Dam Failure Concepts and Modeling

1.0 Introduction

Nagarjuna Sagar Dam across the Krishna River in India

Dams have been part of human civilization for thousands of years and have one basic function – to impound water – though the reasons for this vary greatly. Dams are used for flood control, water supply, and hydropower in a variety of forms. The total number of actual water impoundments is essentially uncountable since they range in size from small ponds to huge reservoirs.

Usoi Dam, a Natural Landslide Dam Along the Murghab River in Tajikistan

In addition to human-engineered dams, dams also form through natural processes such as landslides, glacial moraines, ice, and animal activity.

Dam failures and the resulting floods have been around as long as dams have. Although small dams fail more frequently, the greatest concern is with large dams above densely populated regions.

Kariba Dam, a Hydroelectric Dam in the Kariba Gorge of the Zambezi River Basin Between Zambia and Zimbabwe

This lesson describes the characteristics of large dams and modes of failure including the roles of dam type and age. In the process we will discuss terminology associated with dams and dam breaches. We present simulations of dam failure impacts using the U.S. Army Corps of Engineers’ Hydrologic Engineering Center, River Analysis System (HEC-RAS) model, which is used throughout the world.

Homepage of the Hydrologic Engineering Center,  River Analysis System (HEC-RAS) website

Four simulations will provide examples of how the flows out of a reservoir are sensitive to the characteristics of a dam breach. The characteristics include breach size and the time it takes a breach to form. The simulations are based on the HEC-RAS model run for a dam in the U.S. state of Pennsylvania. Although they are based on the characteristics of that dam, the results illustrate important considerations common to dams around the world.

After completing this lesson, you should be able to:

  • Identify the characteristics of large dams and modes of failure associated with them.
  • Match reservoir and watershed characteristics with the type of model most likely needed to produce a forecast hydrograph during a major dam failure.
  • Explain the changes to headwater, tailwater, and outflow at a dam during a dam failure and how breach size and formation time affect these.
  • Describe how HEC-RAS model simulations can be used to explore the impact of breach size and breach formation time on the resulting downstream flows.
  • Describe the basic characteristics of downstream inundation and hydrographs associated with dam failures.

2.0 Dams and Dam Failure

2.0 Dams and Dam Failure » 2.1 Large Dams

Question

Which variables do you think are used to define a large dam?

The correct answers are a, b, d, f.

The definition of a large dam may be based on the size, storage capacity, flood discharge, and unusual nature of its design.

Please make a selection.
Defining a large dam according to the International Commission on Large Dams (ICOLD)

The World Register of Dams maintained by International Commission on Large Dams (ICOLD) lists nearly 40,000 “large” dams. These dams are at least 15 m in height (measured from the lowest point of the foundation to the top of dam), or those between 10 and 15 m in height that meet at least one of the following conditions:

  • The crest length is not less than 500 m
  • The capacity of the reservoir formed by the dam is not less than 1,000,000 m3
  • The maximum flood discharge is not less than 2000 m3/s
  • The dam had especially difficult foundation problems
  • The dam is of unusual design

Here are the top countries with the greatest number of large dams:

Country

# Dams

China

23,842

United States of America

9,265

India

5,102

Japan

3,116

Brazil

1,392

Korea (Rep. of)

1,305

Canada

1,166

South Africa

1,114

Spain

1,082

Turkey

976

2.0 Dams and Dam Failure » 2.2 Large Dam Failures

Fujinuma Dam failure in Sukagawa, Fukushima

Given the number of dams worldwide, the number of dam failures is relatively low. Small earthen dams have the greatest failure rate due to poor design, construction, and the materials used.

Considering just large dams, the graphic shows the three main causes of failures from 1900 to 1973 according to ICOLD.

Common dam failure modes according to the International Commission on Large Dams (ICOLD)

Overtopping, the most frequent event leading to dam failure, occurs when water pours over the crest of the dam. Although excess runoff into the reservoir is usually a contributing factor, concurrent factors include:

  • Inadequate spillways
  • Debris blockages at the spillway
  • Structural failure, such as settling of the dam
  • Equipment malfunction, such as gate failures

Foundation failure, often due to structural defects, is the second most common mode of failure and accounts for a little over half of concrete dam failures. Foundation failures may be triggered by sinking and settling of the structure due to inferior materials, water seepage beneath the dam, or other problems.

Piping is almost always associated with fill dams. In piping, seepage through the structure causes internal erosion.

Urban runoff

Contributing factors common to large dam failures are:

  • Extreme runoff into the reservoirs from rain, snowmelt, or an upstream dam failure (especially for failures due to overtopping)
seismogram from St Paul Island, Alaska following Nov 15 2006 earthquake in the Kuril Islands
  • Natural phenomenon, such as earthquakes and landslides. A special case is a glacial lake outburst floods (GLOF) where glacial dams (ice-dammed or moraine-dammed) may fail leading to a large release of water.
  • Human error, including poor maintenance or poor design
Remains of St. Francis Dam and Reservoir Floor, Los Angeles County, CA, USA, March 1928

2.0 Dams and Dam Failure » 2.3 Failure for Concrete and Fill Dams

Studies provide failure modes for specific types of dams. Fill dams, sometimes referred to as embankment dams, are the most common type of large dam to fail. A fill dam is a type of gravity dam that relies on the weight of the dam structure to keep it in place.

Conceptual diagram showing an embankment (fill) dam.

Fill dams can suffer from all three main modes of failure, overtopping, foundation defects, and piping.

Concrete dams are less likely to fail but when they do, it typically happens earlier in the dam’s expected lifetime. Concrete dams can be gravity dams, like fill dams, or buttress or arch dams.

Conceptual diagram showing types of concrete dams

2.0 Dams and Dam Failure » 2.4 Dam Failure and Age

Question

Relationship of age, construction material, and failure mode for large dams that failed

According to this graphic that shows the percent of dam failures (of those that have failed) relative to their age, construction material, and mode of failure, which statements are accurate?

The correct answers are b, d.

The statistics used to make this graphic are for dams that have failed. Recall that concrete dams are less likely to fail and can be around for a very long time.

The figure on the left shows dam failure due to overtopping. For concrete dams that failed, 50% occurred in the first few years and 100% occurred within 25 years. For fill dams that failed, 50% failed within the first 10 years, about 75% within 30 years.

The middle figure shows that about 50% of all dams that fail from foundation problems do so in the first year, and even the first few months in the case of concrete dams. All concrete dams in the study that failed from foundation problems did so within 5 years. Fill dams that failed due to foundation problems did so within 12 years.

