Demand library shortly after their live broadcast.

Visit https://portal.damsafety.org/asdso-webinars to view all On-Demand Webinars. ASDSO Webinars Available in On-Demand Format

Don’t compromise your dam’s safety. Worthington barriers mitigate the risk from debris blocking intakes and spillways so you can rest easy and generate more! Talk

Proposed Legislation for the Investigation of Dam Safety Incidents and Failures

ABSTRACT

The purpose of this article is to propose legislation that lays out the requirements of performing dam safety incident and failure investigations. The legislation draws on a review of five industry-sponsored safety incident investigation guidelines: (1) the National Transportation Safety Board (NTSB) for investigations of airplane, rail, marine and highway incidents/ crashes, in addition to pipelines failures and hazardous material incidents; (2) the National Institute of Standards and Technology (NIST) for building failures; (3) The Nuclear Regulatory Commission (NRC) for the nuclear industry; (4)

The Occupational Safety and Health Administration (OSHA) for work safety incidents; and (5) the proposed legislation for the National Disaster Safety Board. These agency policies and practices were reviewed to determine what triggers an investigation, what is done with lesser investigations, what is the timeline to arrive on site, what is the timeline to publish findings, if the report is publicly available, and how recommendations are handled. The proposed legislation defines a dam safety incident as an event where a failure mode initiates and progresses but does not lead to an uncontrolled release from the reservoir. It defines dam failure as an event where a failure mode initiates, progresses, and leads to an

Geologists from the U.S. Army Corps of Engineers and California Department of Water Resources survey the bottom of the damaged Oroville Dam spillway and the huge energy dissipator blocks at the base of the spillway

Nathaniel Gee, P.E. | Rollin Hotchkiss, PhD, P.E.

Brian Baer/ California Department of Water Resources

uncontrolled release of water from the reservoir. Failures are of three types: low consequence, significant consequence, and high consequence, which match the current industry practice for defining dam hazard class. For example, high consequence failure (type III) is a failure where there is loss of life. The legislation proposes that all high consequence failures as well as all failures and incidents that cause evacuation of 500 or more people would be required to have federally conducted and funded investigations.

Introduction

Whenever there is a failure of any type of infrastructure, especially when there is significant property damage and/ or loss of human life, people ask the same question: How did this happen? Therefore it is no surprise that several industries have established laws and policies to investigate and learn from failures. The National Transportation Safety Board (NTSB) investigates airplane, rail, marine, and highway incidents and crashes, in addition to pipeline failures and hazardous material incidents. They have even investigated major infrastructure failures, such as bridges. The National Institute of Standards and Technology (NIST) has the responsibility to investigate major building failures. The Nuclear Regulatory Commission (NRC) is responsible for investigation of incidents in the nuclear industry. And the Occupational Safety and Health Administration (OSHA) has programs for the investigation of work safety incidents. The National Disaster Safety Board legislation, which has passed the House but not the Senate, proposes procedures for investigation of natural disasters. However, there is no federal policy or legislation requiring the investigation of dams when they fail. Recent significant dam failure and incident investigations, such as the Oroville, Spencer, and Edenville Dams, were delayed because there was no standard process for conducting the investigations. This includes funding, scope, and development of teams for the investigation. The purpose of this article is to review the legislation and policies related to the investigation of incidents in the industries and agencies listed above and how they are implemented. The information is then used to inform proposed legislation for the dam safety industry.

Methodology

Earlier research looked at current practices within the dam safety profession when a dam safety incident or failure occurs (Gee, Baker, Mauney, & Hotchkiss, 2023) and found that there

is little consistency within the industry. Given that investigation practices and needs are not unique to dams, we decided to research the investigation practices of other industries. The review included the NTSB for aviation, marine, highway, pipeline, and hazardous materials; NIST for buildings; NRC for the nuclear industry; OSHA for work safety; and National Disaster Safety Board–proposed legislation for natural disasters.

Each agency’s laws, regulations, or internal policy was reviewed based on the following questions:

• How does the agency define an incident?

• What triggers an investigation?

• What did the agency or industry do about investigation of events that did not meet the threshold or trigger for investigation?

• How long did it take for the investigation team to arrive onsite?

• How long did it take to publish the final report?

• What were the agency’s practices for making recommendations within a report and how did those recommendations impact the respective industries?

• Was the final report made public (and how)?

This information was then reviewed to determine how it could apply to the dam safety industry, and then that information was used to craft the proposed legislation.

Policy/Literature Review

Each agency uses different policies and procedures to carry out its mission. The NTSB’s investigation definitions and practices are in the Code of Federal Regulations, CFR 49 part 831 (CFR, 2023). NIST practices for investigating building failures are defined by a specific law, The National Construction Safety Team Act (2002). The NRC uses internal policy document DT-23-06, NRC Incident Investigation Program, to define its investigation practices (NRC, 2023). OSHA’s main investigation program is defined in federal regulations (OSHA, 2004). Additionally, legislation has been proposed for the investigation of natural disasters (National Disaster Safety Board Act, 2021). While this legislation has not yet become law, we reviewed it as well.

In addition to these sources, individual reports, agency websites, databases, and other sources as outlined in each agency’s section below were used to better understand how these policies are being implemented in their respective agencies.

National Transportation Safety Board

The National Transportation Safety Board (NTSB) originated from the 1926 Air Commerce Act, but it was incorporated into the U.S. Department of Transportation (USDOT) in 1967. However, in 1974 the NTSB was made a separate entity based on the reasoning that “No federal agency can properly perform such (investigatory) functions unless it is totally separate and independent from any other … agency of the United States” (NTSB, 2022a).

The NTSB is composed of a board of five members, nominated by the president and approved by Congress. And while the agency is best known for investigations of aircraft incidents, it also has suboffices for highway, marine, railroad, pipeline, and hazardous material investigations. CFR Chapter 49 subpart B, Section VIII (CFR, 2023) defines when the NTSB should be involved in an investigation, which varies by type of incident. Railroad incidents, for example, are under the jurisdiction of the NTSB when there is a fatality or serious injury, or if the incident involves a passenger train. Pipeline failures are under the jurisdiction of the NTSB when there is a fatality, substantial property damage, or significant damage to the environment (NTSB, 2014).

For highway incidents, the NTSB collaborates with states and is often called in for more complex failures. This has led the NTSB to get involved in several infrastructure investigations such as those related to bridge failures. For example, the NTSB

conducted an 18-month-long investigation of the Florida International University pedestrian bridge failure, which occurred on March 15, 2018 (NTSB, 2019), as shown in Fig. 1. Since bridges, like dams, are large infrastructure projects, the NTSB’s investigation practices are of particular interest.

The NTSB states, “The Office of Highway Safety investigates accidents that have a significant impact on public confidence in highway transportation safety, highlight national safety issues, or generate high public interest and media attention” (NTSB, 2014). This scope is clearly limited though. For example, of the thousands of highway casualties that occurred in 2014, the NTSB only opened five field investigations, and only three of those led to full investigations. All three had a bridge or crossing involved (NTSB, 2014).

NTSB investigation reports with recommendations are all available in a single online site that is accessible to the public (NTSB, 2022b). Recommendations are tracked and can be given to manufacturers, private entities, regulators, or other federal agencies. There is, however, no specific authority to enforce these recommendations once they are submitted.

To better understand the NTSB investigation process, for this article, we reviewed all of its investigations initiated in 2019 (NTSB, 2022b; see Fig. 2). That year was selected to ensure that all investigations that had been initiated were completed. There were a total of 71 reports reviewed (10 aviation reports, 6 rail, 6 highway, 46 marine, 2 pipeline, and 1 hazardous material).

Figure 1 Collapse of the Florida International University Pedestrian bridge

NTSB 2019 Investigations

Time to arrive onsite: NTSB has no specific requirements in regard to how soon the team should be onsite. The final reports do not always state how long it takes them to arrive. Of the 71 reports reviewed, only 15 stated the time it took the team to get onsite. However, it is clear that there is a significant effort to arrive quickly. Arrival time averaged one day; the longest arrival time was 4 days.

Time to publish report: The NTSB’s website states, “In general, the NTSB tries to complete an investigation within 12 to 24 months, but … other factors can greatly affect that timing” (NTSB, 2022b). Of the reports reviewed, only 14 took more than 2 years to complete; the average was 1.46 years depending on the type of report. For example, of the 10 aviation reports, 5 took more than 2 years to complete, and a pipeline report and a hazardous material report took more than 3 years to complete. Marine reports were completed most promptly with an average of 1.1 years.

Recommendations: Of the 71 reviewed reports, 18 contained recommendations. When a report included recommendations, including just one recommendation was rare; the norm was to include several recommendations.

Fatalities: The CFR often cites fatalities as a reason for NTSB involvement, but analysis shows that most of the 2019 investigations did not report a fatality (22 of the 71 reports). The reason for the small fraction of reports with a fatality is due to NTSB involvement in minor marine accidents. For example, of the 46 marine investigations, only four had fatalities, and 32 had no injuries at all.

Budget: The NTSB is a very active organization involved in many investigations (71 in 2019). The cost of any specific investigation is not available, but the entire NTSB has a current FY2024 budget request of $145,000,000 (NTSB, 2023).

Partner Agencies

Many of the industries that the NTSB investigates have partner agencies that do their own investigations based on lower tiers or different jurisdictions. For example, pipelines: The Pipeline and Hazardous Materials Safety Administration (PHMSA), under USDOT, has jurisdiction over pipeline safety. The PHMSA’s internal policies help ensure that all pipeline failures are investigated except for those described above that are investigated by the NTSB, such as when a fatality occurs.

Figure 2 NTSB Investigations conducted in 2019

The PHMSA requires that all gas incidents be reported and have an internal report or investigation. An incident in this context is defined as any release that causes death, an injury necessitating hospitalization, property damage of $122,000 or more, or unintentional gas loss of 3 million cubic ft or more. Incidents must be reported within one hour of discovery, and the initial fact-finding report is due to be published within 48 hours of the incident (CFR, 2021).

The PHMSA reported that there were 614 reported pipeline incidents in the United States in 2019. These 614 incidents resulted in the death of a total of 10 people; some of these were also investigated by the NTSB. However, the PHMSA website only has a single inspection report from 2019 available to the public (PHMSA, 2019).

The Federal Aviation Administration (FAA) is also involved in airline incidents. The FAA works closely with the NTSB to determine what investigations are covered by FAA and which by NTSB, as described in FAA Order 8020.11C— Aircraft Accident and Incident Notification, Investigation and Reporting (USDOT, 2010). This very detailed 200-page policy creates clear definitions of aircraft accident, incident, near midair collision, serious injury, substantial damage, surface incident, and occurrence. It even defines fatal injury as any injury that results in a death within 30 days of the

accident. Actual events that meet the definition of aircraft accident are usually investigated by the NTSB, unless it involved an NTSB-owned aircraft, in which case, then FAA does the investigation. All incidents that meet the definition of occurrence, surface incident, or near midair collision are investigated by the FAA.

National Institute of Standards and Technology (NIST)

Another type of large infrastructure that can fail, leading to fatalities and other consequences, comprises buildings. Prior to 2002, building failures and dam failures were investigated in a similar manner on an ad hoc basis, depending on public pressures. An example of this was the Investigation of the Kansas City Hyatt Regency Walkways Collapse, a portion of a building that failed in 1981, killing 114. The investigation was completed by the National Bureau of Standards (1982), which was changed to NIST in 1988 (NIST, 2001/2023a). However, following the 2001 collapse of the World Trade Center Towers, there was desire that a clear authority should be established to perform building failure investigations. In 2002, Congress passed the National Construction Safety Team Act, officially giving NIST the authority to investigate building failures (Fig. 3).

Paul Morse, Courtesy of the George W. Bush Presidential Library

Figure 3 The Collapse of the Twin Towers Led to the Passing of the National Construction Safety Act, Requiring NIST to Investigate Building Failures With Substantial Loss of Life

Within the law, specific criteria require that investigation teams include at least one NIST employee and that the team be onsite within 48 hours of an incident. The law states that investigations should be conducted where, “failure of a building or buildings … has resulted in substantial loss of life or that pose significant potential for substantial loss of life” (National Construction Safety Team Act, 2002). The act also gives NIST authority to enter a site and subpoena authority to gain access to documents and witnesses. Plus, NIST is authorized to use funds otherwise authorized by law to carry out the investigations described in the act.

Since the passing of the act, NIST has performed five specific investigations, shown in Table 1.

The two large meteorological events that NIST investigated impacted multiple buildings.

NIST’s first report under the act on the World Trade Center collapse was an extensive investigation that actually led the to publication of 46 reports (NIST, 2023b). It was clearly a large effort, which is likely why the effort took so long (7 years) to complete.

Table 1 shows that in general all NIST reports are not completed quickly. The law requires a report to be completed 90 days after the investigation is complete; however, there is no time frame for completing the investigation. The fastest (2 years) was the investigation into the 2003 Rhode Island nightclub where 100 people died (NIST, 2005). The Joplin, Missouri, tornado report took slightly longer (NIST, n.d.). The other two investigations have not been completed, and no final reports are available.

One of the requirements in the law is that NIST will have a person onsite within 48 hours after the event. How long it took the NIST team to arrive onsite is also included in Table 1. The only time that NIST met the 48-hour requirement was on the recent Surfside building collapse in Miami, Florida, in 2021 (NIST, 2023). In the other four investigations, NIST was not onsite within the 48-hour requirement. This was in large part because the decision to perform an investigation often did not occur within that time frame. In the case of Hurricane Maria, the decision took months.

Although NIST is able to give recommendations for future building codes, it does not have authority to enforce the recommendations. NIST can propose and promote its recommendations, which has proven effective, as the findings and recommendations in every NIST-published investigation have directly impacted building codes. For example, the investigation into the nightclub fire states, “Ten recommendations to improve model building and fire codes, standards and practices (as they existed in February 2003) resulted from the investigation.”

NIST has recently considered conducting investigations of dam failures. Its 2020 annual report stated:

NIST also considered a dam failure, but did not rate {investigate} this event because it appears to fall outside the scope of NIST. There is an Interagency Committee on Dam Safety, chaired by FEMA [Federal Emergency Management Agency]. There could be circumstances in the future where the performance of buildings or other structures under flooding from dam failures could exhibit failures worthy of the implementation of the National Construction Safety Team Act. (NIST, 2020)

Nuclear Regulatory Commission (NRC)

The Nuclear Regulatory Commission (NRC) has policy in place to investigate any significant operational event. This is defined as any radiological, safeguard, or other safety-related operational event at an NRC-licensed facility that poses an actual or a potential hazard to public health and safety, property, or the environment (NRC, 2023).

When an event occurs, the Executive Director for Operations (EDO) determines what level of investigation is required based on criteria outlined in policy and their judgment. There are three levels of investigation: the highest level is an Independent Incident Investigation (IIT), and the others are an Augmented Inspection Team (AIT) and a Special Inspection (SI).

The highest level, IIT, is as the name states, independent. The policy defines this by saying, “This group consists of technical experts who, to the extent practicable, do not have, and have not had, previous significant involvement with licensing and

inspection activities at the affected facility and who perform the single NRC investigation of a significant operational event as described in Part II of this handbook. An NRC senior manager leads the IIT” (NRC, 2023). The AIT and SI do not require this level of independence.

The investigation program is funded by the NRC; however, the NRC recoups 90% of its costs through fees to licensees, as outlined in the Independent Appropriations Act (1952) and Omnibus Reconciliation Act (1990). The NRC Incident Investigation Program has specific requirements on scope of the investigation, calling for root cause analysis, contributing factors, and human factors among conditions that must be evaluated. The IIT is to be onsite as soon as practicable after the facility has been placed in a safe, secure, and stable condition. The strictest requirement is that from the activation of the team, they only have 45 days to submit their report to the EDO, unless the EDO grants an extension. They also have to make the report publicly available on the NRC public website/database Agencywide Documents Access and Management System (ADAMS).

This requirement to make the document publicly available is likely to encourage transparency. However, in practice, the reports are difficult for anyone to locate. The ADAMS database is used for thousands of documents. Every day a new set of documents is uploaded. For example, on June 16, 2023, alone there were 20 documents posted for public access (U.S. NRC, 2023). The documents are not catalogued or referenced to allow for ease of searching. We accessed database in an effort to review IIT reports, but no IIT reports could be found using standard search functions. We contacted NRC, and an NRC librarian spent extensive time attempting to locate any IIT reports. Only two were found, and that was due to press releases that gave the specific number of the reports, and then only by searching for those specific numbers could the reports be located. One was on a loss of iridium from a Regional Cancer Center in Pennsylvania on November 16, 1992 (NRC, 1993). The other was a near miss with the mishandling of 150 kilograms of uranium at GE Nuclear Fuel plant (NRC, 1991).

Neither of the reports were completed within the 45-day requirement. However, the policy has been updated several times since these reports were completed in the early 1990s. The limited number of the IIT reports that were located made any analysis meaningless of how often or effective the teams were.

