EVALUATION OF NEW MODELS FOR SIMULATING EMBANKMENT DAM BREACH

1
EVALUATION OF NEW MODELS FOR SIMULATING
EMBANKMENT DAM BREACH

• Tony L. Wahl
• Bureau of Reclamation Denver, CO

2
What is CEATI International?

• Since 1891, the Canadian Electrical Association
  (CEA) has been the forum for electrical business
  in Canada
• In 1974, CEA initiated its RD Program to serve
  the research needs of Canadian electric utilities
• In 1998, CEAs RD Program opened its doors to
  international participation
• In 2001, CEA Technologies Inc. (CEATI) was
  separated from the Canadian Electrical
  Association
• CEATI International is now the Centre for Energy
  Advancement through Technological Innovation

3
(No Transcript)
4
Interest Groups

• 14 Interest Groups in the areas of electrical
  energy
• Generation
• Transmission
• Distribution
• Utilization
• Dam Safety Interest Group
• About 40 dam owners
• Jointly sponsors research development projects
• Participants from Canada, the United States,
  Europe, Australia, and New Zealand

5
Dam Safety Interest Group (DSIG)

• Areas of Interest
• Risk assessment for dam safety
• The use of geophysical methods in the diagnostics
  and monitoring of embankment dams
• Erosion and piping in dams
• Reliability of discharge facilities
• Ice loadings
• Probability (frequency) of extremefloods
• Emergency preparedness
• Testing of embedded dam anchors

6
Dam-Break Modeling Recent History

• Lethal Dam Failures in 1970s
• Canyon Lake
• Kelly Barnes
• Laurel Run
• Buffalo Creek
• Teton
• 1977 DAMBRK model developed
• Could route peak breach outflows to determine
  inundation depths, flood consequences
• Could determine peak breach outflow, given a
  description of how a breach would develop

7
Modeling Breach Development

• Concrete dam failure modes (sliding, overturning,
  structural) are usually instantaneous and
  complete
• Embankment dam failures usually involve erosion,
  which takes time and depends on many factors
• Regression equations to relate breach parameters
  to dam and reservoir characteristics
• Many developed in 1980s and refined in 1990s
• Adequate for cases in which the area of interest
  was in the far-field
• Too crude for the near-field

8
Physically-Based Breach Modeling

• Dr. Danny Fread recognizedneed for modeling
  erosionprocesses to obtain betterresults in
  near field
• May 18, 1980 eruption ofMt. St. Helens
  createdlandslide dam on Toutle River
• Dr. Fread developed NWS-BREACH model to analyze
  possible breach of this dam
• NWS-BREACH released to public in 1988

9
Modeling Developments in 1990s

• Flood routing capabilities much improved
• 2D modeling
• Integration with GIS to improve consequence
  analysis
• Little change in breach modeling during this time

10
CEATI Dam Erosion and Breach Project

• Since 2001 the DSIG has had an interest in
  improving the tools used to model embankment dam
  erosion and breaching
• Key Questions
• Will a dam breach?
• What is the outflow hydrograph?
• What is the warning time?
• Available methods mostly unchanged since late
  1980s
• Regression models for predicting peak outflow
• Regression models for predicting breach
  parameters
• Breach erosion models, such as NWS-BREACH

11
Shortcomings of Available Methods

• Regression models for peak outflow
• No aid in determining whether breach occurs
• Little detail about hydrograph shape or warning
  time
• Regression models for predicting breach
  parameters
• Uncertainties are large, especially for time
  parameters
• Breach initiation time
• Breach formation time
• Breach erosion models (e.g. NWS-BREACH)
• Used sediment transport equations, not true
  erosion models
• Poor modeling of erosion of cohesive materials