The main points are that:

  • Concrete dams are less likely to fail than fill dams
  • Overall, dam failures tend to occur early in the expected life of a dam; that’s when material and design flaws are likely to become apparent
  • Concrete dams that fail tend to fail earlier than fill dams
  • Foundation problems tend to occur earlier than other modes of failure
  • Foundation failures may occur during reservoir filling
  • Piping is not a typical mode of failure for concrete dams
Please make a selection.

2.0 Dams and Dam Failure » 2.5 Notable Dam Failures Around the World

2.0 Dams and Dam Failure » 2.5 Notable Dam Failures Around the World » 2.5.1 Banqiao Dam and Shimantan Reservoir Dam near Zhumadian China

Banqiao Dam

Location, River

China, Huai River Basin

Year built

1951-1952

Dam type & material

Embankment, clay fill

Dam dimensions

Height: 24.5 m, Length (crest): 2020 m

Reservoir capacity

492,000,000 m3 (375,000,000 m3 flood storage)

Maximum discharge capacity

1742 cms (estimated 1000 year storm event)

Event triggering failure

Extreme rainfall and runoff, upstream dam failure

Age at failure

23 years

Mode of failure

Overtopping

Breach dimensions

Nearly full depth, 350 m wide

Time to failure

2.25 hours

Outflow from Dam

Peak at 78,000 cms (reservoir drained in 6 hours)

Downstream effects

6 m high, 12 km wide flooding. Downstream dams failed. Flooding 250 km downstream in 1.5 days.

Dam failure toll

170,000-230,000 fatalities

Photo of the Banqiao Dam Collapse in China, Aug 1975

China’s Banqiao Dam was a clay-core fill dam that suffered a rapid full-depth failure due to overtopping in August 1975. It was the largest in a cascade of about 60 dams that failed or were intentionally destroyed along the Huai River due to extreme runoff triggered by a 3-day intense rainfall. This is considered the world’s most catastrophic dam failure with an estimated 170,000-230,000 fatalities.

The breach took out an area 350 m wide and extended vertically through the full depth of the dam and formed in just 2.25 hours. Peak discharge from the dam was about 78,000 cms, or about 45 times the estimated 1000-year event that the dam was designed to pass.

The dam was constructed as a fill dam with a clay core in 1951-1952. Cracks and other construction flaws quickly developed and were repaired. In early August 1975, the remnants of typhoon Nina produced record rainfall in the Huai River basin, with as much as 1060 mm (41.7 inches) in a 3-day period. On the evening of 7 August, the upstream Shimantan Reservoir Dam failed and sent a surge of water to the Banqiao Dam reservoir. The Banqiao Dam breach began around 1:00 AM on 8 August and fully developed in 2.25 hours. The reservoir nearly completely drained in about 6 hours.

Numerous dams and water diversions failed downstream. By the evening of 9 August, floods had reached the Fuyang Area in the Anhui Province, roughly 250 km (156 miles) away.

The Machchhu II Dam failure near Morbi, India on 11 August 1979 was a similar event.

These both are examples of a large volume overtopping breach that develops very rapidly through the full depth of the dam.

Photo of failed Machchhu II dam, which resulted in 1979 Morbi flood disaster

2.0 Dams and Dam Failure » 2.5 Notable Dam Failures Around the World » 2.5.2 Vajont Dam, Italy

Vajont Dam

Location, River

Italy, tributary to River Piave

Year built

1957-1960, filling began in 1960

Dam type & material

Arch, concrete

Dam dimensions

Height: 262 m, Length (crest): 160 m, Width (base): 27 m

Reservoir capacity

150,000,000 m3

Construction notes or concerns

Geologically unstable slopes enhanced by soil saturation from filling the reservoir; significant slope movement noted in weeks leading up to slope failure

Event triggering failure

Filling of reservoir and accompanying soil saturation

Age at failure

3 years

Mode of failure

Landslide (the dam actually survived but a wave of water overtopped it)

Time to failure

Very fast (minutes)

Downstream effects

Flood wave down steep narrow canyon and around the shore of the reservoir

Dam failure toll

2000 fatalities

Photo of Vajont disaster on 9 October 1963 due to a huge landslide from Monte Toc into the alpine lake basin below

The Vajont Dam “failure” is one of the most unique dam-related disasters in history because the dam structure, a concrete arch style, did not actually fail. Instead, a massive, landslide-triggered impulse wave overtopped the dam, causing a catastrophic downstream flood. It is typically included as a “dam failure” because the presence of the dam and reservoir triggered the landslide.

Vajont Valley is in the Italian Alps about 100 km north of Venice. This deep, glacial valley was recognized as a good site for hydroelectric power. The Vajont Dam was completed in 1960 along a minor tributary of the River Piave. At the time, it rose 261.6 m above the valley floor, making it the tallest concrete arch dam in the world. Its total capacity was 150,000,000 m3.

The geologic structure of the valley was concerning since conditions were suitable for slope failures. As the reservoir filled, newly saturated soils surrounding the reservoir mobilized clay layers.The first landslide occurred as the reservoir was filling in 1960. Filling slowed, and the slopes seemed to stabilize somewhat until 1963 when an increase in reservoir levels resulted in increased slope movement. By early October 1963, the slope displacement was 20 cm/day.

At 10:39 PM on 9 October 1963, 270,000,000 m3 of debris slid at 30 m/s into the reservoir resulting in an impulse wave with an estimated volume of 20,000,000 to 25,000,000 m3. It destroyed structures around the reservoir and overtopped the dam by up to 245 m. Downstream impacts lasted mere minutes but were extremely devastating. Surprisingly, the dam survived and it now holds the slide debris instead of water.

UNESCO cited the Vajont Dam tragedy as a cautionary tale of failure of engineers and geologists. Although the landslide and overtopping caused the downstream flood, the overall cause was poor decisionmaking.

This is a special case that was not really a breach. But it would likely be modeled as and a very rapid catastrophic breach. The steepness of the downstream valley should also be taken into account when modeling downstream effects of a dam failure.