What is clear is that the practice of using the ADAMS database does not lead to the transparency that was likely intended.

Occupational Safety and Health Administration (OSHA)

The Occupational Safety and Health Administration (OSHA) has two investigation programs. The first encourages all employers to do internal investigations whenever someone is hurt or there is a close calls/near miss as outlined in a 2015 guidance document published by OSHA, Incident [Accident] Investigations: A Guide for Employers (OSHA, 2015). The second program covers investigations that OSHA performs directly. This program is outlined in the Code of Federal Regulations, CFR Part number 1960 Subpart D.

The criterion for OSHA to perform an investigation is whenever the incident results in a fatality or three or more employees being hospitalized (OSHA, 2004). Further guidance on how to conduct the investigation is described in OSHA internal guidance titled Fatality/Catastrophe Investigation Procedures (OSHA, 2005). While the details of the investigation are different from what might take place in an investigation of infrastructure failure or near

miss, it still involves a root cause analysis. Also, the timing of the investigation is important. The policy states, “The investigation should be initiated as soon as possible after receiving report of the incident, ideally within one working day, by an appropriately trained and experienced compliance officer” (OSHA, 2005).

The timeline to complete the investigation report is governed by the OSHA Act (1970). Part of the investigation requires that the OSHA inspection team determine if the owner should be fined. “No citation may be issued under this section after the expiration of six months following the occurrence of any violation” (OSHA Act, 1970). This means that the report has to be finalized within 6 months from the day of the triggering occurrence. This hard deadline motivates OSHA to complete its investigations to enable it to issue citations.

OSHA does not make these investigation reports public, but a summary of its investigations is on the OSHA website (OSHA, 2023). Given that the reports are not public, little analysis could be completed, but a limited review showed that in 2022 there were 893 investigations from 905 fatalities (OSHA, 2023). This shows that of all the agencies listed, OSHA completes significantly more investigations than the other agencies.

National Disaster Safety Board Act

In August 2023 over 100 people were killed in wildfires on the island of Maui in Hawaii (Mascarenhas, 2023), shown in Fig. 4. This renewed calls for a National Disaster Safety Board. The board had been proposed in Congress and passed the House in 2021 as H.R. 5532, but was not taken up by the Senate (National Disaster Safety Board Act, 2021).

The bill called for creation of a seven-person board. The board would be called upon to investigate all natural disasters that took the lives of 10 or more people. The board was given broad authority to determine the scope of the work moving forward. For example, the board would also be given discretion to define what constitutes a natural disaster. They could also vote to investigate disasters when called upon by state or local government representatives or when they felt that by so doing they could reduce life loss in the future. The bill would fund the investigations through the Secretary of Transportation and called for funding up to $60 million by fiscal year 2025.

The bill would not direct the board regarding a timeline for having staff on site, a timeline for publishing their investigations, or how they were to coordinate with other agencies. Rather, 1 year following the passage of the proposed act, the board was to establish memorandums of understanding with many of the agencies and existing boards that perform investigations, including FEMA, NTSB, and NIST. They also would have the authority to make recommendations, and federal agencies would have 90 days to respond to what extent and how the recommendations would be adopted (National Disaster Safety Board Act, 2021).

It is difficult to determine the level of impact this board would have on the practice of dam incident and failure investigations, since the board, not the legislation, would define what constitutes a natural disaster. However, it is likely that many dam failures would fall under the board’s jurisdiction. For example, Hurricane Matthew killed over 600 people, and following this natural disaster over 71 dams (20 of which were breached) were inspected or reviewed by FEMA (2017). Tropical storms often cause the failure of dams and have significant life loss (Gee et al., 2023). However, not all dam failures or even all dam failures with life loss would likely be investigated by this board. For instance, dams that fail during normal operations, such as the Teton Dam, or due to misoperation, such as the Taum Sauk Dam, would likely not be considered failures related to natural disaster. If this law passes and this board is created, the board would need to work closely with other agencies through a memorandum of understanding to determine a path forward concerning dam failures, as is the case with NIST on building failures.

This would, however, address investigations of flooding incidents that can cause significant loss of life when a dam was not involved. This would include situations like Waverly, TN, where flooding in 2021 killed 20 people (Wikipedia, 2021).

State of Hawaii, 2023

Figure 4 Image From Kula wildfire in Maui

NTSB (Plane)

COMPARISON OF DIFFERENT AGENCIES’ INVESTIGATION POLICIES

Accidents involving a civil or public aircraft (incidents are investigated by FAA)

NTSB (Highway)

No requirement but usually 24-48 hours following the incident

The Office of Highway Safety investigates accidents that have a significant impact on public confidence

NTSB (Rail)

NTSB (Pipeline)

NTSB (Marine)

NTSB (Hazardous Materials)

the incident

When there is a fatality or serious injury No requirement

Fatality, significant environmental or property damage (owners investigate all other failures)

6+ lives lost, or 500K in damage or 100 gross tons or as requested by Coast Guard

They may investigate any transportation accident that involves HazMat

the incident

No requirement but usually 24-48 hours following the incident No requirement

months

CFR 49, subpart B, Chapt. VIll, part 831. NTSB internal policy, The Investigative Process

CFR 49, subpart B, Chapt. VIlI, part 831. NTSB internal policy,

Investigative Process

CFR 49, subpart B, Chapt. VIlI, part 831. NTSB internal policy, The Investigative Process

CFR 49, subpart B, Chapt. VIlI, part 831. NTSB internal policy, The Investigative Process

No requirement

No requirement but guidance to be done within 12-24 months

No requirement

No requirement but guidance to be done within 12-24 months

Federal appropriations Yes

CFR 49, subpart B, Chapt. VIlI, part 831. NTSB internal policy, The Investigative Process

CFR 49, subpart B, Chapt. VIlI, part 831. NTSB internal policy, The Investigative Process

NIST

Building failure that has resulted in substantial loss of life or significant potential for such 48 hours

No criteria (one over 5 years and not completed)

Federal appropriations Yes

National Instruction Safety Act, Public Law 107-231

NRC

Based on man criteria including risk of Core Damage ultimately a judgement call by the Executive Director of Operations at NRC

OSHA

Fatality, or 3 or more employees injured (recommends all other near misses be investigated by the owner)

Only states as soon as practicable after the facility is safe

45 days

90% fees to licensees, 10% appropriations

Yes (however very difficult to search)

US Nuclear Regulatory Comission Management Directive DT14-14

As soon as practicable usually 1 working day

60 days

Federal appropriations and fines

No, summaries are searchable but not the final report

CFR 29, part 1960, Subpart D Accident Investigation National Disaster Investigation Board

National Disaster Investigation Board Act (not yet passed)

Current Dam Safety Legislation

Historically, the development of dam safety regulations and laws have been motivated by dam failures, both for individual states and at the federal level. For example, the Commonwealth of Pennsylvania passed a law regulating the safety of existing dams in 1913 following the Austin Dam failure in 1911 (Rose, 2013). While many of those agencies had significant policies and technical instructions for design and construction of dams, they had few policies related to dam safety once the dams were completed. States acted according to the specific state’s laws, and by 1970 only 24 states had enacted any dam safety laws (Gee, 2023). Prior to 1972 there were no federal laws related to dam safety (Gee, 2018). All federal dams were under the self-regulation of the agency that owned them. The federal government did regulate power dams through Federal Energy Regulatory Commission, but these regulations focused on design and construction. The first federal dam safety law, the National Inspection of Dams Act, was passed in November 1972 following the Buffalo Creek disaster (Gee, Ferguson, Pittman, & Krivanec, 2022). Additionally, the Reclamation Safety of Dams Act was passed in the aftermath of the Teton

Dam failure (Reclamation Safety of Dams Act, 1978). When the original 1972 dam safety law passed, it was called the National Dam Inspection Act because its focus was on the inspection of dams. In the years since then, this law has been expanded, and the governing dam safety law is now the National Dam Safety Program Act of 1996, which has been amended many times, the most recent being in December 17, 2022 (National Dam Safety Program Act, 2022).

Despite the many amendments over time there is nothing in the law requiring or calling for investigations when dams fail. Similarly, the federal guidelines do not currently include any guidance on encouraging investigation of dam failures. This does not mean that the industry is unaware of the need to perform investigations. There was a recent review of 58 dam failure and incident investigations that have been performed since 1960. But that review found very little consistency on how and when these investigations occurred (Gee et al., 2023). While there is no guidance on when or how to fund an investigation, the industry, through the Association of State Dam Safety Officials (ASDSO), has published some technical guidance on how to conduct an investigations (ASDSO, 2021).

Lessons From Other Industries

Table 2 summarizes the investigation practices of other industries and agencies. This section compares and contrasts these practices and introduces insights that are included in proposed legislation discussed later in the article.

Definitions or triggers used to determine action: All agencies use some definition or definitions to define their authority and when they are in charge of a given investigation. They similarly suggest other agencies or owners to perform investigations when the threshold for conducting their own investigations is not met. This is the case with NTSB (pipelines, aircraft, and marine), NRC, and OSHA.

For some of these organizations, the criteria for triggering investigations are very clear and explicit. For example, OSHA requires an investigation when there is a fatality or three or more employees are injured. For those that have clear conditions for when they need to investigate, many of the criteria are based on a single or multiple fatalities (NTSB for rail, pipeline, and marine; NIST; OSHA and National Disaster Review Board [NDSB]).

Some have used criteria that allow for some flexibility, including those that use fatalities. Whether a fatality has occurred is usually clearcut, but the number of fatalities that leads to an investigation is not always so clear. For example, NIST calls for its involvement when a building failure has substantial loss of life, but does not quantitatively define “substantial.”

Also, some of the investigation triggering determinations are left up to a specific person. For example, the NRC’s decision is based on loose criteria but is ultimately left up to the Executive Director for Operations (EDO). While the NDSB has the legislation guidance of 10 or more lives lost, the legislation also gives it broad discretion to investigate when there are fewer fatalities or not investigate in cases when that threshold is surpassed.

All of these are options for dam failures. The determination to investigate could ultimately be left up to a board or governing body. While the flexibility this allows is valuable, it also brings in ambiguity and increased risk of continuing confusion and delays (such as exists today). Therefore, a specific definition of when investigations should be conducted by the federal government and under government funding is advisable.

Life loss is clearly the trigger that Congress has regularly used. Therefore, the proposed legislation outlined below calls for government-funded and government-run investigations for all dam failures and incidents that involve life loss.

However, if this were the only criterion, then two recent major independent investigations within the dam safety industry would not have been required: the investigations of the Edenville Dam failure in Michigan and the Oroville Dam Spillway Incident in California, both of which caused evacuation of thousands of people (France et al., 2018; France, Alvi, Miller, Williams, & Higinbotham, 2022).

Neither had loss of life, but both were important events and had significant impact on the dam safety industry. In both of these cases, the populations below the dam were evacuated. Based on this, we propose that incidents and failures that lead to evacuation be investigated even if there is no life loss.

The level of investigation also may vary if the event is an incident or a failure. Therefore, to ensure clarity within the industry, criteria are included in the definitions to clearly distinguish between an incident and a failure.

Time on site: Table 2 shows that the time to arrive onsite varies considerably across different investigatory agencies. Most do not have a specific requirement, but their documentation notes the importance of arriving onsite “as soon as practicable.” The only agency that has a specific requirement is NIST; the requirement is 48 hours, but NIST has met that requirement only once.

For dams, it was considered that 48 hours is not likely to be a practical requirement. The legislation proposed “as soon as practicable” language but includes “but not later than 7 days following the incident or failure.” This allows for reasonable time with a maximum of 7 days, which will require quick action to form the team but seems practicable.

Time to publish: Most of the agencies understand the importance of publishing in a timely manner. The NTSB puts out internal guidance stating that it should publish final reports within 24–48 months. However, a review of NTSB actual practices shows that it regularly exceeded this guidance. Only two agencies have a specific requirement: OSHA and NRC. We could locate only two investigation reports from the NRC, and both exceeded the NRC’s 45day requirement. The two investigations were published in less than a year, so while the requirement was not met, it appeared to lead to faster publication of reports. OSHA does

regularly meet its 60-day requirement. The law requires that if OSHA does not publish and fine the owner within 60 days, it will lose the opportunity to set fines.

The most concerning is NIST, which has no internal guidance or requirements on time to complete an investigation. NIST reports often come many years after the event. This could be for many reasons, but as we reviewed the reports, it seemed that that scope creep was a significant issue. The delay in publishing these reports and the lessons learned can have significant impact. The public and industry attention can be lost by the time the reports are completed, and this will likely increase the difficulty in getting the information from the reports out. This could easily occur with the investigation of dams, and therefore the recommended legislation calls for 1 year to publish. But recognizing that this is not always possible, there is allowance for the investigation team to request an extension through the administrating agency.

Recommendations: All agencies reviewed had the authority to issue recommendations to the owner and/or the industry. It is imperative that the lessons learned from these investigations are applied. The most effective way to do that is through recommendations that are made and tracked. The proposed legislation gives authority for the investigation team to make recommendations to the dam industry. The dam safety industry is regulated by a mix of federal and state laws and regulations. It is not considered necessary that every state and federal regulator respond to each recommendation by an investigating party. Rather the administrator (FEMA) can require response by the Interagency Committee on Dam Safety (ICODS) or the National Dam Safety Review Board (NDSRB) into how the recommendations will be incorporated into federal guidelines for dam safety. The federal guidelines are used by almost all federal regulators and are often incorporated into state programs as well.

Making investigation reports available to the public: It is important that investigations are accessible. The NTSB has a very effective website that makes all its investigations available and easy to search. This is an effective model for what is being proposed. The NRC is a good example of apparent transparency actually leading to a lack of transparency. NRC investigations are on a public database, but even NRC librarians cannot find them. The proposed legislation calls for the investigations to be available on a searchable public database.

Proposed Legislation

Since the 1972 National Dam Inspection Act (Public Law 92367), further legislative advancements in dam safety have been accomplished by amending the National Dam Safety Program Act of 1996. The proposed legislation is suggested by the authors as an amendment of that act. The current act ends at Section 15, which is about national low-head dam inventory (National Dam Safety Program Act, 2022). The authors propose adding a Section 16 on dam failure investigations.

Sec. 16 Dam Failure Investigations

(A) Definitions

a. Operational Incident: An incident when there are downstream consequences due to flows passed by the dam but not due to the failure of the dam or dam component.

b. Dam Safety Incident: A potential failure mode has initiated and progressed but has not led to an uncontrolled release of water from the reservoir.

c. Dam Failure: A failure mode initiates, progresses, and leads to an uncontrolled release of water from the reservoir.

d. Type I Dam Failure: A failure occurs but the release is contained within downstream banks or levees and there is no environmental/property damage or life loss downstream.

e. Type II Dam Failure: A failure occurs and there is downstream environmental/property damage but no life loss.

f. Type III Dam Failure: A failure occurs and there is downstream life loss.

g. Dam Incident and Failure Investigation Database: A single online searchable database used to access incident and failure investigations.

(B) Dam Incident Investigations

a. Not later than 18 months after the date of enactment of this section, the Administrator in consultation with the heads of appropriate Federal and State agencies shall update Dam Safety Federal Guidelines and State Model Law to include guidance for investigations of all dam safety and operational incidents.

a. Investigations shall include at a minimum:

i. Determination of physical failure mechanism (s) and root cause analysis. This includes hydrologic and hydraulic analysis of an operational incident.

ii. Human and Programmatic factors that impacted the incident.

iii. Review of emergency response.

iv. Operational incidents that include loss of life will include investigation of consequences including determination of factors that led to any life loss.

c. Incident investigation scope, funding, and other aspects will be determined by policies of regulating state and federal agencies.

d. Following the completion of the investigations the regulating agency will submit the final report to FEMA which will include it in the database noted in Section 16.D.

(C) Dam Failure Investigations

a. Not later than 18 months after the date of enactment of this section, the Administrator in consultation with the heads of appropriate Federal and State agencies shall update Dam Safety Federal Guidelines and State Model Law to include guidance for Investigations into all dam failures (Type I-III).

b. Dam failure investigations shall include at a minimum:

i. Determination of physical failure mechanism (s) and root cause analysis.

ii. Human and Programmatic factors that impacted the failure.

iii. Review of emergency response.

iv. Type III failure investigations will include determination of factors that led to any life loss.

c. Type I and II dam failure investigations

i. Type I and II failure investigation scope, funding, and other aspects will be determined by policies of regulating state and federal agencies.

ii. Following the completion of the investigations the regulating agency will submit the final report to FEMA that will include it in the database noted in Section 16.C.

d. Type III dam failure investigations

i. Type III dam failure investigation teams will be selected by National Dam Safety Review Board (NDSRB) and shall be comprised of a minimum of three industry experts.

ii. These failure teams will be selected and onsite within 72 hours following the failure if the site is safe to access.

iii. Type III failure reports will be completed within 1 year following the failure. The administrator has authority to extend the 1 year deadline if requested by the team in writing.

iv. The administrator shall ensure funding of Type III investigations.

v. These failure teams have authority to issue recommendations to the owner and the industry.