12
Large-Scale Physical Tests

• Since 2000, many organizations have been
  performing small-scale and some large-scale
  embankment breach tests
• European IMPACT Project (22 lab tests and
  sponsorship of Norwegian field tests)
• Norwegian tests (23 lab tests, 5 field tests of
  6-m-high dams)
• Agricultural Research Service (7 overtopping
  tests and 4 piping tests of 2-m-high dams)
• New breach erosion models under development
• Physically-based simulation of erosion processes
• Better modeling of the erosion of cohesive soils

13
Project Objectives

• Dam breach erosion project was initiated in 2004,
  with a focus on erosion and breach processes and
  prediction of breach outflow hydrographs at the
  dam
• We want to develop physically-based models for
  overtopping erosion and internal erosion leading
  to dam breach and facilitate the integration of
  those technologies into existing flood routing
  tools like HEC-RAS, MIKE11, Telemac, InfoWorks,
  etc.

14
Participants

• Electricité de France
• Case studieserodimeter and piping erosion
  research
• Hydro Québec / Ecolé Polytechnique Montréal
• Numerical modeling of dam breach, development of
  Firebird breach model
• Bureau of Reclamation
• Laboratory testinginvestigate erodimeters
• Agricultural Research Service
• Large-scale laboratory testing and development of
  SIMBA/WinDAM models
• HR Wallingford
• Large-scale testing (IMPACT project), developers
  of HR-BREACH model
• US Army Corps of Engineers
• Integration of breach modeling technology into
  HEC-RAS suite
• Elforsk AB
• Model evaluation
• Other interested parties and sponsors
• BC Hydro, Churchill Falls, Elforsk AB, EoN
  Vasserkraft, Great Lakes Power, Manitoba Hydro,
  New York Power Authority, Ontario Power
  Generation, Seattle City Light, Scottish
  Southern Energy, National Weather Service

15
Project Overview

• Phase 1 Information Gathering
• Reviewed and assembled case-study and large-scale
  laboratory test data
• Reviewed and identified numerical models under
  development
• Phase 2 Model Development and Implementation
• Phase 3 Model Enhancement

16
Tasks in Phase 2

• Evaluation of three numerical breach models
• SIMBA (ARS)
• HR-BREACH (HR Wallingford)
• FIREBIRD BREACH (Montreal Polytechnic)
• Evaluation of methods for quantifying erodibility
  of cohesive embankment materials
• leading to
• Integration of breach modeling technologies into
  HEC-RAS dynamic routing model
• Potential efforts to facilitate integration with
  commercial flood routing models

17
The Models Common Characteristics

• Models are all physically-based
• Models utilize quantitative input parameters
  describing erodibility of cohesive materials
• Models are intended to perform well without
  specific calibration to a particular case
• Models are not computationally intensive

18
The Models

• SIMBA Simplified Breach Analysis (USDA-ARS)
• Simulates breach by overtopping of homogeneous
  earth embankments with negligible protection on
  the downstream face
• Four stage failure process
• surface erosion leading to development of a
  headcut on the downstream face of the embankment
• headcut advance through the crest to initiate the
  breach
• breach formation as the headcut advances into the
  reservoir
• breach expansion during reservoir drawdown
• Erosion formulas are fixed and most calibration
  factors have been determined from lab testing.
  Complete model is not calibrated to any specific
  data set.

19
The Models

• HR BREACH (HR Wallingford)
• Overtopping or piping-induced breach of cohesive,
  non cohesive and simple composite (i.e. zoned)
  structures.
• Simulated processes
• Initial erosion of embankment surface protection
  (grass or rock cover)
• Headcut erosion through embankment
• Potential failure of breach side slopes by shear
  or bending
• Potential for sliding or overturning of core
  section
• Limited selection of erosion formulas
• Not calibrated to any specific data set

20
The Models

• FIREBIRD BREACH (Montreal Polytechnic)
• Overtopping-induced breach of homogeneous
  earthfill or rockfill dams
• One dimensional unsteady flow, St. Venant
  equations coupled with sediment continuity
• Able to handle transcritical flows
• Side slopes are evaluated for ability to resist
  sliding along a simple inclined face
• Choice of erosion formulas
• Can be more computationally intensive