2.0 Dams and Dam Failure » 2.5 Notable Dam Failures Around the World » 2.5.3 St. Francis Dam, United States

St. Francis Dam

Location, River

San Francisquito Canyon, California U.S.A., San Francisquito Creek

Year built, year filled

1924-1926

Dam type & material

Concrete gravity

Dam dimensions

Height: 56 m, Length (crest): 210m, Width (base): 52 m

Reservoir capacity

47,000,000 m3

Spillway information

1.1 m below top of dam

Construction notes or concerns

Serious seepage developed near base and sides

Event triggering failure

Seepage, undermine foundation

Age at failure

2 years

Mode of failure

Foundation problem

Breach dimensions

Full depth of dam, most of the width

Time to failure

70 minutes

Outflow from Dam

48,138 cms

Downstream effects

Roads, bridges, power stations destroyed

Dam failure toll

420

Before Dam Breach

Photo of the St. Francis Dam before it was destroyed, Feb 1927

After Dam Breach

Remains of St. Francis Dam and Reservoir Floor, Los Angeles County, CA, USA, March 1928

The Saint Francis Dam was a water supply dam in San Francisquito Canyon in the U.S. state of California. This curved-gravity concrete dam was constructed between 1924 and 1926. Its maximum height was 59.5 m, its maximum length was 4.8 km, and it had total reservoir capacity of 47 million m3.

The dam had a history of leaking and seepage, particularly around and under the foundation on the right side. By March 1928, an increased volume of seepage and the water’s muddy appearance suggested erosion from deep within the abutment. At 11:57 PM 12 March 1928, the 2-year old dam failed due to foundation problems. The breach released water through the entire depth of the dam and reached its maximum opening in 70 minutes. A peak discharge of over 48,000 cms entered the San Francisquito River canyon into the Santa Clara River and reached the Pacific Ocean within 4 hours. The initial height of the wave was estimated to be 55 m and the height was recorded at 33.5 m 0.8 km downstream. There were an estimated 450 fatalities.

Although foundation problems were the mode of failure, failure to recognize warning signs contributed to the disaster.

This is an example of a rapid and complete breach associated with a foundation failure.

2.0 Dams and Dam Failure » 2.5 Notable Dam Failures Around the World » 2.5.4 Mosul Dam, Iraq

Mosul Dam

Location, River

Mosul, Ninawa Governorate, Iraq, Tigris River

Year built, year filled

1981-1986

Dam type & material

Concrete gravity

Dam dimensions

Height: 113 m, Length (crest): 3.4 km

Reservoir capacity

11,100,000,000 m3

Spillway information

Controlled chute

Construction notes or concerns

Leakage beneath the dam is causing erosion of the foundation

Event triggering failure

No failure, foundation problems are a concern

Age at failure

No failure

Mode of failure

Hasn’t occurred yet. Foundation failure is very possible

Breach dimensions

No breach

Time to failure

No breach

Outflow from Dam

No breach

Downstream effects

Severe impact on major population centers and agricultural lands

Google Earth view of the Mosul Dam, Iraq

As the world’s population grows, demand for water and power generation increase. Large dams are often part of the solution but can pose great risk if not properly built and maintained. Iraq’s Mosul Dam is a case in point.

According to U.S. Army Corps of Engineers, the dam is not only Iraq’s largest but one of the most dangerous in the world as of early 2017.

Conceptual graphic of a dam with a grout curtain

Located in the Tigris River Valley, the dam was built on unstable ground. Leakage has been causing erosion of the foundation, and the grout curtain needs continual reinforcement to prevent the erosion from undermining the foundation. Without this, it is feared the dam may fail. Recent political unrest has disrupted these vital maintenance activities.

The dam is 113 m high and forms a lake nearly 50 km long. It is estimated that a dam failure flood would impact Mosul, home to more than 700,000 people, as well as Baghdad, some 550 km south.

Al-Taiee and Rasheed (2009) indicate that over 11,000,000 m3 of water are held behind the dam. They used a simplified dam break model to estimate flooding extent in Mosul. Their maximum estimates were a flood wave reaching Mosul in approximately 9 hours, reaching over 25 m in depth.

3.0 Basics of Dam Break Modeling

Guarapiranga Dam and Reservoir in Sao Paulo, Brazil

Modeling a dam failure provides guidance about the depth and velocity of floodwater and the time of its arrival at points downstream. This information can be used for emergency response plans and for engineering flood-resilient infrastructure in downstream communities.

In this section, we will consider three primary issues in modeling a dam failure and its impacts:

  1. Determining the breach characteristics
  2. Simulating the flows leaving the reservoir associated with the breach
  3. Simulating the downstream extent of flooding

3.0 Basics of Dam Break Modeling » 3.1 Dam-Related Terms

Terminology used with dam studies

We’ll refer to the following terms when discussing dams and the associated reservoirs and flows. The panels in the graphic illustrate the terms from three points of view: an aerial view, looking upstream at the dam, and a profile view.

Inflow is the flow entering the reservoir behind the dam.It can be from several sources.

Headwater is the level of the water surface immediately behind the dam. It is the height of the water going through the spillway or the water cresting the dam when overtopping.

Outflow is the flow leaving the dam. This can be through the gates, the spillway, or both. In a breach situation, it also includes the water overtopping the dam, piping through the dam, or flowing through the breach.

Tailwater is the level of the water at the base of the dam on the downstream side.

Dam elevation is the height relative to a datum and is referenced to something like mean sea level.

Dam height refers to the vertical height of the dam, usually from its lowest to highest point.

Dam width refers to the thickness of the dam. The width at the crest is typically less than the width at the base of the dam.

Dam length refers to the length of the crest of the dam as it spans across the river from one side to the other.

Spillway elevation is the elevation, relative to a datum, at its highest point.

Spillway height refers to the vertical height of the spillway above the base of the dam.

Question

Dam height and dam elevation are the same thing.

The correct answer is b.

Although the terms are related, dam height refer to the height of the dam from its base to its crest, and dam elevation refers to the elevation of the dam relative to some reference, such as sea level.

Please make a selection.

Question

Headwater refers to the _____ at the _____ side of the dam.

The correct answer is a.

Headwater refers to the water level on the upstream side of the dam. Tailwater is the water level on the downstream side.

Please make a selection.

3.0 Basics of Dam Break Modeling » 3.2 Breach Terminology

Terminology used with respect to dam breaches during dam failures

Determining breach characteristics is essential in dam failure modeling. A number of components of the breach must be estimated. The terms used are:

Breach width typically refers to the width of the bottom of the breach unless the top width is specified.

Side slopes is the slope of the sides of the breach.

Bottom elevation is the elevation of the bottom of the breach.

Time at beginning of the breach.

Time to fully develop or time to breach, is the time from the beginning of the breach to when it reaches its fullest extent.