1. The administrator will work with the Interagency Committee on Dam Safety (ICODS) and National Dam Safety Review Board (NDSRB) to respond to all recommendations within 180 days of a finalized report.

The response is to include how/whether federal guidelines will be updated based on the recommendations.

vi. Operational incidents that result in loss of life will follow the investigation requirements of Type III dam failure investigations.

vii. Dam failures or dam safety incidents that result in the evacuation of downstream communities of over 500 people will follow the investigation requirements of Type III dam failure investigations.

(D) Dam Incident and Failure Investigation Database

a. Not later than 18 months after the date of enactment of this section, the administrator in consultation with the heads of appropriate federal and state agencies shall –

i. Compile existing failure investigations on past dam failures.

ii. Post the investigation reports to a public website.

iii. Make future investigations available on said public website.

iv. Maintain the database of failure investigations for the industry and public.

*

Conclusion and Future Actions

The proposed legislation will greatly improve dam failure investigations in the future and improve the industry’s ability to learn from failures and incidents. It is understood that enacting legislation is a long process, and it may take many years for this proposed legislation or parts of it are adopted by Congress. In the meantime, many of the proposed solutions can be implemented and encouraged by organizations within the industry. For example, individual regulators and owners can adopt internal investigation policies to ensure that incidents and failures that occur at the dams they own or regulate are investigated.

Similarly, industry organizations such as United States Society on Dams (USSD), ASDSO, and the Center for Energy Advancement through Technological Innovation (CEATI) can work together to encourage investigations when these incidents and failures occur. What has been presented here is a proposal. As industry works together to find solutions, there may be other solutions that meet the same needs. There is a current effort within USSD to draft a white paper putting forward proposals on future investigations.

There are also efforts to create a database of past incidents, regardless of the status of investigations.* All these efforts will be valuable as the industry seeks to learn from incidents and failures.

Acknowledgments

It’s impossible to thank all the people who supported this article and related work. The authors want to give special thanks to ASDSO and USSD. We love being a part of these organizations. They have given opportunities for growth in the field that we could have never achieved without them. They are full of people who care deeply about their work and give their time to try to improve the life of others. Lee Mauney has been a sounding board for so much of this work and constantly amazes us with how much he accomplishes. Finally, this works leans heavily on the past work of Mark Baker. He first gave us the idea of the importance of dam incident and failure investigations and has done tremendous work in this area. To them and many others who have helped in this work, thank you.

HEAVY CIVIL

HYDROPOWER & RENEWABLES

POWER DELIVERY

UNDERGROUND

OIL & GAS

ENVIRONMENTAL

barnard-inc.com

Editors Note: ASDSO maintains the Dam Incident Database on its website. The ASDSO Dam Incident Database provides basic information on dam safety incidents to ASDSO members, dam safety stakeholders, the media and the public. While it does include both historic and recent data, the database is not considered comprehensive of all dam safety incidents and reflects only the data that ASDSO has been able to collect.

References

Association of State Dam Safety Officials. (2021). Dam failure and incident investigation guide.

Code of Federal Regulations (CFR). (2021). CFR Title 49 Part 191 Transportation of Natural and other Gas by Pipeline; Annual, Incident and other Reporting. https://www.ecfr.gov/current/title-49/subtitle-B/chapter-I/subchapter-D/part-191

Code of Federal Regulations (CFR). (2023, May 20). Code of federal regulations. https://www.ecfr.gov/current/title-49/subtitle-B/ chapter-VIII/part-831

Federal Emergency Management Administration (FEMA). (2017). Hurricane Matthew in North Carolina dam risk management assessment report. FEMA. https://damfailures.org/wp-content/uploads/2022/05/IR_Hurricane-Matthew.pdf

France, J., Alvi, I., Dickson, P., Rigbey, S., Trojanowski, J., & Falvey, H. (2018). Independent forensic team report: Oroville dam spillway incident. California Department of Water Resources. https://cawaterlibrary.net/wp-content/uploads/2018/03/Independent-ForensicTeam-Report-Final-01-05-18.pdf

France, J., Alvi, I., Miller, A. C., Williams, J. L., & Higinbotham, S. (2022). Independent forensic team report: Investigation of failures of Edenville and Sanford Dams. Federal Energy Regulatory Agency. https://damsafety-prod.s3.amazonaws.com/s3fs-public/files/ Edenville-Sanford_Final%20Report_Main%20Report%20and%20Appendices.pdf

Gee, N. (2018). The impact of dam failures on the development of dam safety legislation and policy in the 1970s. Journal of Dam Safety, 15(4), 9-16.

Gee, N. (2023). Dam safety 1.0, 2.0, 3.0 and the failures that got us there. FEMA Technical Seminar (p. 39). FEMA.

Gee, N., Baker, M., Mauney, L., & Hotchkiss, R. (2023). Analysis of dam failure and incident investigations in the United States from 1960 through 2022 and a framework for improving dam failure and incident investigations in the future. ASCE Journal of Water Resources and Planning, 150(3).

Gee, N., Ferguson, K., Pittman, J., & Krivanec, C. (2022). Failure of Buffalo Creek Dam, February 26, 1972. Journal of Dam Safety, 19 (3), 14–30.

Independent Offices Appropriation Act (1952).

Mascarenhas, L. (2023, August 16). Maui wildfires death toll rises to 110 as official says using the warning sirens wouldn’t have saved lives. CNN. www.cnn.com: https://www.cnn.com/2023/08/16/us/hawaii-maui-wildfires-death-toll-wednesday/index.html

National Bureau of Standards. (1982). Investigation of the Kansas City Hyatt Regency walkways collapse. U.S. Department of Commerce.

National Construction Safety Team Act, Pub. L. 107-231 (2002). https://www.govinfo.gov/app/details/PLAW-107publ231

National Dam Safety Program Act. (2022).

National Disaster Safety Board Act of 2021. (2021). https://www.congress.gov/bill/117th-congress/house-bill/5532/text

National Institute of Standards and Technology (NIST). (2005). Report of the technical investigation of the Station Nightclub fire. U.S. Department of Commerce.

National Institute of Standards and Technology (NIST). (2020). National construction safety team advisory committee 2020 annual report.

National Institute of Standards and Technology (NIST). (2023a). From NBS to NIST. https://www.nist.gov/pao/nist-100foundations-progress/nbs-nist#:~:text=Founded%20as%20the%20National%20Bureau,%2C%20or%20NIST%2C%20in%201988 (Original work published 2001)

National Institute of Standards and Technology (NIST). (2023b, May 27). World Trade Center investigation. NIST. https://www.nist.gov/world-trade-center-investigation

National Institute of Standards and Technology (NIST). (n.d.). Technical investigation of the May 22, 2011, Tornado in Joplin, Missouri.

National Transportation Safety Board (NTSB). (2014). Annual report to Congress 2014.

National Transportation Safety Board (NTSB). (2019). Highway accident report—Pedestrian bridge collapse over SW 8th street. Miami FL, March 15, 2018. https://www.ntsb.gov/investigations/AccidentReports/Reports/HAR1902.pdf

National Transportation Safety Board (NTSB). (2022a). History of the National Transportation Safety Board. https://www.ntsb.gov/ about/history/Pages/default.aspx

National Transportation Safety Board (NTSB). (2022b). NTSB Investigations. https://www.ntsb.gov/investigations/Pages/ Investigations.aspx

National Transportation Safety Board (NTSB). (2023, May). Letter from Jennifer Homendy, Chair NTSB, to Speaker Kevin McCarthy, reauthorization request. NTSB. https://www.ntsb.gov/news/Documents/NTSB-Reauthorization-Proposal.pdf

Nuclear Regulatory Commission. (1991). Potential criticality accident at the General Electric nuclear fuel and component manufacturing facility, May, 29, 1991.

Nuclear Regulatory Commission (NRC). (1993). Loss of an iridium-192 source and therapy misadministration at Indiana Regional Cancer Center Indiana, Pennsylvania, on November 16, 1992.

Nuclear Regulatory Commission (NRC). (2023). NRC Incident Investigation Program, DT-23-06.

Occupational Safety and Health Act of 1970, Pub. L. 91-596 (1970). https://www.osha.gov/laws-regs/oshact/completeoshact

Occupational Safety and Health Administration (OSHA). (2004, November 26). Standard 1960.29 – Accident investigation. OSHA.

https://www.osha.gov/laws-regs/regulations/standardnumber/1960/1960.29

Occupational Safety and Health Administration (OSHA). (2005, April 14). Fatality/catastrophe investigation procedures. OSHA.

https://www.osha.gov/enforcement/directives/cpl-02-00-137

Occupational Safety and Health Administration (OSHA). (2015). Incident [accident] investigations: A Guide for employers.

Occupational Safety and Health Administration (OSHA). (2023, May 25). Fatality inspection data: Work-related fatalities for cases inspected by federal or state OSHA. OSHA.

https://www.osha.gov/fatalities#&sort[#incSum]=0-1-1-0

Omnibus Budget Reconciliation Act of 1990. (1990). Pub. L. 101-508. https://www.congress.gov/101/statute/STATUTE-104/ STATUTE-104-Pg1388.pdf

Pipeline and Hazardous Materials Safety Administration (PHMSA). (2019). Pipeline failure investigation reports.

https://www.phmsa.dot.gov/safety-reports/pipeline-failure-investigation-reports

Reclamation Safety of Dams Act, Pub. L. 95-578 (1978). https://www.congress.gov/95/statute/STATUTE-92/STATUTE-92-Pg2471.pdf

Rose, A. T. (2013). The influence of dam failures on dam safety laws in Pennsylvania. Dam Safety 2013, Providence, RI, United States. https://damsafety.org/content/influence-dam-failures-dam-safety-laws-pennsylvania

September 11, 2001 Photographs. (n.d.). George W. Bush Library. https://www.georgewbushlibrary.gov/explore/photo-videocollections/september-11-2001-gallery

State of Hawaii. (2023, August 11). Revised: DLNR News release-aerial photos and video of all four Maui fires, August 11, 2023. Office of the Governor [of Hawaii]. www.governor.hawaii.gov: https://governor.hawaii.gov/newsroom/revised-dlnr-news-release-aerialphotos-and-video-of-all-four-maui-fires-august-11-2023/

U.S. Nuclear Regulatory Commission (NRC). (2023, June 16). NRC Incident Investigation Program [document DT-23-06]. NRC Webbased ADAMS. https://adams.nrc.gov/wba/

U.S. Department of Transportation. (2010, February 2). Aircraft accident and incident notification, investigation, and reporting [Order 8020.11C]. FAA. https://www.faa.gov/documentLibrary/media/Order/8020.11C_.pdf

Wikipedia. (2021, August 21). 2021 Tennessee floods. https://en.wikipedia.org/wiki/2021_Tennessee_floods#cite_note-WKRN-2

Nathaniel Gee, P.E.

Manager, Dam Safety Governance & Oversight, Tennessee Valley Authority 1101 Market St. Chattanooga, TN, 37402 • (702) 378-8554 • ngee@tva.gov

Nathaniel Gee has over 15 years’ experience as a professional engineer with a focus on dam safety. He has performed work on some of nation’s largest and oldest dams including Roosevelt Dam, Fontana Dam, Grand Coulee Dam, and Hoover Dam. He received a bachelor’s degree in civil engineering from Brigham Young University, and a master’s degree from the University of Nevada – Las Vegas. In his spare time he loves to write and recently published his second book, Of Pigs and Priests, a romantic comedy. He lives in Signal Mountain, Tennessee, with his wonderful wife, Jeanine, and nine energetic children.

Rollin H. Hotchkiss, PhD, P.E., D.WRE, F.ASCE

Professor, Civil and Construction Engineering Department, Brigham Young University 430 Engineering Building, Provo, UT 84663 • (801) 422-6234 • rhh@byu.edu

Rollin H. Hotchkiss is a water resources engineering professor in the Civil and Construction Engineering Department at Brigham Young University. His current research areas include mitigating drowning at low-head dams, extending the useful life of dams and reservoirs, and the economics of sustainability. He is a co-leader of a task force to create an inventory of low-head dams in the United States. He has authored more than 50 refereed journal articles and more than 100 conference papers. He teaches courses in fluid mechanics, 2D river modeling, hydraulic structures, and leadership.

ASDSO Peer Reviewers

The article was peer reviewed by John France, P.E. (JWF Consulting LLC) and Bryant Robbins, PhD, P.E. (U.S. Army Corps of Engineers)

Delivering sustainable legacies

AECOM draws upon our network of dam, hydropower, levee and mining experts to provide technical excellence for the most complex project. Our network of global experts work collaboratively with our clients to provide tailored and innovative solutions to meet their unique needs.

Dennis J. Hogan, P.E. Americas/Global Dams & Hydropower Market Sector Leader dennis.hogan@aecom.com

A Numerical Study of Historical Dam Failures Using DSS-WISE Lite Web-Based System

Nuttita Pophet, PhD | Marcus McGrath, MS | Mohammad Al-Hamdan, PhD

Paul Smith | Gokhan Inci, PhD, P.E. | James Demby

ABSTRACT

In the context of dam safety, understanding past dam failures through numerical modeling holds significant importance. This not only helps unravel the complexities and consequences of historical incidents but also plays a crucial role in enhancing our preparedness for potential future events. This study used the Decision Support System for Water Infrastructural Security Lite (DSS-WISE™ Lite) web-based system to investigate two historical dam breach incidents: the 2005 Taum Sauk Upper Dam failure and the 2020 Edenville Dam and Sanford Dam failure. The numerical simulations were conducted using two-dimensional modeling techniques, considering distinct

scenarios for each incident. For the Taum Sauk Upper Dam failure case, simulations were performed for a sunny-day scenario, focusing on the breach of only the upstream dam in a two-dam series. The results showed a high level of agreement between the simulated and observed inundation extents, with an overall F statistic value of 84.1%. In the case of the Edenville Dam and Sanford Dam failure, simulations were executed for a wet-day scenario involving cascading failures of two dams. The cascading characteristic is currently not supported by the DSS-WISE Lite web-based version due to its restriction to breaching only one structure per simulation. To address this limitation, two approaches were employed to study the 2020 Edenville Dam and Sanford Dam cascading failure case. The first approach

used the DSS-WISE Lite web-based version in conjunction with a combined hydrograph technique. The second method involved using the full version of the DSS-WISE solver, which permits multiple breaches but is not publicly accessible. For this case, the results showed varying degrees of agreement between the simulated and observed extents, with certain areas demonstrating good agreement. The full version of the DSS-WISE model slightly outperformed the combined hydrograph approach in replicating observed extents.

INTRODUCTION

Although dams provide many benefits to society, dam failures have caused catastrophic floods, leading to loss of life, property damage, environmental degradation, and cascading failures (Altinakar, Matheu, & McGrath, 2009). Thus, dam-break modeling has become increasingly important to identify the impacts of potential dam failures. Numerical simulations of dam-break floods play a crucial role in the assessment of potential impacts of dam failures and in the development of effective mitigation strategies. These simulations involve use of mathematical models and numerical methods to predict the flow characteristics, including flow depth, flow velocities, flood arrival time, and inundation areas, resulting from a dam breach. The modeling process requires consideration of various factors, including the physical properties of the dam and reservoir, topography of the surrounding area, and the hydrological conditions at the time of failure. With advancement of computational tools and availability of high-resolution topographic data, numerical simulations have become powerful tools for predicting dam-break flood impacts and have been extensively used in engineering design, emergency management planning, and risk assessment.

Numerous two-dimensional modeling software products such as HEC-RAS 2D, FLO-2D, TELEMAC-2D, MIKE 21, MIKE FLOOD, and DSS-WISE Lite are available to simulate flood characteristics due to dam failure. Haltas, Tayfur, and Elci (2016) employed HEC-RAS and FLO-2D to investigate flood inundation in an urban area resulting from a possible failure of Urkmez Dam in Izmir, Turkey. Yakti et al. (2018) applied HEC-RAS 2D to simulate the flooding induced by the failure of the Way Ela Natural Dam. Dat, Tri, Truong, and Hoa (2019) used MIKE FLOOD to simulate and construct inundation maps for the downstream area of Tra Khuc – Song Ve River basin. Pilotti et al. (2020) tested HEC-RAS 2D and TELEMAC 2D against the discharge hydrographs measured

in a historical physical model to analyze the consequences of the hypothetical collapse of the Cancano I dam (northern Italy) and the propagation of the resulting dam-break wave along the downstream alpine valley.