21
Model Evaluation

• Evaluate model performance against large-scale
  laboratory tests and case-study data
• 2 ARS outdoor laboratory tests 2.3-m high
  homogeneous dams, overtopping 1 breach, 1
  non-breach
• 3 overtopping breach tests performed in Norway
  during the IMPACT project (5- to 6-m high dams)
• homogeneous clay
• homogeneous gravel
• zoned embankment
• 2 real dam failures
• Oros (Brazil)
• Banqiao (China)

22
ARS Tests

• Two overtopped embankments, 2.3 m high
• SM Silty Sand, complete breach in 51 minutes
• CL Lean Clay, headcut damage, but no breach after
  20 hours
• 2.5 orders of magnitudedifference in erodibility
  ofmaterials
• Constant inflow, smallreservoir
• Hanson, G.J., Cook, K.R., Hunt, S. 2005. Physical
  modeling of overtopping erosion and breach
  formation of cohesive embankments. Transactions
  of the ASAE, 48(5)1783-1794.

23
Norwegian Tests - Part of IMPACT

• Three overtopped embankments, 5 to 6 m high
• Homogeneous clay, placed very wet
• Homogeneous gravel, surface frozen
• Zoned rockfill with moraine core
• Inflow regulated at upstream reservoir
• Clay dam Peak inflow arrivedshortly after
  initial breachreservoir level went back
  uppeak outflow driven by peakinflow
• Flow regulation not attempted forgravel dam test
• Inflow was too little, too latefor zoned test

24
Oros Dam (Brazil, 1960)

• 35-m high dam, failed by overtopping during
  construction
• Core material probably a Sandy Lean Clay, with
  PI10
• Well-compacted, except maybe last lifts

25
(No Transcript)
26
Oros Dam - Summary

• Thick, erosion-resistant embankment, large
  reservoir
• Slow erosion
• 12 hrs to initiate breach
• 6.5 to 12 hrs to form breach and drain reservoir

27
Banqiao Dam (China, 1975)

• Hand-built dam with homogeneous earth shells and
  clay core wall of arenaceous shale
• Assumed to be poorly compacted and highly
  erodible
• 1 hr breach initiation
• 2 to 2.5 hrs to fully form breach

28
Evaluation Criteria

• Evaluate performance using
• initial inputs (best available information and
  judgment)
• optimized inputs
• Objective criteria
• Time to initiate breach (erode through crest)
• Time to form breach (reach full width)
• Final breach width
• Breach widening rate
• Peak outflow
• Subjective criteria
• Do models exhibit appropriate sensitivity?
• Ease of determining input data and selecting
  parameters
• Ease of operation

29
Current Status

• Team met at last years USSD meeting in Portland
• Members have been working this summer to perform
  the evaluation runs
• Group will meet again later this week to compare
  results and try to reach consensus on
• Which models and model components are working
  well?
• What technologies are presently ready to be
  integrated into state-of-the-art models?
• Where is more work needed?
• SIMBA and HR-BREACH models are being integrated
  into USDA WinDAM and Wallingford Software
  InfoWorks products

30
Challenges

• TIME Too many models, cases, scenarios
• Each case study presents unique evaluation
  challenges
• Real failures have questions about dam materials
  and erodibility, and about observed breach and
  outflow characteristics
• Lab tests have real-world logistical
  complications and limitations related to
  reservoir size
• Failure to accurately model breach initiation
  phase can require judgment to evaluate how well
  the model reproduced later stages of the breach
  process
• Evaluation process has already been extremely
  valuable

31
(No Transcript)
32
CEATI InformationChris HayesDirector, Business
Development1155 Metcalfe St., Suite
1120Montreal, QC H3B 2V6 [phone]
www.ceatech.ca info_at_ceatech.ca