3.0 Basics of Dam Break Modeling » 3.3 Simulating the Breach

Photograph of the breached Möhne Dam taken by Flying Officer Jerry Fray of No. 542 Squadron from his Spitfire PR IX, six Barrage balloons are above the dam

The breach is one of the most difficult processes to predict during the dam failure and subsequent flooding. The development and size of the breach and the nature of the reservoir can all contribute to the flow that will pass through the breach. Wahl (1998) published an excellent summary of breach parameters associated with embankment dams and references a number of other studies. Some of the findings include:

  • Breach simulation has the greatest uncertainty of all of the aspects of dam-breach flood wave modeling. Reservoir size has an impact on this uncertainty.
  • Large reservoirs with large volumes of water are often associated with relatively small drops in water level during breach formation. Peak outflow often coincides with the maximum breach opening. The term ‘large’ is somewhat relative. When a dam has a large volume of storage, only a portion of the reservoir’s volume can empty during the initial stages of the breach.
  • Small reservoirs tend to be associated with significant drops in water level during breach formation, and peak outflow often occurs before the breach is fully formed.

In our simulations, we’ll address two very important breach parameters: the time it takes to fully form, and the size of the breach. The other parameters are beyond the scope of this lesson.

3.0 Basics of Dam Break Modeling » 3.4 Simulating Flows

Hydrograph attenuation downstream of a dam failure

For any dam failure, the impacts of downstream flooding are affected by the total volume and rate of water leaving the reservoir. Simulating these impacts, and how the flood wave will attenuate as distance from the dam increases, requires the use of streamflow routing. There are two primary methods for routing the flow: hydraulic routing and hydrologic routing.

Hydraulic routing is more complicated and involves solving unsteady flow equations, often referred to as the St. Venant Equations. For more information on hydraulic routing, see the two-part COMET lesson:

It is generally accepted that hydraulic routing provides a more accurate solution due to the nature of a rapidly changing dam break hydrograph.

Compare long, narrow with short round reservoirs during dam break

Characteristics of the reservoir need to be considered in selecting a method. For example, a long narrow reservoir with a dam breach may develop a sloped water surface throughout the length of the reservoir, making hydraulic routing the better choice for simulating the movement of water. But for smaller, more circular reservoirs where the water surface may decrease more uniformly, the simpler hydrologic routing may perform adequately.

Note that hydraulic routing may not provide a better simulation than hydrologic routing if the input data are not available.

Question

Impact on hydrograph as floodwave from a dambreak travels downstream

The flood hydrograph downstream of a dam failure that represents a steep basin and little floodplain storage will attenuate _____ the flood hydrograph associated with a mildly-sloped basin with large areas of floodplain storage.

The correct answer is c.

A flood hydrograph downstream of a dam failure that represents a steep narrow basin with little floodplain storage will attenuate more slowly than a flood hydrograph associated with a mildly-sloped basin with floodplain storage.

Please make a selection.

HEC (2014) notes several considerations when routing flow through reservoirs for dam failure simulations:

  • The distance downstream: If the population that may be affected is far enough downstream, the routing method through the breach is less important because of attenuation. Attenuation will act to make the peak discharge from both routing methods converge when far enough downstream. What is “far enough?” If the downstream reach is steep and narrow with little or no off-channel storage, the peak discharge will not attenuate quickly, and hydraulic routing is preferred for many downstream points. But if the downstream reach is flat and has a large floodplain and off-channel storage, the peak discharge will attenuate more rapidly and hydrologic routing may be sufficient for many downstream points.
  • Reservoir details: The additional benefits of hydraulic routing are only realized if the reservoir is represented with detailed bathymetric data, numerous cross sections, and information about inflows and tributaries.

For many dam locations, the amount of data may well determine the type of routing used. Fread (2006) and HEC (2014) provide estimates of potential errors associated with various reservoir configurations and inflowing hydrographs. The error estimates in using level pool (hydrologic routing) versus hydraulic routing range from as high as 45% down to just a few percent, with hydraulic methods typically producing less error.

3.0 Basics of Dam Break Modeling » 3.5 Determining Downstream Extent of Flooding

Flooding in Rexburg, ID downstream of Teton Dam failure

Determining the extent of downstream flooding is perhaps the most important aspect of the overall simulation in terms of societal impact. But there are difficulties. If a location is close enough to the failed dam so that flood wave attenuation is relatively small, hydraulic routing should be used to capture the movement of the flood wave downstream. This requires detailed data about the physical characteristics of the waterway. Many issues can impact the routing of the flood wave through the downstream reaches, such as:

  • Accuracy and spacing of cross-section data
  • Computational time step used in the simulation
  • River valley characteristics
  • Off-channel storage areas or large floodplain areas
  • Levees
  • Hydraulic structures (bridges, culverts, and other dams, etc.)

As mentioned earlier, the attenuation of the flood wave will eventually reach a point where the method of routing has minimal effect. This distance is difficult to determine since it depends on the characteristics of the river valley.

3.0 Basics of Dam Break Modeling » 3.6 Overview of HEC-RAS Software

Homepage of the Hydrologic Engineering Center,  River Analysis System (HEC-RAS) website

In our simulations, we will use the U.S. Army Corps of Engineers (USACE) River Analysis System (RAS) modeling software or HEC-RAS. It was developed and is maintained by the Hydrologic Engineering Center (HEC) in California, USA. The software is used internationally and is free for download.

HEC-RAS is used to perform hydraulic computations on waterways. Among its features and capabilities are analyses of steady and unsteady flow, floodplain, dam break modeling, levee modeling, and flood mapping. This lesson is not intended to provide in-depth HEC-RAS training but uses the simulations to illustrate several elements and scenarios in dam break modeling.

The HEC-RAS website has more information about running a dam failure simulation, including input data, simulation data, and data displays.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations

We will run several scenarios of a widely used case study dam to illustrate important dam break parameters. Specifically, we will explore the sensitivity of dam failure to the size of the breach and the time it takes for the breach to fully develop.

Terminology used with respect to dam breaches during dam failures

To understand the sensitivities of important parameters, we need to vary those parameters for the same dam. The dam chosen for our scenarios has been the focus of numerous sensitivity studies.

An aerial view of the extensive flooding in Richmond, Virginia from Hurricane Agnes

We will simulate a major dam failure on our case study dam that occurred in 1972 after a large rainfall associated with Hurricane Agnes in the east-central United States. We will use four simulations.

First, we will run the simulation with no breach so we can see how it differs from the breach scenarios. Then we will run the simulation with a large breach that forms rapidly. Next, we will run another simulation where the final size of the breach opening is smaller. Finally, we will run a simulation of the large breach but with a slower time to full formation.