The software products mentioned above differ in various aspects such as governing equations, numerical schemes, computation methods, ease of use, input customizability, and result format. DSS-WISE Lite, a web-based system developed by the National Center for Computational Hydroscience and Engineering (NCCHE) at the University of Mississippi, stands out for its ability to set up simulations quickly with a relatively small amount of user input. DSS-WISE Lite automatically generates input files using input data and national data sets. Salt (2019) noted that the DSS-WISE Lite runtime is incredibly fast and comparable in accuracy to similar HEC-RAS models. They only differed when modifications to the underlying terrain data were needed due to potentially flow-altering features, because back in 2019 DSS-WISE Lite did not include the capability to modify terrain when setting up a simulation. However, this capability has been added to DSS-WISE Lite since then, in a subsequent release in November 2021. Correa (2019) also found comparable simulation results between the HEC-RAS and DSS-WISE Lite models, particularly for inundation extents for a dam-break flow at Bidwell Lake Dam. Larrauri and Lall (2020) used DSS-WISE Lite to predict breach hydrographs and inundation areas for dams in the Cumberland River Basin and proposed a framework to complement Flood Insurance Rate Maps (FIRMs) to identify “hot spots” beyond the current dam hazard classifications using the analytic hierarchy process (AHP) as a ranking method.

DSS-WISE Lite has also been used for emergency simulations during the Oroville Dam spillway incident in 2017, dam safety emergencies in Texas during Hurricane Harvey in 2017, emergency dam-break flood simulations in Puerto Rico during Hurricane Maria in 2017, and emergency simulations during Hurricane Florence in 2018.

We present a two-dimensional analysis of two historical dam breach incidents: the 2005 Taum Sauk Upper Dam failure and the 2020 Edenville Dam and Sanford Dam failure. The former happened on a sunny day, and the latter occurred during a period of heavy rainfall. In both cases, dams in series were involved, with breach of only the upper dam in the Taum Sauk Upper Dam failure, and a cascade failure of both dams for the Edenville and Sanford Dams. The DSS-WISE Lite web-based system was utilized to study both events, while the DSS-WISE full version was also employed for the Edenville Dam and Sanford Dam failures.

MATERIALS AND METHODS

Site Description

and Failure Sequence

TaumSaukDamFailure

The Taum Sauk Pump Storage Powerplant (Figure 1), which is owned and operated by Ameren UE (now known as Ameren Missouri), was constructed in the early 1960s in Reynolds County, Missouri. The powerplant consists of a lower reservoir sited along the East Fork Black River and an upper reservoir formed by a kidney-shaped rockfill dike

approximately 50 to 87 ft high capped by a 10-ft-high concrete parapet wall set on crest that is 12 ft wide (Rogers, Watkins, & Chung, 2010). The plant uses reversible turbines that operate as pumps pumping water into the upper reservoir during off-peak hours and releasing it from the upper reservoir to the lower reservoir to generate electricity during high-demand periods. The upper reservoir, at approximately 95 ft in depth, covers a surface area of roughly 55 acres, with a capacity to hold nearly 1.5 billion gallons of water.

According to the Federal Energy Regulatory Commission (FERC) Taum Sauk Investigation in 2006, the upper reservoir

Figure 1 Physical Layout of the Taum Sauk Pumped Storage Scheme; the Red Circle Represents the Breach Location

Jeff Spooner, USGS

Figure 2 Edenville and Sanford Dam Locations and Images of (a) Edenville Dam Breach (Coleman, 2020), (b) Edenville Dam After the Failure (Google Earth, 2022), (c) the Fuse Plug Spillway of Sanford Dam Being Overtopped (CFI Media (Checkered Flag Interactive Media), 2020), and (d) the Entire Embankment of Sanford Dam Being Overtopped (Michigan Drone Services)

The dam failures occurred near the end of a 3-day rain event starting on May 16, 2020 (Figure 2). On May 19, 2020, at approximately 5:35 p.m. EDT, when a section of the eastern side of Edenville Dam collapsed suddenly during the rainfall event that resulted in a record-high water level within the impounded Wixom Lake (France et al., 2022). Much of Wixom Lake drained into the Tittabawassee River and flowed into Sanford Lake. At about 7:19 p.m., the Sanford Lake level rose above the crest of the fuse plug spillway at the Sanford Dam, and the fuse plug began to erode. At around 7:45 p.m., the fuse plug began to wash out, and there was an increase in releases from the gated spillway at Sanford Dam. However, the combined releases from the fuse plug spillway and the gated spillway were insufficient to prevent the continued rise of the Sanford Lake level, leading to an overtopping breach failure of the Sanford Dam at about 8:20 p.m. The combined outflows from the failure of Edenville Dam and Sanford Dam and runoff from the natural watershed inundated downstream areas of Midland County and Saginaw County. The flooding resulted in widespread damage and destruction of buildings, homes, roads, utility infrastructure, and natural resources. Approximately 11,000 residents were successfully evacuated with no serious injuries or loss of life. More than 4,000 structures were reportedly impacted by the floodwaters, with estimated losses of roughly $245 million (Galvin 2020).

Data Sources

Terrain elevation data used in this study is sourced from the 3D Elevation Program (3DEP) 1-m resolution Digital Elevation Model (DEM) available on the United States Geological Survey (USGS) website, with a reported vertical accuracy root-mean-square error (RMSE) of 0.53 m. The DEM was resampled to resolutions of 1-m and 6-m cell sizes for simulations of Taum Sauk Dam failure and Edenville and Sanford Dam failure, respectively. The impact of the cell sizes on simulation results is briefly discussed in the results section, but further insights about that can be found in Pophet et al. (2023), who studied the sensitivity of the model to cell size and other input parameters. The presence of levee crest elevations in the DEM varies depending on data sources and the specific region. In some cases, the DEM includes comprehensive levee information. However, in certain regions or data sets, the DEM may not explicitly represent levees due to limited availability of detailed data or lower resolution. In the DSS-WISE system, automated input data preparation modules seamlessly integrate levees from the National Levee Database (NLD) into the DEM.

Flow resistance experienced by water as it travels through a flood area is represented by Manning’s n-value. The Manning values are estimated based on engineering judgment using land cover data from the National Land Cover Database (NLCD), which classifies land into 21 different types.

Numerical Models

The DSS-WISE Lite web-based system’s simulation engine (DSS-WISE solver) solves the conservative form of shallow water equations that govern the flood propagation over complex topography. The solver uses a finite volume method to discretize the shallow water equations over a regular Cartesian computational grid, with the elevation represented at the center of the cell. The fluxes between adjacent grid cells are computed using a Godunov-type upwind scheme based on the approximate solution of generalized Riemann problems at each cell interface using the first order Harten-Lax-van Leer-Contact (HLLC) Riemann solver. The HLLC method handles wetting/ drying and does not require entropy fixes to avoid nonphysical solutions. It captures shocks and allows the coexistence of mixed flow regimes (subcritical, transcritical, and supercritical) in the computational domain. The solver is parallelized and takes advantage of the multicore architecture of modern computers. Special programming techniques further increase the computational speed by tracking and computing only wet cells. These techniques are described in Altinakar et al. (2012, 2017).

Initial Conditions

Input parameters used in the simulations are presented in Table 1.

Simulation Setup

TaumSaukDamFailure

The simulation setup for the Taum Sauk Dam failure event did not account for any inflow discharge into the reservoirs or downstream channels, as it occurred on a sunny day. It was assumed that the channels were dry at the time of the dam breach. The simulation ran for 24 hours with a cell size of 1 meter.

EdenvilleandSanfordDamFailure

The Edenville and Sanford Dam failure occurred during a period of heavy rain and was a cascading failure, where the Edenville Dam failed first, causing the Sanford Dam downstream to overtop and fail. To simulate this event, it is necessary to use a model that can accommodate a wet-day scenario and be able to

3 Breach formation time after model calibration.

4 Breach formation time reported by Team, FERC Taum Sauk Investigation, 2006.

TABLE 1 INPUT PARAMETERS FOR THE TAUM SAUK UPPER RESERVOIR DAM FAILURE AND THE EDENVILLE AND SANFORD DAM FAILURE

handle multiple breaches. It is noted that the DSS-WISE solver, which is a full version of the web-based system, has this capability. However, it is not publicly available. The current web-based version of DSS-WISE Lite does not have direct support for such situations. Therefore, to study this event, the model was set up in two different ways. The first approach involved using a combined hydrograph in which the DSS-WISE Lite web-based version can be used. The second approach involved using DSS-WISE full version.

Usingacombinedhydrograph

To conduct the simulation, a hydrograph-type simulation approach was used, where a combined hydrograph was input into the DSS-WISE Lite model at the Sanford Dam breach location. This simulation is referred to as Sim3. The combined hydrograph (Hyd6) used in Sim3 was produced by summing up four hydrographs (Hyd1, Hyd2, Hyd3, and Hyd4; see Figures 3a, b, d, and e). The resulting hydrograph (Hyd6; see Figure 3f) was then used as input in the model.

• Hyd1 was generated from a simulation of the Edenville Dam failure that was carried out using a reservoir-type simulation. This simulation is referred to as Sim1. Sim1 assumed that only the Edenville Dam had breached, and a sunny-day scenario was used, which means no additional inflows into the lake and downstream channels. Wixom Lake was assumed to be at its maximum capacity at the time of failure, and the Edenville Dam was breached at time 0, corresponding to May 19, 2020, at 5:35 p.m. To collect data from Sim1, an observation line was placed at the Sanford Dam (see Figure 2) to measure the discharge hydrograph (Hyd1) for subsequent use.

• Hyd2 was generated through a reservoir-type simulation of the Sanford Dam failure. This simulation is referred to as Sim2. In Sim2, a sunny-day scenario was assumed. Unlike Sim1, Wixom Lake was not considered, and only Sanford Lake was filled up to its maximum capacity. The Sanford Dam was assumed to have breached at time 0, corresponding to May 19, 2020, at 8:20 p.m., 2.45 hours after the Edenville Dam breached. An observation line was placed at the Sanford Dam (see Figure 2) to measure the discharge hydrograph (Hyd2).

• Hyd4 was obtained from a spillway gate discharge at Sanford Dam. According to France et al. (2022), the spillway gates at Sanford Dam were opened throughout the day on May 18, beginning at 7:00 a.m. Ayres Associates (2021) applied HEC-HMS to calibrate spillway gate outflows from Sanford Dam and calibrated the model by comparing the model’s calculated outflows to the outflows indicated by the opertors’ log records. Their calibrated model’s outflows are presented in Figure 3(d) and were used in the simulation (Sim3).

• Hyd5, representing streamflow at Pine River near Midland, Michigan, was obtained from USGS station 04155500. There are two other streams, Chippewa River and Salt River, flowing into Tittabawassee River dowstream of Sanford Dam, but there is no stream flow data available for these two rivers at the time of the events. To approximate the stream flow for these two locations, the stream flow data from USGS station 04155500 were used. It is assumed that the discharges from these three streams were released at Sanford Dam location. Therefore, the discharge from this step was calculated as 3 × Hyd5 and combined with the previous mentioned discharges. This approximation allows for a more complete representation of the total stream flow that would have been discharged during the event. However, it is important to note that this approximation may not fully capture the complexity of the actual flow conditions and should be interpretted with caution.

• Hyd6 was obtained from summing up Hyd1, Hyd2, Hyd4, and 3 × Hyd5. The resulting hydrograph was then used in Sim3, where it was placed at Sanford Dam location. This indicates that the combined discharges from the various sources, including the Edenville and Sanford Dams, rainfall runoff, and stream flow, were taken into account in the simulation. Typically, hydrographs would undergo transformation when located at different downstream locations, but in this study, no such adjustments were made.

UsingDSS-WISE

The simulation model for the Edenville and Sanford Dam failure was started on May 17, 2020, at 12:00 midnight, about 65.58 hours before the first dam failure, and is referred to as Sim4. The model domain’s upstream section includes Wixom Lake, which is impounded by the Edenville Dam. The model assumed that inflows into Wixom Lake were equivalent to the outflows released by the gated spillway (Hyd3 in Figure 3c), along with the water filling up the lake. Therefore, in the simulation, Wixom Lake was filled to its maximum capacity without inflows, and the dam spillway discharges were located immediately downstream of the dam. The water in Wixom Lake was kept untouched until it was released due to the Edenville Dam breaching at time 65.58 hours in the simulation, corresponding to May 19 at 5:35 p.m.

At the start of the simulation, Sanford Lake was filled to its normal operation level. During the simulation, water was both being filled into Sanford Lake from Wixom Lake and released through its gated spillway. To simulate this process, a control release option with a single area hydrograph in the DSS-WISE solver was used. Two hydrographs were placed, one

upstream and one downstream of Sanford Dam. The upstream hydrograph had a negative discharge equal to the positive discharge provided downstream (Hyd4 in Figure 3d), ensuring that the water taken from Sanford Lake was the same as the water released downstream. Sanford Dam was set to breach at 68.33 hours, corresponding to May 19, 2020, at 8:20 p.m., which was 2.45 hours after the Edenville Dam breached. To account for tributary inflows, a stream flow hydrograph (Hyd5 in Figure 3e) was placed at three locations along the Salt River, Chippewa River, and Pine River.

RESULTS

Validation of the Flood Prediction

TaumSaukDamFailure

The release of water from the reservoir due to embankment failure can be analyzed by examining the breach outflow hydrograph and the reservoir level over time. Figure 4 depicts the calculated and simulated outflow hydrographs and reservoir stages for Taum Sauk Upper Dam breach. The calculated outflow

Figure 3 (a) Hyd1: Discharge Hydrograph From Sim1 (Edenville Dam failure) at Sanford Dam, (b) Hyd2: Discharge Hydrograph From Sim2 (Sanford Dam failure) at Sanford Dam, (c) Hyd3: Spillway Outflow Hydrograph at Edenville Dam (Ayres Associates, 2021), (d) Hyd4: Spillway Outflow Hydrograph at Sanford Dam (Ayres Associates, 2021), (e) Hyd5: Inflow Hydrograph From Tributary at USGS Gage Station 04155500, and (f) Hyd6: a Combined Hydrograph Used as Input Hydrograph in Sim3

hydrograph (Team, FERC Taum Sauk Investigation, 2006) was computed at 1-minute intervals on December 14, 2005, from 5:15 a.m. until the reservoir was mostly empty at 5:50 a.m. The change in stage for each one-minute interval was interpolated on the stage-storage curve to a volume in acre-feet per minute and then converted to cubic feet per second (cfs). In the simulation, an observation line was defined at the crest line of the dam, and the breach outflow discharge was computed by dividing the line into equidistant points and calculating the discharge at each point using the nearest velocity vector. The total discharges crossing the observation line were obtained by integrating the discharges computed at each point.

The calculated outflow hydrograph has a zig-zag shape, which corresponds to the outflow expected from the loss of sections of the parapet wall and multiple cantilever collapses of panels during the event. These factors were not considered in the simulation. According to FERC Taum Sauk Investigation (2006), the time to fully formed breach was 0.33 hour (20 minutes). By using this breach formation time, the simulated hydrograph had a peak discharge of 155,247 cfs at about 5:33 a.m.. However, by trial and error, it was found that a time to fully formed breach of 0.18 hour (10 minutes) provides a peak discharge of 272,887 cfs at 5:25 a.m., which is closer to the peak discharge of 273,000 cfs at 5:23 a.m. reported by Team, FERC Taum Sauk Investigation, 2006. As depicted in Figure 4, the simulated reservoir stage matches very well with the observed stage obtained from the Druck pressure transducers.

where A obs and A mod represent the inundation extent of the observed and modeled data, respectively. F varies between 0 for model with no overlap between observed and predicted inundation areas and 100 for a model where these coincide perfectly. In addition to F value, the commission (C) and omission (O) metrics (Bates & De Roo, 2000) were used to evaluate the performance of the simulated results relative to the observed maximum flood extent. They are calculated as follows:

where Pe is the number of common flooded cells, Pt is the total number of model wet cells, and Pu is the total number of observed wet cells. The commission metric measures the percentage of area simulated to be inundated that is not observed to be inundated, also known as false positives. In other words, it represents the amount of overestimation in the simulated flood extent. The lower the commission, the better the model performs. The omission metric measures the percentage of observed flooded area that was not simulated by the model, which is known as the false negative rate. This metric indicates how much of the actual flood extent was missed by the model. A lower omission error indicates a better performance of the model.

Table 2 shows that DSS-WISE Lite simulation demonstrated a high level of agreement with the observed inundation extent, as indicated by an overall F statistic value of 84.1%. Of the total flood extent, only 12.2% was overpredicted (commission), and 4.7% was underpredicted (omission). In comparison, Kalyanapu et al. (2011) used Flood2D-GPU to study the same event and obtained an F statistic value of 75.1% with a commission value of 15.3% and an omission value of 13.1%. Similarly, Judi (2009) employed FIT2D, achieving an F statistic value of 76.3% with a commission value of 16.6% and an omission value of 9.5%.

To quantify the deviation of simulated results from observations, three metrics are used. The first metric is a measure of fit, the F statistic presented by Bates and De Roo (2000):

To gain insight into how simulation results are affected by different cell sizes, simulations were conducted using various cell sizes: 1 m (results presented in Table 2), 3 m, 6 m, and 10 m. As the cell size was increased, the F statistic decreased from 84.1% to 83.5%, 82.2%, and 80.7%, respectively, indicating just a slight reduction in accuracy, while the simulation run time was shortened from 60.1 minutes to 24.2, 3.1, and 0.9 minutes, respectively. In other words, increasing cell size reduces simulation run times; however, this reduction comes at a slight expense of accuracy, particularly in specific areas, such as immediately downstream of the breach. Further detailed

Figure 4 Simulated and Observed Breach Outflow Hydrographs and Reservoir Stages for Taum Sauk Breach

information regarding the sensitivity of the model to cell size and other input parameters can be found in Pophet et al. (2023).

Figure 5 illustrates that the simulated results are generally in good agreement with the observed inundation extents, except certain locations where the simulation either over- or underpredicted the observed inundation extents. However, it can be inferred that these deviations were due to changes in local topography or flow conditions that were not captured in the model. For instance, the overestimation and underestimation immediately downstream of the breach may have been caused by changes in flow resistance or direction due to obstructions such as debris, vegetation, or other factors. It is noted that the DEM used in the simulation was generated after this event, and the terrain may not be identical to the pre-event conditions. The underestimation near the western end of the confluence and at the entrance to the lower reservoir may have been due to changes in channel geometry or sediment transport. The reasons for the overestimation at the northern side of the confluence and the eastern side of the lower reservoir are unclear. Although there were no apparent obstructions or changes in the DEM at those locations, it is possible that the field observations were limited to that extent.

The arrival time of the flood could not be validated due to lack of observed data. However, Rydlund (2006) conducted a post-event modeling study of the dam break that included flood wave routing from the upper reservoir embankment failure to the spillway of the lower reservoir using the dynamic wave unsteady flow models Dam Break (DAMBRK) and Unsteady NETwork (UNET). According to Rydlund’s simulations, the flood wave was estimated to have entered the floodplain of East Fork Black River in approximately 5.5 to 6 minutes and entered the lower reservoir 29 minutes after the breach. In comparison, DSS-WISE Lite simulated the flood wave entering the East Fork Black River at 6 minutes after the breach and entering the lower reservoir

at 28 minutes after the breach. Overall, the comparisons of the simulated results with the observation and other modeling efforts suggest that DSS-WISE Lite reasonably simulates the breach hydrograph, flood inundation, and flood arrival time.

Figure 5 Comparison of Simulated Inundation Extent (Gray Fill) With Observation for the Taum Sauk Dam Breach

EdenvilleandSanfordDamFailures

It is essential to compare the simulated discharge with USGS stream gage data to evaluate the accuracy of the dam breach model. Figure 6 shows the simulated and measured time series of discharge hydrographs at the Tittabawassee River in Midland, Michigan, which is about 10 miles downstream of Sanford Dam. The discharge was measured using a stream gage at this location during the event. The simulated discharge before the arrival of the dam breach wave at this location is composed of two sources: the spillway release and stream flow from three tributaries. These were included in the simulation to represent the natural flow conditions before the dam breach event occurred. The simulation results show that the contribution from these sources alone underestimated the observed discharge, indicating the presence of additional water sources, such as unknown streams or other sources of water flow, that were not accounted for in the model. The measured peak discharge was 51,800 cfs on May 20 at 11:00 a.m. In contrast, the simulated peak discharges using two different approaches were slightly different from the observed peak discharge. The peak discharge simulated with the combined hydrograph approach model (Sim3) was 54,187 cfs on May 20 at 07:48 a.m., while the peak discharge simulated with the full version DSS-WISE model (Sim4) was 51,583 on May 20 at 07:12 a.m.. However, the recession limb of the simulated hydrograph using Sim3 was almost the same as the observed one, indicating a close agreement between the simulated and measured data. The recession limb of the simulated hydrograph using Sim4 was slower than the observed one, suggesting that the flow was slower. This could be due to different flow resistances in the model. Overall, comparing the simulated and measured discharge at the USGS stream gage station helps evaluate the accuracy of the dam breach models.

The assessment of flood extent is crucial for understanding the impact of water disasters, and satellite imagery provides valuable insights. For instance, the Dartmouth Flood Observatory (DFO, 2020) used Landsat-8 imagery from May 20, 2020, to map the flood extent. The observed inundation area was then compared with the simulated extents produced by two modeling approaches: the combined hydrograph method and the full version of the DSS-WISE solver. The results showed that the simulated extents demonstrated some level of agreement with the observation. The simulation with the combined hydrograph approach (Sim3) had F statistic value of 39.5%, with 41.7% overpredicted and 44.9% underpredicted. The low F statistic value could be due to the accuracy of the observed data.

The simulation with the full version DSS-WISE (Sim4) had an F statistic value of 47.1%, with 49.3% of the extent overpredicted and 13.3% underpredicted. Although the overall F values were not high, certain areas showed good agreement between the simulated and observed extents, for instance, around Midland and near Bay City, as seen in Figure 7(a-b, e-f). However, the simulation underpredicted the observation at Saginaw, possibly due to the need for calibration of the flow resistance used in the model to define Manning’s roughness for a particular land cover type.

The accuracy of flood extent mapping using satellite imagery can be affected by several factors, including cloud cover, shadow, and land cover type. Vegetation cover can also impact the accuracy of flood extent mapping, as it may hinder the satellite's ability to detect the presence of water. In areas with dense vegetation, the satellite may not be able to penetrate through to detect flood water, which could result in an overprediction of the flood extent in the simulations. Figure 8 demonstrates this phenomenon, where the simulated extents overpredicted the flood in areas with dense vegetation, and the actual flood extent may not have been detected by the satellite imagery.

Figure 6 Simulated and Observed Discharge Hydrographs at the USGS Gage Station 04156000 Tittabawassee River at Midland, MI

CREATING BETTER DAM SAFETY SOLUTIONS, TOGETHER.

Engineering Design

Risk Analysis

Hydraulics & Hydrology

Geotechnical & Seismicity

Construction Services

Physical & Cyber Security

Public Safety & Emergency Management

Regulatory Compliance

Figure 7 Landsat-8 Satellite Imagery From May 20, 2020 (Lucey, 2020), and Simulated (blue) and Observed (black solid line, DFO, 2020) Extents From the Same Day

CONCLUSION

This study conducted a two-dimensional numerical modeling simulation using the DSS-WISE Lite web-based system to analyze two historical dam breach incidents: the Taum Sauk Upper Dam failure and the Edenville Dam and Sanford Dam failure. The simulations were performed as a sunny day and considered a single breach of two dams in series for the Taum Sauk Upper Dam failure and as a wet day for the Edenville and Sanford Dam failure, which involved a cascading failure. The study compared the simulated results with the observed inundation extent from the field measurement and the Dartmouth Flood Observatory using satellite imagery to evaluate the accuracy of the DSS-WISE Lite model in

simulating the dam breach events. The study also analyzed the breach outflow and discharge hydrographs and compared them with the measurements.

The simulation of the Taum Sauk Upper Dam failure showed a high level of agreement with the observed inundation extent, with an overall F statistic value of 84.1%. The deviations between the simulated and observed results were attributed to changes in local topography or flow conditions that were not captured in the model. The results of the Edenville Dam and Sanford Dam failure study showed that while the overall F statistic values for the simulated extents were not high, certain areas exhibited good agreement between the simulated and observed extents. It is important to note that the accuracy of

Figure 8 Simulated (red) and Observed (blue, DFO, 2020) Extents

flood extent mapping using satellite imagery can be influenced by various factors, such as cloud cover, shadow, and land cover type. The simulated extents using the full version DSSWISE model demonstrated slightly better agreement with the observed extents compared to the simulated extents produced by the combined hydrograph approach. The simulated discharge hydrograph at the USGS stream gage station near Midland agreed well with the measured peak discharge, time to peak, and the shape of the hydrograph.

The study findings suggest that the DSS-WISE Lite model can be a useful tool for risk assessment and emergency management planning for potential dam breach events. Further improvements in the model's calibration and validation, such as incorporating actual rainfall runoff and flow resistance data, can lead to better accuracy in simulating discharge hydrographs and flood extents. Overall, this study contributes to the ongoing efforts toward improving the understanding and management of potential dam breach events.

ACKNOWLEDGMENTS

We gratefully acknowledge the funding from the U.S. Department of Homeland Security (US-DHS) Science and Technology Directorate and the Federal Emergency Management Agency (FEMA) National Dam Safety Program, which have supported the original development of the DSS-WISE Lite and DSS-WISE Web since 2011, and its subsequent improvements and operations and maintenance.

References

Altinakar, M. S., Matheu, E. E., & McGrath, M. Z. (2009). New generation modeling and decision support tools for studying impacts of dam failures. Proceedings of the Association of State Dam Safety Officials (ASDSO) Dam Safety 2009 Annual Conference, 3, pp. 1256–1288.

Altinakar, M. S., McGrath, M. Z., Matheu, E. E., Ramalingam, V. P., Seda-Sanabria, Y., Breitkreutz, W., & Yezierski, M. (2012). Validation of Automated Dam-Break Flood Simulation Capabilities and Assessment of Computational Performance. Proceedings of the Association of State Dam Safety Officials (ASDSO) Dam Safety 2012 Annual Conference, 1, pp. 549–571.

Altinakar, M., McGrath, M., V.P., R., & Demby., J. (2017). Web-Based, Automated Two-Dimensional Dam and Levee Failure Flood Modeling and Mapping. Proceedings of the Association of State Dam Safety Officials (ASDSO) 2017 Southeast Regional Conference, (pp. 1–19).

Ayres Associates. (2021). Design Flood Hydrologic Analyses: Secord, Smallwood, Edenville, and Sanford Dams. Ayres Associates (Ayres).

Bates, P. D., & De Roo, A. P. (2000). A simple raster-based model for flood inundation simulation. Journal of Hydrology, 236(1–2), 54–77.

CFI Media (Checkered Flag Interactive Media). (2020). Sanford, Michigan Dam Failing - May 19th 2020 Flooding. Retrieved from https://www.youtube.com/watch?v=1pfRqM2KyI4/

Coleman, L. (2020). Michigan dam failure caught on video. Retrieved from https://www.youtube.com/watch?v=Hc3u_CHVHJ8

Correa, E. S. (2019). Bidwell Lake Dam breach model and flood map analysis using HEC-RAS. Doctoral dissertation, Department of Civil Engineering, California State University, Sacramento.

Dat, T. T., Tri, D. Q., Truong, D. D., & Hoa, N. N. (2019). Application of mike flood model in inundation simulation with the dambreak scenarios: A case study of Dak-Drinh Reservoir in Vietnam. Journal of Earth Sciences and Engineering, 12(01), 60–70.

DFO. (2020). DFO Flood Event: 2020-USA-4915. Retrieved from https://floodobservatory.colorado.edu/Events/4915/2020USA4915.html

France, J. W., Alvi, I. A., Miller, A. C., Williams, J. L., & Higinbotham, S. (2022). Investigation of Failures of Edenville and Sanford Dams. Independent Forensic Team – Final Report.

Galvin, L. (2020). Nine months after devastating flooding, Central Michigan residents are left abandoned. Retrieved from World Socialist Web Site (WSWS): https://www.wsws.org/en/articles/2020/12/30/mich-d30.html

Google Earth. (2022). Retrieved from https://earth.google.com/web

Haltas, I., Tayfur, G., & Elci, S. (2016). Two-dimensional numerical modeling of flood wave propagation in an urban area due to Ürkmez dam-break, İzmir, Turkey. Natural Hazards, 81, 2103-2119.

Judi, D. R. (2009). Advances to fast-response two-dimensional flood modeling. Doctoral dissertation, Department of Civil and Environmental Engineering, University of Utah.

Kalyanapu, A. J., Shankar, S., Pardyjak, E. R., Judi, D. R., & Burian, S. J. (2011). Assessment of GPU computational enhancement to a 2D flood model. Environmental Modelling & Software, 26(8), 1009–1016.

Larrauri, P., & Lall, U. (2020). Assessing the exposure of critical infrastructure and other assets to the climate induced failure of aging dams in the US Final Report for the Global Risk Institute.

Lucey, R. (2020, July 10). Landsat 8 Imagery Products for the Midland, Michigan Dam Failure. Retrieved from https://maps.disasters.nasa. gov/arcgis/home/webmap/viewer.html?layers=1d10057145894462be99cfce43fdfc82

Pilotti, M., Milanesi, L., Bacchi, V., Tomirotti, M., & Maranzoni, A. (2020). Dam-break wave propagation in alpine valley with HECRAS 2D: experimental cancano test case. Journal of Hydraulic Engineering, 146(6), 05020003.

Pophet, N., McGrath, M., Al-Hamdan, M., Smith, P., Inci, G., & & Demby, J. (2023). Sensitivity analysis of input parameters to flood characteristics for historical dam failures using DSS-WISE Lite web-based system. Proceedings of the Association of State Dam Safety Officials (ASDSO) Dam Safety 2023 Annual Conference, (pp. 1–8).

Rogers, J. D., Watkins, C. M., & Chung, J. W. (2010). The 2005 upper Taum Sauk dam failure: A case history. Environmental & Engineering Geoscience, 16(3), 257–289.

Rydlund, P. H. (2006). Peak discharge, flood profile, flood inundation, and debris movement accompanying the failure of the upper reservoir at the Taum Sauk pump storage facility near Lesterville, Missouri. U. S. Geological Survey.

Salt, D. V. (2019). A comparison of HEC-RAS and DSS-WISE Lite 2D hydraulic models for a Rancho Cielito Dam breach. California State University: Sacramento, CA, USA.

Team, FERC Taum Sauk Investigation. (2006). Report of Findings on the Overtopping and Embankment Breach of the Upper Dam-Taum Sauk Pumped Storage Project. FERC No. 2277.

Yakti, B. P., Adityawan, M. B., Farid, M., Suryadi, Y., Nugroho, J., & Hadihardaja, I. K. (2018). 2D modeling of flood propagation due to the failure of way Ela natural dam. In MATEC Web of Conferences, 147, p. 03009.

Nuttita Pophet, PhD

Post-Doctoral Research Associate

National Center for Computational Hydroscience and Engineering (NCCHE),

The University of Mississippi

220 South Oxford Center, Oxford, Mississippi 38655 • (662) 915-7788 • nuttita@ncche.olemiss.edu

Nuttita Pophet is a post-doctoral research associate at NCCHE. Her expertise spans numerical modeling of tsunami wave propagation, wind-generated random sea phenomena, wave breaking in deep ocean waters, and dam-break flows of grain-water mixtures. During her time at NCCHE, she has contributed to multiple projects, including the Decision Support System for Water Infrastructural Security (DSS-WISE), focused on flood simulation, and the Web-Based Agricultural Integrated Management System (AIMS), aimed at watershed management. She has significant experience in hydraulic and hydrologic modeling.

Marcus McGrath, MS

Senior Research and Development Engineer

National Center for Computational Hydroscience and Engineering (NCCHE), The University of Mississippi

220 South Oxford Center, Oxford, Mississippi 38655 • (662) 915-7788 • mzmcgrat@ncche.olemiss.edu

Marcus earned his BS and MS at the University of Mississippi. During his time at the National Center for Computational Hydroscience and Engineering at the university, he has worked on numerous projects in the field of dam and flood safety, resilience, hazard mitigation, and numerical flood model development. He is the lead developer and technical administrator of DSS-WISE Web, a web-based, automated, dam-break flood inundation modeling and mapping system supported by FEMA and DHS S&T.

Mohammad Z. Al-Hamdan, PhD

Director and Professor

Director, National Center for Computational Hydroscience and Engineering (NCCHE), The University of Mississippi

2301 S. Lamar Blvd, Oxford, MS 38655 • (662) 915-3783 • mzalhamd@olemiss.edu

Dr. Mohammad Al-Hamdan is the director of the National Center for Computational Hydroscience and Engineering (NCCHE) and a professor of civil engineering at the University of Mississippi (UM). Before joining the UM in 2020, Dr. Al-Hamdan was a principal research scientist with the Universities Space Research Association at NASA Marshall Space Flight Center/National Space Science and Technology Center, where he worked for 21 years. He received his PhD and MS degrees in civil and environmental engineering from the University of Alabama in Huntsville. His research expertise includes remote sensing and GIS applications to environmental modeling and assessment for water, air quality, urban heat island, ecological, and public health applications. Throughout his career, he has conceived, led, and worked on numerous research projects funded and supported by NASA, EPA, NIH, USDA, USDHS, FEMA, DOD, USGS, CDC, NOAA, and USDOT with total grants of over $29 million, which resulted in numerous peer-reviewed journal, conference and technical report publications, and several national and international scientific achievement awards. They also resulted in developing and/or improving several impactful national decision support systems for water and environmental management and assessment, emergency management and response, and several other societal benefits.