So our four scenarios are:

  1. No breach of the dam (No-breach scenario)
  2. Large, rapidly forming breach (Large-fast scenario)
  3. Smaller, rapidly forming breach (Small-fast scenario)
  4. Large, slowly forming breach (Large-slow scenario)

For each scenario we will examine the outflow hydrograph and downstream inundation. We will not focus specifically on the mode of failure, such as overtopping, piping, or foundation failure, although HEC-REC allows you to choose a mode of failure. You’ll want to keep the mode in mind since it’s often related to the type of dam (concrete versus fill) and its age.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.1 Overview of Case Study Dam

Foster Joseph Sayers Dam

Location, River

Pennsylvania, U.S.A., Bald Eagle Creek

Year built

1969

Dam type & material

Fill dam, earthfill

Dam dimensions

Height: 30.5 m, Length (crest): 2,083 m

Reservoir capacity

122,000,000 m3

Notes

The storage is equal to approximately 14 cm of runoff from the upstream drainage area of 878 km2.

The primary discharge means is through a series of gates located near the center of the dam.

There is an uncontrolled emergency spillway on the left side of the dam face.

The closest downstream population centers are the town of Mill Hall about 14 km downstream, and the city of Lock Haven 22 km downstream.

Aerial view of Foster Joseph Sayers Dam

Location of Foster Joesph Sayers Dam on GoogleEarth

Map/Location of Foster Joseph Sayers Dam

Drainage area and river basin

Map showing the Chesapeake Bay drainage area, Susquehanna River basin, and the 878 km2 drainage above the Foster Joseph Sayers Dam

Photo of Foster Joseph Sayers Dam

Photo of the Foster Joseph Sayers Dam and Lake at Bald Eagle State Park in Liberty Township, PA

The Foster Joseph Sayers Dam is located on the Bald Eagle Creek in the state of Pennsylvania, eastern U.S.A.. Completed in 1969, it is an earthfill dam operated by the USACE, which rises 30.5 m above the streambed and is 2,083 m in length. The reservoir has a surface area of approximately 700 hectares and a storage capacity of 12,000,000 m3. The storage is equal to approximately 14 cm of runoff from the upstream drainage area of 878 km2.

The primary discharge means is through a series of gates located near the center of the dam. There is an uncontrolled emergency spillway on the left side of the dam face.

Map showing the Foster Joseph Sayers dam and 2 towns downstream, Mill Lock and Lock Haven

We will look at simulated flood inundations in two locations: the town of Mill Hall about 14 km downstream, and the city of Lock Haven 22 km downstream.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.2 No-breach Scenario

To see how a very large runoff event with no dam breach compares to scenarios with dam breaches, our first scenario will simulate downstream effects when no breach occurs.

Profile of Foster Joseph Sayer Dam in Pennsylvania with heights, spillway, and gates

In this HEC-RAS cross section of the dam and spillway, the gray area (on the right) represents the dam. It rises to 208.5 m above sea level and has a maximum height from the base to the crest of 30.5 m. The spillway (on the left) begins uncontrolled discharge when the reservoir level reaches 200.25 m above sea level.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.2 No-breach Scenario » 4.2.1 Simulations in and near the Dam

Inflow hydrograph to the Bald Eagle Cr. reservoir behind the Foster Joseph Sayers Dam, June 1972.

The inflow hydrograph to the reservoir shows that incoming flow peaked at 438.8 m3/s on 23 June 1972 at 04:50 (green dashed line). The water surface elevation, or stage entering the reservoir (blue), was at 192 m above mean sea level when the storm inflow began on 22 June and continued to rise to a peak of 200.65 m at 22:15 on 25 June. The flow into the reservoir peaked on 23 June but the water surface in the reservoir continued to rise. The backwater effect from the full reservoir is seen in the inflow stage leveling off on 25 June.

Inflow and outlflow for Foster Joseph Sayers Dam, no-breach scenario

Now we’ll look at response in the reservoir itself, specifically the flow out of the gate (red line), the flow over the dam spillway (green-dashed line), the tailwater elevation (yellow line), and headwater elevation (blue line).

A gated orifice on the structure is programmed to remain open for the entire event. As the water in the reservoir rises, the flow out of the gate increases slightly but not much compared the flow that leaves via the spillway.

The flow out of the dam increases greatly at approximately midnight on 25 June when the levels in the reservoir become high enough to initiate flow over the emergency spillway. The spillway discharge increases rapidly, peaking at 22:25 on 25 June. The headwater (HW) stage increases late on 22 June as the inflow begins and the reservoir level starts to rise. The stage of the water immediately below the dam, called tailwater (TW), increases slightly at that time due to flow out of the gate only.

Hydrograph (Stage and Flow) ~150 m Downstream of Foster Joseph Sayers Dam, No-breach Scenario

This downstream hydrograph is approximately 150 m downstream of the dam and is influenced by both the flow through the gate and the flow over the spillway. Late on 22 June there is a rise of approximately 0.75 m in the river system below the dam due to backwater from a small tributary entering the river system below that dam. Its flow enters the system, peaks, and is gone before the mainstem flooding passes through the dam. In the hydrograph, you can see that the flow in this section just below the dam (green dashed line) peaks at approximately 94 cms, with a stage 179.71 m above mean sea level (blue line) at approximately 22:15-22:20 on 25 June.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.2 No-breach Scenario » 4.2.2 Downstream Effects

Flood Hyrdograph (Stage and Flow) at Mill Hall, 14 km downstream of Foster Joseph Sayers Dam, No-breach scenario

The flood hydrograph at Mill Hall, approximately 14 km downstream, shows that the stage rises approximately 2.7 m (blue line). Note that the peak in both stage (blue) and flow (green-dashed) occur early on 23 June, prior to the main outflow from the dam.

Question

Why would the peak flow downstream occur before flow over the dam spillway had begun?

The correct answer is c.

The peak flow on 23 June is from the local runoff from the storm reaching the creek below the dam. The outflow from the dam can be seen moving through as the slight increase on 25 June.

Please make a selection.
Flood hydrograph (Stage and Flow) at Lock Haven, 22 km downstream of the Foster Joseph Sayers Dam, No-breach scenario

At Lock Haven, 22 km downstream, the stage (blue line) rises approximately 3.3 m. Note that the peak flow (green-dashed line) and stage peak on 23 June during the local runoff. A slight leveling of the recession can be seen on 25-26 June as the main outflow for the dam moves through.

Next we’ll look at the aerial extent of flooding from the reservoir down to Mill Hall and Lock Haven on a digital elevation map.

Inundation map downstream from Foster Joseph Sayers Dam at Mill Hall and Lock Haven, no-breach scenario

As you can see, the large amount of local runoff resulted in some inundation in the downstream communities even without a dam failure (blue shading). The darker blue shows the greater depth of flood water.

On the following pages, we’ll investigate the same storm and setting but with three scenarios of dam failure.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.3 Large-fast Breach Scenario

Our second scenario simulates a large breach through the entire 30.5-m height of the dam. It’s a rapidly forming breach that fully develops in 3.2 hours.

Initial piping failure and full extent of the breach of the Foster Joseph Sayers Dam. This is the Large-Fast breach scenario.

Here’s how the breaching of the dam happens. Excessive storm runoff enters the reservoir behind the dam, and the water level in the reservoir rises. When the water surface in the reservoir reaches 200.5 m above mean sea level, a piping breach forms 11.5 m below the water level at an elevation of 189 m. It’s 9:00 on 25 June. The initial piping, or opening, is treated as a pressurized orifice while it grows but eventually the material above it collapses, forming a trapezoidal opening. The flow is now treated as weir flow (flow that is being channeled by the breach). The trapezoid grows until it reaches a bottom elevation of 178 m above mean sea level. The trapezoid is 136 m wide at the bottom and has steep side slopes. The breach reaches its maximum opening size 3.2 hours after the initial piping began.

In HEC-RAS, the breach parameters are entered in the geometry data editor for the dam. In particular, notice the following:

  • Starting elevation of breach: 189 m (above mean sea level)
  • Bottom extent of breach: 178 m (above mean sea level)
  • Bottom width of trapezoidal-shaped breach: 136 m
  • Time to failure (from start until full extent of breach): 3.2 hours
  • Reservoir stage when breach begins: 200.5 m

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.3 Large-fast Breach Scenario » 4.3.1 Large-fast Breach and Associated Hydrographs

Graphics showing major changes in stages and flows at the Foster Joseph Sayers Dam. This is the Large-Fast breach scenario.

The inflow hydrograph to the reservoir does not change from the no-breach scenario. But there are significant changes to the reservoir outflow hydrographs and reservoir levels as a result of the dam failure. As a consequence of the breach formation, headwater (HW, blue line) and tailwater (TW, red line) levels experience rapid changes, with the HW decreasing and TW increasing.

Recall that the breach was triggered when the water surface elevation in the reservoir reached 200.5 m. That’s only about 0.25 m over the emergency spillway, which is at 200.25 m. The piping failure began shortly after the spillway activated, but the spillway flow did not cause the breach. Once the breach began it proceeded rapidly, reaching its maximum size 3.2 hours later. This caused a rapid emptying of the reservoir as seen in the rapid rise and fall in the outflow hydrograph (green dashed line).

Comparison of simulated hydrographs for the large-fast breach versus the no-breach scenarios immediately below the Foster Joseph Sayers Dam.

A simulated outflow hydrograph immediately below the reservoir shows the stage (blue line) and flow (green dashed line). The peak outflow from the reservoir was 11,143.7 m3/s at 11:30 on June 25. Overlaid are the hydrographs for the no-breach scenario with stage (red) and flow (yellow). In the red line you can see the relatively small peak stage that occurred below the dam prior to the breach on 23 June.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.3 Large-fast Breach Scenario » 4.3.2 Downstream Effects

Comparison for large-fast breach scenario with no-breach scenario for Mill Hall 14 km downstream of the Foster Joseph Sayers Dam

Moving downstream to Mill Hall, we see a very large difference in river stage compared to the 2.7 m rise associated with the no-breach scenario. The breach case has more than 9 m rise in water in the Mill Hall area, including a rapid rise of ~8 m on 25 June. In addition, the no-breach flow is quite low compared to the breach flow, which peaks at nearly 9,000 m3/s.

Recall that the breach began to occur at approximately 09:00 on 25 June. The Mill Hall area saw the first rise in stage due to the failure at approximately 11:00 hours, only 2 hours later when the breach was not yet fully formed. The maximum depth of flooding occurs at 12:55 or about 4 hours after the breach initiated.

Comparison of the large-fast breach scenario  with the no-breach scenario for Lock Haven about 22 km downstream of the Foster Joseph Sayers Dam

We see similar results in the Lock Haven area, about 8 km further downstream from Mill Hall.

Question

The hydrograph at Lock Haven shows slightly attenuated flow. The peak is 5644 m3/s compared to ~9000 m3/s upstream at Mill Hall. What would cause this?

The correct answer is a.

The attenuation of the flow between Mill Hall and Lock Haven is mainly due to large off-channel storage areas. If these were ignored, there would be considerably less attenuation.

Please make a selection.
Inundation map of Mill Hall and Lock Haven downstream of the Foster Joseph Sayers Dam. It compares the simulated flood of a large-fast dam breach versus a no breach scenario.

Here’s the inundation map for the area around Mill Hall and Lock Haven. The area in green is from the no-breach scenario, while the blue shading indicates additional flooded areas from the breach. Darker blue is deeper water.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.4 Small-fast Breach Scenario

Now we’ll change the size of the final breach so it’s not as wide as the breach in the previous scenario and does not extend through the full depth of the dam. We’ll use the same fast development time of 3.2 hours.

Initial piping failure and final extent of the breach of the Foster Joseph Sayers Dam. This is the Small-Fast breach scenario.

As the cross section shows, the final bottom width is 90 m, 46 m less than in the large-fast breach scenario, and the final bottom elevation stops at 183 m, 5 m higher than in the large-fast scenario.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.4 Small-fast Breach Scenario » 4.4.1 Small-fast Breach Compared with Large-fast Breach

Now let’s compare the results of the small-fast breach scenario with the large-fast scenario. Let’s begin by looking at the flows and stages for both scenarios.

Headwater and Tailwater stage comparison between Large-Fast and Small-Fast breach at the Foster Joseph Sayers Dam.

The graphic shows the reservoir headwater (pink) and tailwater (yellow) stages as a result of the smaller breach. The red and blue lines show the results from the larger breach for the headwater and tailwater respectively.

Question

Why would the Headwater and Tailwater remain at higher levels for a longer time in the in the smaller breach than in the large breach?

The correct answers are b, c.