Paul Smith

Coordinator of Computing Facility Operations

National Center for Computational Hydroscience and Engineering (NCCHE), The University of Mississippi

220 South Oxford Center, Oxford, Mississippi 38655 • (662) 801-2140 • cvpsmith@ncche.olemiss.edu

Paul Smith earned his BSc in computer science at the University of Mississippi in 1998. He joined NCCHE in May 1999 as a computing facilities specialist and was appointed to coordinator of computing facility operations in January 2001. His current research interests include parallel hardware systems (GPGPU), SMP and load-balancing systems, Linux networking, and network penetration testing. Throughout his time at NCCHE, Mr. Smith has also worked on various projects as DevOps engineer and systems programmer. These projects include the DSS-WISE (Decision Support System for Water Infrastructural Security) Web system, as well as the AIMS (Agricultural Integrated Management System) project.

Gokhan Inci, PhD, P.E., PEng, PMP

Civil Engineer, FEMA National Dam Safety Program

400 C Street SW, Washington, DC 20472 • (202) 436-1721 • gokhan.inci@fema.dhs.gov

Dr. Inci is a graduate of Bosporus University (BS), University of New South Wales (BEngSc), and Wayne State University (PhD). His major was in civil engineering, and he specializes in dams and other infrastructure. After 15 years of practice in industry analyzing, inspecting, or designing more than 70 water and waste retaining dams, including large international projects, he joined FEMA. He is contributing to the success of the program as an emergency manager, contracting officer representative, research lead, and subject matter expert.

James E. Demby, Jr., P.E.

National Design Engineer, U.S. Department of Agriculture Natural Resource Conservation Service 1400 Independence Ave SW, Washington, DC 20250 • (771) 233-6247 • james.demby@usda.gov

James Demby is the national design engineer for the USDA Natural Resources Conservation Service, where he provides leadership and direction for the agency’s policy and guidance dam safety engineering community of practice. He is a professional engineer registered in Virginia. Mr. Demby has over 30 years of federal service as a civil engineer. Over his career, he has worked in the field of dam safety for various federal agencies, including the U.S. Army Corps of Engineers, the USDA Forest Service, the Federal Emergency Management Agency, and the Department of the Army Headquarters.

ASDSO Peer Reviewers

This article was peer reviewed by James Gallagher, P.E. (MA Office of Dam Safety, Retired) and Sushil Chaudhary, P.E. (NM Office of the State Engineer).

Screening Tool for Predicting Drowning Potential at Low-Head Dams

ABSTRACT

Recirculating currents downstream from low-head dams can trap recreational river users where they are continuously and repeatedly impacted by the high-velocity flow coming over the dam and eventually drown. Case studies and previous laboratory research have shown that the tailwater level is a primary determinant of the danger, with medium tailwater depths common to a wide range of flows creating the nearly inescapable submerged hydraulic jump. This study investigates a set of low-head dams at which fatalities occurred to evaluate the effectiveness of a previously developed spreadsheet tool for identifying the dangerous flow range. The study considered

58 fatal incidents at 29 low-head dams across the United States and also describes site visits to six low-head dams in Utah where there was no history of fatalities. Discharge information was available for all sites from nearby gaging stations. Downstream channel slopes were estimated from the National Hydrography Dataset (NHD), and tailwater depths were estimated using the Manning equation with a uniform assumption of n = 0.030. With these settings, the algorithm made correct predictions in 75% of the cases. For the successfully modeled cases, flow-duration curves were developed and used to estimate the frequency of the predicted submerged hydraulic jumps. These dams were determined to have submerged hydraulic jumps for an average of 343 days per year, or 94% of the time.

Caleb McCurry, M.S., EIT | Tony L. Wahl, M.S., P.E. | Rollin Hotchkiss, PhD, P.E.

A fisherman casts his line below the Scott Street Dam in Des Moines, Iowa AP Photo/Charlie Neibergall

Similar frequencies were obtained using streamflow estimates obtained from the National Water Model (NWM) and the GEO Global Water Sustainability Initiative (GEOGLOWS) model (GEO stands for the Group on Earth Observations). Sensitivity of predicted results was tested using the channel roughness defined by the Manning’s n value, the downstream channel slope, and a streambed elevation adjustment factor to account for aggradation or degradation at the dam toe. Prediction success improved with increased channel roughness, decreased channel slope, or aggradation of the downstream channel above the dam toe. Accuracy decreased slightly with a negative streambed adjustment (degradation), decreased channel roughness, and increased channel slope. Further testing is recommended using field-derived values for inputs wherever possible.

Introduction

Low-head dams are a nearly ubiquitous feature on rivers and streams across the country, constructed for a wide range of purposes throughout history, such as irrigation diversion, recreation, small hydroelectric power, and thermal cooling, as well as to provide power for mills. Many of these simple dams, like the one shown in Figure 1, provide no significant storage function, pass flow continuously, and span the full width of a river channel from bank to bank without separate spillways, gates, or other structures (Federal Register, 1997). High dams are imposing to most river users, but structures smaller than 15 ft in height can appear relatively benign. In some cases, paddlers may even be attracted to the challenge of going over the dam while underestimating the hidden danger. For a range of tailwater conditions, flow over the crest of a low-head dam can create a submerged hydraulic jump downstream from the dam (a “hydraulic” in the parlance of many river users) that produces a strong, recirculating current that traps objects and people and repeatedly carries them back upstream into the high-velocity flow passing over the dam (similar to the trapped log shown in Figure 1). The continuity of the structures from bank to bank eliminates mid-channel refuge and hinders escape. With the growing popularity of outdoor recreation and the simultaneous growth of dam safety programs that have led to a reduction in dam failures, drownings at low-head dams are now accounting for most of the dam-related fatalities in the United States (Tschantz, 2014). The deceptive appearance of these structures can turn family outings to tragedy, and first responders and good Samaritans often become additional victims (Elverum & Smalley, 2012; Wahl & Svoboda, 2023).

The U.S. Army Corps of Engineers maintains the National Inventory of Dams (NID), a database of about 92,000 documented dams across the United States. However, the NID was developed primarily to understand flooding risk due to failure of large dams (uncontrolled release of the stored reservoir) rather than risks associated with regular daily operations of small dams. As a result, most low-head dams do not meet the minimum height or storage volume requirements of the NID and are not included. In response to the lack of data on low-head dams, a task force including professionals from the Environmental and Water Resources Institute of the American Society of Civil Engineers (ASCE), the Association of State Dam Safety Officials (ASDSO), the U.S. Society on Dams, American Whitewater, and American Rivers created a national inventory of low-head dams that documents more than 13,500 structures (Figure 2; Hotchkiss et al., 2023). Researchers at Brigham Young University (BYU) have also developed an online database of low-head dams where fatalities have occurred (https://krcproject.groups.et.byu.net/). As of January 2023, this list included 315 low-head dams with 437 associated fatalities.

Not all low-head dams are dangerous, nor are individual dams necessarily dangerous at all flow rates. To assess the danger presented by specific dams and flow conditions, Wahl and Svoboda (2023) developed a spreadsheet application that evaluates low-head dam dangers based on dam and river properties for a range of flow rates. The work presented here

Figure 1 Low-Head Dam on the Jordan River in Salt Lake City, Utah

(McCurry, 2023) tests this spreadsheet tool against a subset of the BYU fatalities database with known discharges. Anticipating potential application of the tool to the larger database being compiled by the low-head dams task force and wishing to avoid the need for individual site visits to thousands of dams, the study also considers the use of next-generation, GIS-based, online data sources for the necessary input data.

SubmergedHydraulicJumps

Flow over a typical low-head dam exhibits several zones with dramatically different flow conditions. Water passing over the crest transitions smoothly from subcritical to supercritical flow, as seen in Figure 1. Near the downstream toe of the dam, the flow usually returns to subcritical through a hydraulic jump. The tailwater depth downstream from the dam strongly affects the nature and location of the hydraulic jump.

For shallow tailwater depths, the hydraulic jump is swept downstream, away from the low-head dam; this is a Type A jump (Figure 3). This jump exhibits a violent appearance that usually deters river users from trying to swim or boat over the dam or wade or swim in the tailwater pool. When the tailwater depth matches the depth required to produce the jump (conjugate depth), the jump occurs at the base of the dam (Type B). For tailwater depths greater than the conjugate

Figure 2 National Inventory of Low-Head Dams, Alpha Version

Figure 3 Types A, B, C, and D Hydraulic Jumps (top to bottom) Wright et al. (1995) with permission from Kenneth Wright

Note: H is the depth of water (head) on the low-head dam crest; P is the height of the low-head dam above the downstream streambed; Y1 is the supercritical depth at the base of the low-head dam; Y2 is the conjugate depth required to produce a hydraulic jump; Yflip is the tailwater depth that submerges the low-head dam; YT is the actual tailwater depth as controlled from downstream; and ΔZ is the change in streambed elevation above or below the toe of the low-head dam.

depth (Type C), the hydraulic jump moves upstream onto the downstream face of the dam and submerges the supercritical flow coming down the dam face. In this condition, the jump appears less violent, but the submerged jump is characterized by an unrelenting recirculation pattern that carries objects near the surface upstream, toward the dam. This is the dangerous condition that traps victims. Finally, at very high tailwater levels (Type D), the trajectory of water passing over the dam is said to “flip”; that is, it detaches from the submerged downstream surface of the dam and instead is directed downstream along the water surface. In this condition, the jump is said to be “drowned out” and the trapping recirculation is gone, so that swimmers in the water are consistently carried downstream, away from the dam. Type D jumps are relatively safe for river users.

Dangerous conditions for a low-head dam primarily depend on the calculated conjugate depth Y2, the flip depth Yflip, and the actual tailwater depth YT. In the spreadsheet tool developed by Wahl and Svoboda (2023), Y2 is calculated from theoretical equations by Bélanger (1849) and empirical equations developed from laboratory testing by Leutheusser and Fan (2001). The flip depth is computed from an empirical equation developed by Leutheusser and Fan (2001), Yflip = 0.91(H + P), where H is the head above the dam crest and P is the dam height from crest to toe. YT is calculated using the Manning equation for normal

depth flow in a uniform channel. This equation is applicable when the downstream channel is uniform in size, shape, slope, and roughness for a sufficient distance to cause the flow depth to become a unique function of those variables. Figure 4 illustrates the important parameters of the dam and the hydraulic jump, including a variable ΔZ that accounts for any difference in elevation between the downstream channel bed and the toe of the dam due to long-term aggradation (increased bed level) or degradation (decreased bed level). The ΔZ term shown in Figure 4 is negative because the downstream bed is lower than the toe of the dam; a higher bed elevation than the toe produces a positive value of ΔZ so that the effective tailwater depth against the dam is greater than that calculated for the downstream channel by the Manning equation.

Using the input data, the spreadsheet determines the jump type and estimates the magnitude of the upstream-directed surface velocities in the tailwater pool based on the lab tests by Leutheusser and Fan (2001). Type A, B, and D jumps are considered safe, while all Type C jumps are considered dangerous, regardless of the calculated upstream velocity. Type C jumps at very low flow rates may present less danger due to weak upstream velocities and/or shallow flow depths at the toe of the dam, but all Type C jumps are considered dangerous, since a shallow flow at high velocity can still sweep

Figure 4 Low-Head Dam and Hydraulic Jump Parameters

Modified from Leutheusser (1991) and Wahl and Svoboda (2023)

a person off their feet. Drownings in a few inches of water have occurred when swimmers could not regain their footing and became disoriented by the violent spinning action of the flow. For a given dam, low flow rates typically produce Type A or B jumps, moderate to high flows produce Type C jumps, and very large flows produce Type D jumps. A large, positive value of ΔZ may create enough additional submergence that there is no range of Type A or B jumps; all jumps are submerged, and the dangerous zone begins at very low flow rates.

Methods

The spreadsheet analysis approach was tested by compiling a list of candidate dam sites where submerged hydraulic jumps were expected. Two data sources were used. First, the BYU database of low-head dam fatality incidents was screened to identify cases that:

1. Had at least one recorded fatality with a known incident date

2. Met the definition of a low-head dam, that is, water flows uncontrolled over the dam crest (not gatecontrolled) with no significant water storage volume

3. Had available discharge records from a nearby U.S. Geological Survey (USGS) stream gage for the date of the incident

4. Had a known dam height, either the hydraulic height listed in the NID (maximum reservoir water level above original stream bed) or a general height value from an alternate source such as accident reports

A total of 58 fatal incidents at 29 low-head dams met these criteria (Figure 5). Additionally, 13 site visits were made to six low-head dams in Utah that had never caused fatalities. The height of each of these dams was estimated during the site visits, and each site also had a nearby USGS gage with available discharge data. A Type C jump was observed at one of the Utah dams during three site visits, each with a different flow rate, and these were counted as additional dangerous cases. The other five Utah sites were each visited at low and high flow rates, and Type A swept-out jumps were observed each time and were thus classified as instances of safe conditions.

Additional data that were needed to apply the spreadsheet to each incident included the dam crest length, downstream channel width, bed slope, channel roughness, crest shape, and tailwater bed height. Because dam crest length was inconsistently reported in the NID, the crest length was measured using GIS tools (e.g., Google Earth) to ensure consistency. For locations where a dam had been removed or significantly modified since the incident, a measurement of the channel width at the original location was used for the crest length. Downstream channel width, which controls

Figure 5 Map Showing Low-Head Dams With Known Fatalities, Stream Gages at Dams With NID Records, and Low-Head Dams Used in This Analysis

tailwater level, was also measured using GIS tools at a distance of one crest length downstream from each dam to obtain the shortest distance bank to bank. Islands that were often present in the downstream channel were excluded from the channel width calculation. Bed slope downstream from each dam was obtained from the National Hydrography Dataset (NHD), which provides the minimum and maximum elevation and lengths of stream segments. Care was taken to correctly identify the first segment downstream from each dam. A potential source of error was choosing a stream segment that included the dam, since the elevation drop across the dam would distort the calculated stream slope. For locations where a simple downstream slope was not available, such as areas with exceptionally long stream segments or multiple stream channels, a stream segment was chosen that was most representative of the main downstream channel.

Lacking site-specific data sources, values for Manning’s n, the low-head dam crest shape, and ΔZ were all assumed. Manning’s n roughness values for the downstream channels were initially assumed to have the constant value n = 0.030 at all sites. Later sensitivity analysis considered a range of values. Possible crest shapes include ogee-shaped crests, rectangular

crests (i.e., broad-crested weirs), or shorter crests, such as concrete weir walls with a thickness much smaller than the water depth (head) on the crest. The laboratory experiments by Leutheusser and Fan (2001) used to estimate Yflip are based on models of sharp-crested weirs. While the spreadsheet allows selection of a sharp-crested weir or ogee-crest shape and accounts for some differences in expected head losses within the tailwater pool, results are generally insensitive to this parameter. The sharp-crest option was assumed for all sites in this study. Finally, the elevation difference ΔZ between the toe of each dam and the downstream channel bed was assumed to be zero. The impacts of this assumption were tested later.

The site parameters and flow conditions for each fatal incident and the Utah site visits were used in the spreadsheet to test whether a Type C submerged hydraulic jump was predicted. At locations where submerged hydraulic jumps were correctly predicted, flow-duration curves were calculated using gage data and streamflow estimates obtained from the National Water Model (NWM) and the GEO Global Water Sustainability Initiative (GEOGLOWS) model, and these were used to estimate the likely frequency of dangerous conditions occurring throughout a typical year.

Results

For the fatal incident cases, the spreadsheet result was considered accurate if a Type C jump was predicted.

For the Utah site visit cases, the result was accurate if the spreadsheet prediction matched the hydraulic jump condition observed during the site visits. The predictions were accurate for 75% of the modeled cases (Table 1).

Predictions were correct for all tested flow conditions at 24 of the dams, or 69% of the 35 dams considered (29 dams with fatal incidents and the six Utah dams).

For the fatal incident cases, the tailwater depth corresponding to the mean flow from the day of the incident was plotted in Figure 6 relative to the conjugate depth and flip depth for that same flow.

For the incorrect predictions, the tailwater was predicted above the flip depth for the shallow slope cases or below the conjugate depth for steeper slopes. Among the Utah site visits, the five dams with swept-out Type A jumps were correctly predicted for both low and high flow rates. The sixth Utah dam was observed to have a Type C jump at three different flow rates. The lowest of these was correctly predicted, but for the two higher flow rates, the spreadsheet predicted the tailwater to exceed the Yflip depth and produce a Type D jump.

Note: The tailwater depth for correctly predicted cases is indicated by a red X between the lower bound (conjugate depth) and the upper bound (flip depth). Incorrectly predicted cases are shown with a gray X for predictions of a flipped jet (Type D jump) or blue X for a swept-out jump (Type A).