The smaller breach maintains a higher headwater because the reservoir is not emptying as rapidly or as completely. The tailwater for the smaller breach does not peak as quickly or as high. But it stays higher for longer than the tailwater for the large breach. That’s because the reservoir did not fully empty and stored water pushed through the gates for a longer period.

Please make a selection.
Comparison of reservoir outflows of the large-fast breach versus the small-fast breach at the Foster Joseph Sayers Dam.

Let’s look at outflows from the reservoir, comparing the lower outflow from the smaller breach (pink) to that from the larger breach (green). As you’d expect, the larger breach caused a peak outflow of over 11,000 m3/s, while the smaller breach had a reduced peak outflow of just slightly over 7300 m3/s.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.4 Small-fast Breach Scenario » 4.4.2 Downstream Effects

As we move 14 km downstream to Mill Hall, we continue to see the effects of the smaller peak flows exiting the reservoir.

Comparison of Flows 14 km downstream of the Foster Joseph Sayers Dam in Mill Hall. We are looking at the flow that results from a large-fast breach vs a small-fast breach.

The smaller breach is still producing smaller peak flows. However, the difference in peak flows between the large-fast breach (green) and small-fast breach (pink) is less than the difference immediately below the dam. At the outflow of the reservoir, the differences were approximately 4000 m3/s. But by the time the flow reaches Mill Hall, it is on the order of 2700 m3/s. These differences will continue to decrease as the flood routes downstream. Eventually, if we modeled the breach far enough downstream, there would effectively be no difference between the scenarios.

It is important to note that the smaller breach scenario is associated with a peak flow that arrives later but with high water that continues longer. This will be seen in the downstream hydrographs at Lock Haven as well.

Comparison of stages 14 km downstream of the Foster Joseph Sayers Dam in Mill Hall. We are looking at the stage that results from a large-fast breach vs a small-fast breach.

The stages at Mill Hall also show that the difference between the large-fast (blue) and small-fast (pink) breach scenarios decreases as you move downstream. There’s approximately a 1 m difference between the breaches.

We see similar results at Lock Haven. There’s approximately 1500 m3/s difference in the flow between the large-fast and small-fast breach scenarios, though the stage differences are still approximately 1 m.

Lockhaven Flow

Comparison of Flows 22 km downstream of the Foster Joseph Sayers Dam in Lock Haven. We are looking at the flow that results from a large-fast breach vs a small-fast breach.

Lockhaven Stage

Comparison of Stages 22 km downstream of the Foster Joseph Sayers Dam in Lock Haven. We are looking at the stage that results from a large-fast breach vs a small-fast breach.

Next, we’ll look at the aerial extent of flooding. The extent of flooding with the small-fast breach scenario is shown in pink. The pink plus the blue is for the large-fast breach scenario.

Inundation map of Mill Hall and Lock Haven downstream of the Foster Joseph Sayers Dam. It compares the simulated flood of a large-fast dam breach versus a small-fast dam breach.

The Mill Hall area sees basically the same extend of flooding from an aerial extent, but the depths are about 1 m less with the small-fast breach scenario when compared with the large-fast breach scenario. The Lock Haven area sees a reduction in both the depth and areal extent of flooding. Local topography around Lock Haven likely allowed a larger extent of flooding with the additional 1 m of water when compared with Mill Hall.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.5 Large-slow Breach Scenario

Initial piping failure and final extent of the breach of the Foster Joseph Sayers Dam. This is the Large-slow breach scenario.

The large-slow breach scenario uses the large breach size but breach formation develops more slowly. The time to full breach is doubled from 3.2 hours to 6.4 hours.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.5 Large-slow Breach Scenario » 4.5.1 Large-slow Breach Compared With Large-fast Breach Scenario

Let’s start by looking at the reservoir headwater and flows out of the reservoir for the large-slow breach compared to the large-fast breach.

Headwater and Tailwater stage comparison between Large-Fast and Large-slow breach at the Foster Joseph Sayers Dam.

As may be expected, the slower forming breach allows the headwater in the reservoir to drop more slowly and the tailwater to peak later compared to the fast forming breach of the same size. In the graph, the headwater for the fast-forming breach (red) shows a much faster drop in stage than the slow forming breach (purple). The tailwater for the slow forming breach (yellow) rises more slowly than the tailwater for the fast forming breach (blue), and the high water lasts longer.

Comparison of reservoir outflows of the large-fast breach versus the large-slow breach at the Foster Joseph Sayers Dam.

The peak flows leaving the reservoir in the large-slow breach (purple) are reduced by more than 3000 m3/s compared to the large-fast breach (green). By increasing the time to full breach formation, the peak flow is reduced from approximately 11,300 m3/s to 8233 m3/s. Recall that the previous simulation with a small-fast breach also had a reduction in flow, in that case to approximately 7300 m3/s.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.5 Large-slow Breach Scenario » 4.5.2 Downstream Effects

Comparison of flows 14 km downstream of the Foster Joseph Sayers Dam in Mill Hall. We are looking at the flow that results from a large-fast breach vs a large-slow breach.

Moving 14 km downstream to Mill Hall, the flows show similar results to those with the small-fast breach scenario. The peak flows for the large-slow breach (purple) are smaller than the large-fast breach (green), and the peak comes later.

Comparison of stages 14 km downstream of the Foster Joseph Sayers Dam in Mill Hall. We are looking at the stage that results from a large-fast breach vs a large-slow breach.

The corresponding stage at Mill Hall is not much different for the slow-developing breach (purple) than it is for the fast-developing breach (blue). It’s less than a meter lower for the slow breach.

Comparison of flows 22 km downstream of the Foster Joseph Sayers Dam in Lock Haven. We are looking at the flow that results from a large-fast breach vs a large-slow breach.

22 km downstream from the dam at Lock Haven the flows associated with the two large breaches—the fast-forming (green) and slow-forming (purple)—are beginning to coincide. The slow breach scenario peaks slower and about 500 m3/s lower.

Finally, we’ll look at the aerial flooding using both a terrain and map background. There’s little difference in areal extent between the faster forming breach and the slower forming breach.

Inundation map of Mill Hall and Lock Haven downstream of the Foster Joseph Sayers Dam. It compares the simulated flood of a large-fast dam breach versus a large-slow dam breach.

The purple shows the extent of floods in the large-slow breach with the small areas of blue indicating the additional flooding for the large-fast breach. The inundation is not as deep in the slow breach, but it lasts a little longer.

4.0 Sensitivity of Breach Parameters in Dam Failure Simulations » 4.6 Breach Scenario Summary

Photo of 1976 Teton Dam failure

In these scenarios we explored the impacts of both the size of a breach and the time it took for the breach to fully develop. To summarize, match the following statements:

Question 1 of 3

Large, rapidly-developing breaches:

The correct answer is a.

Large, rapidly-developing breaches cause the highest peak flows and stages in downstream areas.

Please make a selection.

Question 2 of 3

Relatively smaller breaches:

The correct answer is c.

Relatively smaller breaches will not be able to pass as much water from the reservoir and the peak discharge associated with the flood wave will attenuate more rapidly.

Please make a selection.

Question 3 of 3

Large, but slower-forming breaches:

The correct answer is b.

If the breach is large but develops slowly, the volume of water will be spread over a longer period of time relative to a fast-forming breach. Therefore the peak discharge for the slow-forming breach should be smaller but it may last longer.

In summary:

Large, rapidly-developing breaches cause the highest peak flows and stages in downstream areas.

Relatively smaller breaches will not be able to pass as much water from the reservoir and the peak discharge associated with the flood wave will attenuate more rapidly.

If the breach is large but develops slowly, the volume of water will be spread over a longer period of time relative to a fast-forming breach. Therefore the peak discharge for the slow-forming breach should be smaller but it may last longer.

Please make a selection.

5.0 Lesson Summary

Gordon Dam in Tasmania, Australia

The failure of large dams poses a great risk to life and property.

Fill (or embankment) dams have a higher rate of failure than concrete dams.

Failure modes of large dams tend to fall into three categories:

  1. Overtopping
  2. Foundation defects
  3. Piping and seepage

When dams do fail it tends to be early in the expected lifetime of the dam, especially for foundation failures.

In addition to excessive runoff, other factors important when considering dam failures include mechanical failure, design flaws, and lack of maintenance.

Aerial view of dam break in Quebec, Canada

The characteristics of a flood wave are sensitive to the breach dimensions and development time, and to both the reservoir and downstream basin characteristics. As a flood wave moves downstream it will attenuate. At some point downstream the flood peak and inundation area will not differ much between a scenario with a large, rapidly-forming breach compared to a scenario with a smaller, slowly-forming breach. The downstream attenuation will occur more rapidly in situations with shallow slopes and large floodplain storage.

In this lesson we explored the impacts of both the size of the breach and the time it took for the breach to fully develop. Relatively smaller breaches will not be able to pass as much water from the reservoir and the peak discharge associated with the flood wave will attenuate more rapidly. If the breach is large but develops slowly, the volume of water will be spread over a longer period of time relative to a fast-forming breach. Therefore the peak discharge for the slow-forming breach should be smaller than the peak with a fast-forming breach, but it may last longer.

You have reached the end of the lesson. Please complete the quiz and share your feedback with us via the user survey.

6.0 References and Websites

Al-Taiee, T. M. and A. M M. Rasheed. 2009: Simulation Tigris River flood wave in Mosul City due to hypothetical Mosul Dam break. 13th International Water Technology Conference, Hurghada, Egypt. http://www.iwtc.info/2009_pdf/4-1.pdf

Costa, John E. 1985: Floods from dam failures. United States Department of the Interior, Geological Survey, Open File Report, 85-560, Denver, CO

Fread, D. L., 2006: Course notes on Dam Break Floods.

Fujia, T. and Yumei, L. , 1994: Reconstruction of Banqiao and Shimantan Dams. International Journal of Hydropower and Dams, 49-53. (Report No. T032700-0207B, 79 pp.)

Hydrologic Engineering Center (HEC), 2014: HEC-RAS River Analysis System, User’s Manual. CPD-68. U.S. Army Corps of Engineers, Hydrologic Engineering Center, Davis, CA, USA.

ICOLD (International Commission on Large Dams), 1973: Lessons from dam incidents, Boston, Massachusetts, abridged edition, U.S. Commission on Large Dams.

Links:

Banqiao Dam, China: http://www.hurricanescience.org/history/storms/1970s/typhoonnina/.

Foster Joseph Sayers Dam: http://www.nab.usace.army.mil/Missions/DamsRecreation/FosterJSayersDam.aspx

Mosul Dam: U.S. Army Corps of Engineers study: Iraq's Mosul dam at 'higher risk' of failure: http://www.militarytimes.com/story/military/2016/02/09/us-army-study-iraqs-mosul-dam-higher-risk-failure/80074752/.

Contributors

COMET Sponsors

MetEd and the COMET® Program are a part of the University Corporation for Atmospheric Research's (UCAR's) Community Programs (UCP) and are sponsored by NOAA's National Weather Service (NWS), with additional funding by:

To learn more about us, please visit the COMET website.

Project Contributors

Program Manager
  • Bruce Muller — UCAR/COMET
Project Lead
  • Matt Kelsch — UCAR/COMET
Instructional Design
  • Bruce Muller — UCAR/COMET
Science Advisors
  • Dennis Johnson — Juniata College
Graphics/Animations
  • Steve Deyo — UCAR/COMET
Multimedia Authoring/Interface Design
  • Gary Pacheco — UCAR/COMET

COMET Staff, October 2017

Director's Office
  • Dr. Elizabeth Mulvihill Page, Director
  • Tim Alberta, Assistant Director Operations and IT
  • Paul Kucera, Assistant Director International Programs
Business Administration
  • Lorrie Alberta, Administrator
  • Tara Torres, Program Coordinator
IT Services
  • Bob Bubon, Systems Administrator
  • Joshua Hepp, Student Assistant
  • Joey Rener, Software Engineer
  • Malte Winkler, Software Engineer
Instructional Services
  • Dr. Alan Bol, Scientist/Instructional Designer
  • Tsvetomir Ross-Lazarov, Instructional Designer
International Programs
  • Rosario Alfaro Ocampo, Translator/Meteorologist
  • Bruce Muller, Project Manager
  • David Russi, Translations Coordinator
  • Martin Steinson, Project Manager
Production and Media Services
  • Steve Deyo, Graphic and 3D Designer
  • Dolores Kiessling, Software Engineer
  • Gary Pacheco, Web Designer and Developer
  • Sylvia Quesada, Production Assistant
Science Group
  • Dr. William Bua, Meteorologist
  • Patrick Dills, Meteorologist
  • Bryan Guarente, Instructional Designer/Meteorologist
  • Matthew Kelsch, Hydrometeorologist
  • Erin Regan, Student Assistant
  • Andrea Smith, Meteorologist
  • Amy Stevermer, Meteorologist
  • Vanessa Vincente, Meteorologist

Back to Top