ACCURACY OF PREDICTIONS OF HYDRAULIC JUMP TYPES

Figure 6 Tailwater Depths vs. Dangerous Tailwater Range for the 58 Fatal Incidents Sorted by Slope, Which Increases From Left to Right

NEXT-GEN DATA MANAGEMENT

Take your dam monitoring to the next level

MLWEB3

The all-new MLSuite 3rd generation data visualization tool for dams

In addition to predicting the submerged hydraulic jump condition for the flow rate matching each fatality incident or site visit condition, the spreadsheet analysis also readily shows the range of discharges that will produce dangerous Type C hydraulic jumps for each site. Considering only those sites with correct predictions of submerged hydraulic jumps (19 dams) and pairing this information with flow-duration curves developed from each site’s associated USGS stream gage data, the dangerous sites produced submerged hydraulic jumps for an average of 343 days per year. The analysis also showed that the dangerous sites all produced Type C jumps for the lowest 70% of flows; at even higher flow rates, the flip depth began to be exceeded for some structures, producing Type D jumps. For the highest 10% of flows, 15 out of 19 sites remained dangerous. For the top 1% of flows, only nine of the dams were dangerous.

Anticipating possible application of this approach to sites without a source of gaged streamflow data, the flow-duration analysis was also performed using data obtained from online resources provided by GEOGLOWS and the NWM. The flow-duration curves and results from these sources were similar to those obtained using gaged streamflow; NWM output predicted 343 days of dangerous conditions, and GEOGLOWS simulations predicted 344 days.

Discussion

Although the spreadsheet was successful for most cases, missed predictions may be the result of inaccurate assumptions underlying the analysis and/or inaccurate input data. The most significant assumption is the application of Manning’s equation, which requires uniform, normal depth flow in the downstream channel. Changes in stream slope, channel shape, channel width, or roughness along a reach produce gradually varied flow that does not match the Manning’s equation calculation of flow depth. Examination of the cases considered here via aerial imagery showed some changes in channel width for both accurately and inaccurately predicted cases. Other factors that could produce gradually varied flow are difficult to evaluate without detailed investigation, including site visits.

SensitivityAnalysis

The spreadsheet analysis is dependent on eight variables. Discharge values used in this study are probably accurate, since stream gage data were available near all sites. A sharpcrested dam shape and fixed values of Manning’s n = 0.030 and ΔZ = 0 were used in this analysis. The dam height, dam

crest length, and channel width were obtained from the NID or GIS-based resources. The dam height is probably the most uncertain of these, as it is difficult to observe from aerial imagery or even during site visits when flow is occurring and because there are a range of ways in which dam height may be measured and reported. Nevertheless, no sensitivity testing was performed for NID or GIS-based data, because results were shown to be relatively insensitive to these variables (McCurry, 2023). The channel slope estimates from the NHD are subject to significant uncertainty and could perhaps be improved by site-specific survey work. Sensitivity to channel slope was tested due to its impact on results (McCurry, 2023).

The effects of the ΔZ = 0 and n = 0.030 assumptions made for the fatality incident cases were examined in more detail, since these assumptions did not incorporate any site-specific information. Considering bed elevation changes of ΔZ = +1 ft or ΔZ = –1 ft, an increased bed elevation significantly improves prediction accuracy by shifting 9 of the 10 Type A swept-out jump cases into the Type C submerged jump category; only two cases shift from Type C to Type D jumps, so the prediction accuracy increases from 72 to 84%. In contrast, assuming a bed elevation decrease, ΔZ = –1 ft, 11 cases move from Type C to Type A jumps, while there is no change in the number of Type D cases. This reduces the prediction accuracy to 53%. Similarly, the effect of the Manning’s n on prediction accuracy was evaluated by considering n values from 0.015 to 0.060 at increments of 0.005. Increased channel roughness values of n = 0.040 to 0.045 optimizes the prediction accuracy by shifting 8 of the 10 Type A swept-out jump cases into the Type C submerged jump category; two cases are shifted from Type C to Type D jumps, so there is a net gain of eight more accurate predictions and the accuracy increases from 72 to 83%. The prediction accuracy decreases with smoother roughness values. For example, at n = 0.020, nine additional cases are predicted to have Type A jumps, but only two additional cases shift from Type D to Type C jumps, so the prediction accuracy drops to 60%. The sensitivity to n shows the potential value of using site visits and any available survey data to refine channel roughness estimates when possible.

The effect of channel slope on prediction accuracy was also evaluated. Reducing the slope to 90% of the value obtained from the NHD improves the prediction accuracy by shifting 3 of the 10 Type A swept-out jump cases into the Type C submerged jump category without shifting any cases from Type C to Type D jumps. This increases the prediction accuracy from 72 to 78%. In contrast, increasing the slope to 110% of the NHD value shifts three cases from Type C to

Type A jumps, while there is no change in the Type D cases. This reduces the prediction accuracy to 67%.

Input data for the dams visited in person in Utah were obtained by methods similar to those used for the fatal incident dams (slope estimated from the NHD, with ΔZ = 0 and n = 0.030 assumed), except for the dam height, which was estimated from observations made during the site visits. Five of the six sites in Utah had swept-out Type A jumps for all observed flow rates, and these were accurately predicted by the spreadsheet.

One site had a Type C jump for all three observed flow rates but was predicted to be drowned out with a Type D jump for the two highest flows. The channel slope estimated at this site was relatively low, S = 0.0002. Predictions for this site could be improved by changes that would reduce the tailwater depths, such as a negative ΔZ value, a lower n value, a steeper downstream channel slope, or a combination of these changes.

Conclusions

An effective hazard classification model for low-head dams should accurately predict the occurrence of dangerous conditions to help responsible parties effectively mitigate the danger and prevent future deaths. The spreadsheet analysis tested here was successful in 75% of cases and could potentially be improved with more accurate estimates of downstream bed slope, channel roughness, and local bed elevation differences around the dam toe. The dams with known fatalities that were included in this study were estimated to operate in dangerous flow ranges about 94% of the time based on flow-duration curves developed from either local streamflow data or from next-generation GIS-based streamflow data sources such as the NWM or GEOGLOWS. Further testing is recommended using field-derived values for inputs wherever possible.

References

Bélanger, J.B. (1849). Notes Sur le Cours D’hydraulique [Notes on a Course in Hydraulics], session 1849–1850. Paris, France: Mém. Ecole Nat. Ponts et Chaussées.

Brigham Young University. Locations of fatalities at submerged hydraulic jumps. Retrieved December 18, 2023, from https://krcproject. groups.et.byu.net/browse.php

Elverum, K. A., & Smalley, T. (2012). The drowning machine [Tri-fold pamphlet]. Minnesota Department of Natural Resources.

Federal Register. 2017. Vol 82(4): 1997.

Hotchkiss, Rollin H., Karl Kingery, & Kathleen Hoenke. (2023). National inventory of low-head dams: Completion and next steps. Proceedings, Dam Safety 2023, Association of State Dam Safety Officials conference, Palm Springs, CA, Sept. 17–20.

Leutheusser, H. J. & Birk,W. M. (1991). Drownproofing of low overflow structures. Journal of Hydraulic Engineering 117(2), 205–213.

Leutheusser, H. J., & Fan, J. J. (2001). Backward flow velocities of submerged hydraulic jumps. Journal of Hydraulic Engineering, 127(6), 514–517.

McCurry, C. (2023). Development of a submerged hydraulic jump prediction method using documented fatal incidents at low-head dams [unpublished master’s thesis]. Brigham Young University.

Tschantz, B., 2014. What we know (and don’t know) about low-head dams. Journal of Dam Safety, 12, 37–43.

U.S. Army Corps of Engineers National Inventory of Dams. https://nid.sec.usace.army.mil

Wahl, T. L., & Svoboda, C. D. (2023). A spreadsheet tool for defining dangerous flow ranges of low-head dams. Water, 15, 1032. https://doi.org/10.3390/w15061032

Wright, Kenneth R., Kelly, J. M., Houghtalen, R. J., & Bonner, M. R. (1995). Emergency rescues at low-head dams. Proceedings, 12th Annual Conference of the Association of Dam Safety Officials. 327-336.

Caleb McCurry

Research Assistant, Brigham Young University Provo, UT 84604 • (801) 404-0727 • cmccurry@byu.edu

Caleb McCurry is finishing his master’s degree at Brigham Young University in civil engineering with a focus on dam safety. Prior to graduate school, he worked for 3 years as a hydrologist for the Bureau of Reclamation in Boulder City, NV, where he worked on the Consumptive Uses and Losses Report for the Lower Colorado River Basin. Recently, he has accepted employment as an engineer for the Utah Division of Water Rights.

Tony L. Wahl, M.S., P.E.

Technical Specialist, Bureau of Reclamation, Hydraulics Laboratory (86-68560) PO Box 25007, Denver, CO 80225 • (303) 445-2155 • twahl@usbr.gov or tony.wahl@gmail.com

Tony Wahl has worked for 34 years at the Bureau of Reclamation Hydraulics Laboratory on a variety of hydraulic structures topics including flow measurement, fish screening, spillways, embankment dam breach, and soil erodibility measurements. He has recently been focusing on spillway issues, including water surface profile calculations, hydraulic jacking, and stepped spillways. In 2019 he was awarded the ASCE Hydraulic Structures Medal.

Rollin H. Hotchkiss, PhD, P.E., D.WRE, F.ASCE

Professor, Civil and Construction Engineering Department, Brigham Young University 430 Engineering Building, Provo, UT 84663 • (801) 422-6234 • rhh@byu.edu

Rollin H. Hotchkiss is a water resources engineering professor in the Civil and Construction Engineering Department at Brigham Young University. His research efforts to reduce fatalities at low-head dams are documented in several journal papers and conference proceedings. Current work on this topic aims to create user-friendly tools to determine the potential occurrence of submerged hydraulic jumps for any of the more than 13,000 low-head dams in a recently completed national inventory. He has authored more than 50 journal papers and more than 100 conference papers. He was the 2017 ASCE Hydraulic Structures Medal recipient.

ASDSO Peer Reviewers

This article was peer reviewed by Greg Paxson, P.E., D.WRE (Schnabel Engineering), Jim Gallagher, P.E. (MA Office of Dam Safety, Retired) and Bill Sturtevant. P.E. (Colorado Springs Utilities).

Dam Safety Inspections & Assessments

Dam Design & Rehabilitation

Dam Safety Inspections & Assessments

FERC Part 12 Inspections & Reports

Dam Design & Rehabilitation

Dam

Safety

Dam Safety is what we do

is what we do

Potential Failure Modes Analysis (PFMA)

FERC Part 12 Inspections & Reports

Semi-Quantitative Risk Analysis (SQRA)

Potential Failure Modes Analysis (PFMA)

Site-Specific PMP and PMF Services

Semi-Quantitative Risk Analysis (SQRA)

CFD Hydraulic Modeling

Site-Specific PMP and PMF Services

Dam Break Analyses & Inundation Mapping

CFD Hydraulic Modeling

Emergency Action Plans and Exercises

Dam Break Analyses & Inundation Mapping

Tainter Gate Inspections

Emergency Action Plans and Exercises

Penstock Inspections

Tainter Gate Inspections

Public Safety Plans

Penstock Inspections

Public Safety Plans

Utica, NY - Buffalo, NY

Albany, NY - Henniker, NH

Utica, NY - Buffalo, NY

www.gomezandsullivan.com

Albany, NY - Henniker, NH

315-724-4860

www.gomezandsullivan.com

315-724-4860

Rising Tide: The Great Mississippi Flood of 1927 and How it Changed America

by John M. Barry

Review by Nathaniel Gee, P.E. Manager Dam Safety Governance & Oversight, TVA ngee@tva.gov

“It is so much easier to believe than to think ; it is astounding how much more believing is done than thinking.”

– Rising Tide

Dams are intrinsically linked with levees. If you want to understand levees and levee policy in the U.S., you need to study the Mississippi River. No flood had a greater impact on levee and water policy than the flood of 1927. And no book outlines that history more expertly than John Barry’s Rising Tide.

John Barry effectively outlines the politics and people that led to managing the Mississippi floodplain before and after the great flood of 1927. How could it not be a great book when the first section of the book is titled “The Engineers”? To better illustrate the interesting aspects of this book, I’ll provide you with a summary of this section.

No one knew the Mississippi River better than James Buchanan Eads, who began work on the river when he was just a teenager. Rather than leave St. Louis when his drunk father ran out of town, he decided to stay and work on a steamboat instead. By age 22, he had saved enough to have a recovery barge of his own design and started a business in salvaging. He also designed his own diving bell and spent day after day pulling out brokendown ships – even walking along the bottom of the turbulent river. He quickly became a rich man, but not without great cost. He was on this river when his son died, and then a short time later missed the death of his wife.

His success extended to shipbuilding, building several for General Ulysses S. Grant. His boats were key in securing Grant’s victory at Fort Henry and Fort Donelson – the first major victories of the Civil War. This gave Eads a personal relationship with Grant – one he would need years later.

In the mid-1800s, there was no position considered as elite as a civil engineer (some things never change). West Point had the only formal engineering training in the U.S., and only the top two cadets of each class were allowed to enter the U.S. Army Corps of Engineers. One of the most renowned civilian engineers at the time was Charles Ellet. Having studied in France, he was one of the only European-trained engineers in the U.S. Although he had many engineering feats, nothing would make him as famous as the metal rope he strung across Niagara Falls with a basket, becoming the first human to

cross. The federal government commissioned a survey of the Mississippi River and an engineer named Andrew Humphreys was determined to make it his claim to fame; however, many engineers demanded Congress commission a civilian instead of someone from the Corps – particularly Ellet.

The $50,000 appropriation was split in two, and both men were asked to do separate and independent reports. Each took the job seriously, and Humphreys worked harder on it than anything he had in his life. But as he was nearing completion, he fell so ill he almost died. And while Humphreys was ill, Ellet finished his survey.

One prominent theory beginning to gain popularity in the basin was the “levees only” theory – the idea that levees alone could be built deep enough to hold back the entire Mississippi. Ellet dismissed this as a dream and said the only possible way

Egremont, Mississippi rail station on May 2, 1927.

to keep the Mississippi from regular flooding would require limiting side inflow during storms, with large storage basins, and selectively choosing locations to create natural outflows from the river.

Ellet’s report outraged Humphreys – not because he disagreed (much of his data aligned with Ellet’s conclusions) but by completing his report first, Ellet took the glory Humphreys felt belonged to him. It was almost a decade later in 1861 that Humphreys would submit his report, and in order to ensure his views were taken over Ellet’s, he made sure to point out where Ellet was wrong. Even though it contradicted much of his own data, Humphreys pushed strongly a "levees only” theory. Ellet, having died fighting for the Union in the Civil War, was unable to defend himself. Humphreys became Chief of the Engineers in the Corps and could begin deploying the actions he had recommended, namely a “levees only” policy.

The economy was run by the men who owned transportation at that time, and there was a battle for which mode of transportation would be “king.” Rail was quickly beginning to dominate, but to really progress, they needed to put railroad bridges across the Mississippi. The first bridge that crossed the river was quickly destroyed when it was struck by a steamboat. The steamboat operators were quick to stop any future bridges from being built. Eads was smart enough to expand his empire in both realms. He got the job to build a railroad bridge. Since it crossed the waterway of the U.S., he needed the approval of Humphreys. The bridge was reviewed and accepted, but at the last minute, the permit was pulled due to pressure from steamboat operators.

Eads tried to get the permit worked out but was rebuffed at every turn. When no one could provide him with a good explanation, he marched into President Ulysses S. Grant’s office and asked why the permit was revoked. Grant called in his Secretary of War, who tried unsuccessfully to explain. Before they left the office, Eads had a permit to build the bridge. Eads quickly went to work, but less than a month later, the Corps pulled the permit again. This time, Eads just ignored them and kept constructing. The bridge became a successful crossing of the Mississippi.

The Mississippi River carried thousands of tons of sediment daily down to the delta, and all this sediment made the gulf very difficult to enter. Boats often had to wait days, even weeks, for the river to rise high enough for them to travel (or pay exorbitant prices for a tug). There was pressure for the U.S. government to solve the sediment issue. For years, the Corps (with Humphreys as the lead) advocated for a canal that could be built, up to 18-feet deep and regularly dredged. The canal would only allow one boat to cross at a time and was estimated to cost $13 million.

Eads told Congress he would build the jetties. The theory was if you decreased channel width, the increased velocity would “self-dredge.” He guaranteed a 28-foot-deep shipping channel wide enough for multiple ships and said he could do it for $10 million. Humphreys, fearing he couldn’t get Congress to say no, pressured them to offer Eads $5 million, assuming the amount was impossible. Eads took the offer. But in order for Eads to get paid, he had to pass a survey controlled by the Corps.

Eads quickly went to work and spent his capital. By the time he was down to 16-feet deep (when he was to receive his first payment), he was desperate. The only issue was that the

Andrew Atkinson Humphreys, Chief of the Corps of Engineers in 1866.

United States Army Corps of Engineers

Corps had snuck a ship through and supposedly already did the survey, and someone had leaked to the papers the channel was not sufficiently deep. Eads was outraged, demanding they let him know what data they had collected or redo the survey when he was there to verify it was done fairly. The Secretary of War refused. Investors began to pull out. It looked like he’d go bankrupt before the Corps even published their survey. Eads was determined to save the project. He paid for the largest boat he could get, one with an 18-foot draft, and invited the press and investors on a ride into the Gulf. Many were convinced it would only end in disaster, which increased press coverage. His internal survey showed they were at least 16-feet deep, and he hoped 18-feet deep. The boat safely passed into the Gulf.

Eads successfully made the passage over 30-feet deep and got his $5 million, a price he would have lost significant money on if it wasn’t for his great business acumen. Knowing the new passage would create great demand for a rail/boat connection, he bought a portion of the control of the railroads that went to the newly created ports.

This is just one of the nine sections of the book. Each one is full of interesting facts and historical figures – whether it’s the elite in New Orleans, a senator from Mississippi, or the Secretary of Commerce (who would eventually be President and have one of the world’s most iconic dams named after him). Barry shows how all tie together and impacted the floods of 1927. This book is both enlightening and entertaining. It is impossible

to read this book and not see the influence of the 1850-1927

Mississippi Delta on our world and industry today. For anyone involved in water policy or dam/levee safety, it is a must-read.

The ASDSO Quarterly Book Club is an ongoing review of books that may be pertinent to ASDSO membership. New or old, the hope is to share a reading list and promote lifelong learning in the dam/levee safety industry.

Please share feedback or future book suggestions to lee.mauney@hdrinc.com or check out our ASDSO Collaborate site.

Brady-Handy photograph collection, Library of Congress, Prints and Photographs Division.

James Buchanan Eads, Civil Engineer

Each quarter, ASDSO recognizes one ASDSO volunteer or volunteer group in the Journal of Dam Safety. Through this recognition, ASDSO hopes to spotlight some of the outstanding efforts being made by our members and thank them for their contributions. The ASDSO Annual Awards Committee oversees this effort, and the Board of Directors selects honorees.

ASDSO Board of Directors Recognizes Greg Paxson, Chair of the Technical Journal Editorial Committee

The Journal of Dam Safety is one of the top benefits of ASDSO membership and it serves as an important resource for those in the industry. To continue to meet these high standards, each issue of the journal undergoes months of detailed preparation before publishing. For the past three years, Greg Paxson has led this effort as the chair of the Technical Journal Editorial Committee.

Producing the journal is a year-round commitment with little time, if any, between issues. To meet the quarterly schedule, it is common for the committee to have multiple issues in various stages of development at the same time. Greg plays a vital role in keeping each issue moving through production while also ensuring the quality of the articles. For each issue, he oversees author recruitment, peer-review oversight, and on-going coordination between authors, staff, and the committee.

The Journal has expanded in many ways under Greg’s leadership. He has led the development of multiple themed issues, where all technical articles center around a common topic. Examples include last issue’s focus on estimating life loss from flooding and the popular dam failures issue published in Summer 2022. Under his leadership, the committee also implemented a new cover art concept and the integration of new non-technical content such as the quarterly book club.

In addition to his contributions to the journal, Greg has contributed to ASDSO through the Advisory Committee, including previously holding the chair position, and through the Annual Conference Program Committee.

Thank you, Greg, for making a difference!

Greg Paxson, PE, BC.WRE Schnabel Engineering

NORTHEAST REGIONAL SPOTLIGHT

Each quarter, the ASDSO Regional Representatives from one region recognize an individual, organization, or group that has made outstanding contributions to dam safety in their region or nationally as a representative from their region. The ASDSO Annual Awards Committee oversees the effort. If you have an idea for a regional spotlight that you would like to be considered, please email awards@damsafety.org.

2023 Vermont Floods Highlight Dam Safety Collaboration

The State of Vermont suffered a widespread rain event in July 2023 with three to nine inches of rain across the state and four to eight common along the steep, mountainous terrain of the Green Mountains. The resultant extreme flooding tested the state's approximately 1,200 dams and the dedicated Vermont Dam Safety team.

In Montpelier, the state capital, floodwaters rose to within 10 inches of the auxiliary spillway crest of the 115-foot high Wrightsville Dam, located upstream from the city, resulting in a flood-of-record for the flood control dam, constructed in 1935. The Vermont Dam Safety program, also located in Montpelier, took action to triage both state-owned and regulated dams across the state to ensure public safety all while dealing with post-event impacts to their homes and communities.

As in similar widespread storm events, the dam safety community rallied to support the Vermont program. ASDSO Executive Director Lori Spragens arranged a call with the program, Bill McCormick (Colorado Dam Safety, Black & Veatch), and Jill Stewart (South Carolina Dam Safety) to share advice on response, recovery, and lessons learned after major storm events in their respective states. Ben Green (Vermont Dam Safety) noted recommendations to develop an After-Action Report to describe and memorialize the event, actions taken, challenges faced, and opportunities for improvement were very helpful.

The Vermont Dam Safety program immediately began to develop and implement a Rapid Inspection Program to target 400 dams and determine their post-flood status in approximately two weeks. This included the development and launch of a smartphone and GIS-based inspection application to document findings. Under the Emergency Management Assistance Compact (EMAC) program, which is the national interstate mutual aid agreement that enables rapid deployment of shared resources during times of disaster, the Vermont Dam Safety program was able to deploy nine

inspection teams with support from the New York Power Authority, Massachusetts Office of Dam Safety, and New York Department of Environmental Conservation - Dam Safety.

The teamwork displayed through the rollout and implementation of the Rapid Inspection Program revealed five failed dams (2 significant and 3 low hazard), approximately 50 dams with notable damage, and 57 dams that overtopped. Without the teamwork and support of the broader dam safety community, these great accomplishments would not have been possible.

Clarks Sawmill Dam, Cabot, LOW Hazard

East Calais Mill Dam, East Calais, HIGH Hazard

Save the Date for Dam Safety 2024

Save the date and make plans to join us for Dam Safety 2024, September 22-26, 2024, at the Colorado Convention Center in Denver, CO.

The conference agenda and registration information will be available in May 2024.

Recognize Excellence & Submit an ASDSO Award Nomination

ASDSO annually honors those individuals and organizations making exemplary contributions to the improvement of dam safety in the U.S. through its awards program. The nomination window for the 2024 awards program is now open, and ASDSO is accepting nominations for seven of its awards:

NATIONAL REHABILITATION PROJECT OF THE YEAR

REGIONAL DAM SAFETY AWARDS

MEDIA OUTREACH AWARD

BRUCE TSCHANTZ PUBLIC SAFETY AT DAMS AWARD

YOUNG PROFESSIONAL OF THE YEAR

TERRY L. HAMPTON MEDAL

DANNY MCCOOK MEDAL

Nominations are due May 3, 2024. Award recipients will be recognized in numerous ASDSO publications and at a luncheon in conjunction with Dam Safety 2024. Full award information, nomination forms, and information on past recipients are available at Damsafety.org/Awards.

ASDSO Members at ASCE Legislative Fly-In

In February, ASDSO members Keith Conrad, John Roche, Gregory Daviero, John Moyle, and Clint Oman visited Washington, DC, to participate in the American Society of Civil Engineers (ASCE) Congressional Fly-In. Together, they met with over 25 House and Senate offices to promote ASCE and ASDSO's legislative initiatives, which include reauthorization of the National Dam Safety Program and addressing difficulties surrounding the State Assistance and High Hazard Potential Dams Rehabilitation grants programs. Thank you to each of them for continuing ASDSO's work to highlight the importance of dam safety for members of Congress.

(Left to Right) Cody Holt (ASDSO Government Affairs Manager), John Moyle (DE Dam Safety), Greg Daviero (Schnabel Engineering), Clint Oman (NJ Dam Safety), Keith Conrad (NV Dam Safety)

(Left to Right) New Jersey Dam Safety Program Representatives John Ciccarelli, Clint Oman, Ian McKane, and Bilal Awan

(Left to Right) Greg Daviero (Schnabel Engineering), Chris Carter (Legislative Committee Chair for ASCE New York State Council), Congressman Molinaro, Eva Lerner-Lam (ASCE Transportation Policy Committee), and Eleanor Lamb (Senior Manager of Federal Government Relations at ASCE)

(Left to Right) Keith Conrad (NV Dam Safety), Representative Mark Amodei, Dr. Shashi Nambisan (University of Nevada Las Vegas), Humberto Franco Rivas (Clack County Public Works), David Swallow (Southern Nevada RTC)

Partner News

U.S. ARMY CORPS OF ENGINEERS:

The National Levee Safety Program, a joint program between USACE and FEMA currently under development, released the draft first edition of the National Levee Safety Guidelines on April 1, 2024. These guidelines, developed with stakeholder input, are intended to serve as a resource of best practices to help achieve nationwide consistency in improving the reliability of levees and resilience of communities behind levees throughout the United States.

Comments on the draft National Levee Safety Guidelines must be submitted by July 31. View the guidelines and learn how to submit feedback at LeveeSafety.org.

DHS CYBERSECURITY AND INFRASTRUCTURE SECURITY AGENCY:

Registration is open for the 2024 Dams Sector Information Sharing Drill. It is hosted by the Cybersecurity and Infrastructure Security Agency (CISA) as the Sector Risk Management Agency (SRMA) for the Dams Sector, and serves as a learning environment for sector stakeholders to identify and assess informationsharing issues and concerns that could potentially affect the sector during heightened threat conditions. During the drill, conducted virtually, a detailed scenario will stress-test the effectiveness of the sector’s information-sharing mechanisms, processes, and procedures.

The 2024 Dams Sector Information Sharing Drill is open to all sector stakeholders representing owners and operators (e.g., operations, emergency managers, and engineers), first responders responding to dam and levee incidents, and others that support the sector (e.g., associations, regulators, consultants). With a scenario focused on both cyber and physical aspects, the 2024 Dams Sector Information Sharing Drill provides an opportunity for organizations to identify a wide range of players with safety and security (physical and cyber) expertise necessary for testing relevant plans (e.g., Emergency Action Plan, Cyber Incident Response Plan).

The drill will be held May 7-8, with a kickoff meeting on April 30. To register, visit https://bit.ly/3xdCEjJ.

Membership Appreciation Day is May 31

The third annual ASDSO Membership Appreciation Webinar will be on May 31, 2024, in conjunction with National Dam Safety Awareness Day. This year’s webinar, “Dam Perspectives—Building a Career in Dam Safety,” will be held from 12:00 PM to 2:00 PM Eastern. Watch your membership newsletter for registration information and resources for National Dam Safety Awareness Day outreach.

ASDSO Releases 2023 State Performance Reports

The 2023 State Performance Reports can now be downloaded on the individual state pages on ASDSO's website. Reports include 2022 data on condition ratings, state staffing, state budgeting, EAP completion, a comparison to the national benchmark, and more. State Performance Reports are a useful tool to educate the public, media, and other stakeholders on specific state regulatory programs and the dams they regulate.

To view the reports, visit DamSafety.org/States.

The member companies of the Dam Safety Group offer proven solutions for the evaluation and monitoring of safety and subsurface ground characterization issues for new build or existing Dams and related hydrological projects, together with market-leading real-time seismic Earthquake Early Warning Systems supported by a wide range of alerting technologies.

• Non-invasive or destructive techniques

• Regular monitoring of the Dam

• Early warning detection of geotechnical problems

• Assists the design and build of remedial projects

• Assists to assess remedial work after completion

• Identifies damaged areas within the body of the Dam

• Locates fractures, voids and zones of seepage

www.damsafetygroup.com

Resistivity Imaging Seismic Tomography

Ground Penetrating Radar (GPR)

Wireline Borehole Logging Seismic Monitoring

ADVERTISING RATES AND INSERTION

The Journal of Dam Safety accepts advertising. Because of its status as a charitable, educational institution and because of postal laws regarding postage rates for nonprofit organizations, ASDSO cannot accept advertising from insurance, travel, and credit card companies.

ASDSO Sustaining Members receive 25% off Full and Half page ads.

For complete information on advertising rates, visit www.damsafety.org/advertise, or contact Ross Brown, rbrown@damsafety.org, (859) 550-2788.

Advertising Index

*As of March 1, 2024

» Learn about becoming a Sustaining Member at DamSafety.org/Sustaining

UNIT CONVERSIONS

It is intended that articles in the ASDSO Journal of Dam Safety be in English measurements. In order to assist our international partners in dam safety, we also want to include the international SI system of measurements.

Length Conversions

1 Inch (in) = 25.40 Millimeter (mm)

1 Feet (ft) = 30.48 Centimeter (cm)

1 Feet (ft) = 0.3048 Meter (m)

1 Yard (yd) = 0.9144 Meter (m)

1 Mile (mi) = 1.609 Kilometer (km)

Mass Conversions

1 Pound-Mass (lbm) = 453.6 Gram (g)

1 Pound-Mass (lbm) = 0.4536 Kilogram (kg)

1 Ounce (oz) = 28.35 Gram (g)

1 Slug = 14.59 Kilogram (kg)

Force Conversions

1 Pound-Force (lbf) = 4.448 Newton (N)

1 Kip = 1000 Pound-Force (lbf)

1 Ton = 2000 Pound-Force (lbf)

1 Ton = 4.448 Kilonewton (kN)

1 Kip = 4448.2 Newton (N)

Pressure and Stress Conversions

1 Ibf/Inch2 (psi) = 6.89 kilopascal (kPA)

1 Atmosphere (atm) = 1.013x105 Newton/meter2

Area Conversions

1 Inch2 (in2) = 6.451 Centimeter2 (cm2)

1 Foot2 (ft2) = 0.0929 Meter2 (m2)

1 Yard2 (yd2) = 0.836 Meter2 (m2)

1 Mile2 (mi2) = 2.590 Kilometer2 (km2)

1 Mile2 (mi2) = 640.0 Acre

Volume Conversions

1 Inch3 (in3) = 16.39 Centimeter3 (cm3)

1 Foot3 (ft3) = 0.0283 Meter3 (m3)

1 Yard3 (yd3) = 0.764 Meter3 (m3)

1 Pint = 0.473 Liter (L)

1 Gallon = 3.785 Liter (L)

1 Acre-Foot (acre-ft) = 1233 Meter3 (m3)

Velocity Conversions

1 Feet/Second (fps) = 0.3048 Meter/Sec (m/s)

1 Miles/Hour (mph) = 1.609 Kilometer/Hr (km/s)

Flow Conversions

1 Gallons/Minute (gpm) = 0.0022 Foot3/Second (cfs)

1 Acre-feet/Second = 1233.48 Meter3/Sec (cms)

1 x106 Gallons/Day (mgd) = 1.547 Foot3/Second (cfs)

Temperature Conversions

n Celsius (°C) = ([°Fahrenheit]-32)/1.8

n Kelvin (°K) = ([°Fahrenheit]+459.7)*(5/9)

1 Millimeter (mm) = 0.04 Inch (in)

1 Centimeter (cm) = 0.03 Feet (ft)

1 Meter (m) = 3.281 Feet (ft)

1 Meter (m) = 1.094 Yard (yd)

1 Kilometer (km) = 0.621 Mile (mi)

1 Gram (g) = 0.0022 Pound-Mass (lbm)

1 Kilogram (kg) = 2.2046 Pound-Mass (lbm)

1 Gram (g) = 0.04 Ounce (oz)

1 Kilogram (kg) = 0.07 Slug

1 Newton (N) = 0.2248 Pound-Force (lbf)

1 Pound-Force (lbf) = 0.001 Kip

1 Pound-Force (lbf) = 1x10-4 Ton

1 Kilonewton (kN) = 0.1020 Ton

1 Newton (N) = 2.23x10-4 Kip

1 Pascal (Pa) = 1.45x10-4 Ibf/Inch2 (psi)

1 Newton/meter2 = 9.87x10-6 Atmosphere (atm)

1 Centimeter2 (cm2) = 0.1550 Inch2 (in2)

1 Meter2 (m2) = 10.76 Foot2 (ft2)

1 Meter2 (m2) = 1.196 Yard2 (yd2)

1 Kilometer2 (km2) = 0.386 Mile2 (mi2)

1 Acre = 0.0016 Mile2 (mi2)

1 Centimeter3 (cm3) = 0.0610 Inch3 (in3)

1 Meter3 (m3) = 35.31 Foot3 (ft3)

1 Meter3 (m3) = 1.308 Yard3 (yd3)

1 Liter (L) = 2.113 Pint

1 Liter (L) = 0.2642 Gallon

1 x106 Meter3 (m3) = 811 Acre-Foot (acre-ft)

1 Meter/Sec (m/s) = 3.281 Feet/Second (fps)

1 Kilometer/Hr (km/s) = 0.6214 Miles/Hour (mph)

1 Foot3/Second (cfs) = 450 Gallons/Minute (gpm)

1 Meter3/Sec (cms) = 35.32 Foot3/Second (cfs)

1 Foot3/Second (cfs) = 0.65 million gallons per day

n Fahrenheit (°F) = [°Celsius]x1.8+32

n Fahrenheit (°F) = [°Kelvin]x(9/5)-459.7