Impact of Explosions on Embankment Dams and Levees

James Parkes, P.E. Senior Professional Associate—Geotechnical & Tunneling

Finalist, William Barclay Parsons Fellowship ProgramParsons Brinckerhoff

March 2014

ii

First Printing 2014

Copyright © 2014, Parsons Brinckerhoff

All rights reserved. No part of this work may be reproduced or used in any

form or by any means—graphic, electronic, mechanical (including photo-

copying), recording, taping, or information or retrieval systems—without

permission of the publisher.

Published by:

Parsons Brinckerhoff

One Penn Plaza

New York, New York 10119

Graphics Database: V212

iii

CONTENTSForeword ...............................................................................................................vii

Acknowledgements .............................................................................................viii

1.0 Introduction ....................................................................................................1

1.1 Background ................................................................................................21.2 Previous Analyses and Shortcomings .......................................................31.3 Examples of Explosive Impacts on Dams .................................................31.4 Research Objectives ..................................................................................51.5 Approach ....................................................................................................5

2.0 General Background on Dams and Levees .................................................7

2.1 Dam Type ...................................................................................................82.2 Dam Size ....................................................................................................82.3 Hazard Classification .................................................................................9

3.0 Development of the Trial Dam Cross Section ...........................................11

3.1 Type of Dam ............................................................................................ 123.2 Geometry of Trial Dam ............................................................................ 14

3.2.1 Dam Slopes ................................................................................. 143.2.2 Crest Width ................................................................................. 153.2.3 Dam Height ................................................................................. 163.2.4 Freeboard .................................................................................... 163.2.5 Internal Dam Geometry ............................................................... 163.2.6 Summary of Trial Dam Section ................................................... 17

4.0 Development of the Explosive Load and Location ...................................19

4.1 Size of Explosive Load ............................................................................ 204.2 Explosive Location .................................................................................. 20

5.0 Simplified Failure-Mode Analysis ...............................................................23

5.1 Possible Modes of Failure ...................................................................... 245.2 Considerations Regarding Other Failure Modes .................................... 25

6.0 Blast-Impact Assessment Approach .........................................................27

6.1 Conventional Approach for an Embankment Dam Stability Assessment ...................................................................... 286.2 Proposed Approach for Blast-Impact Assessments of Embankment Dams ..................................................... 29

iv

7.0 Software for Blast-Impact Analysis ...........................................................31

8.0 Soil Material Models for Blast Analysis .....................................................35

8.1 Suitability of Soil Models in Software Libraries ...................................... 368.2 Soil Behavior Under Explosive Loading and Soil Models ...................... 368.3 Sources of Soil Material Models for Analyses ........................................ 398.4 Validation Models .................................................................................... 408.5 Selection of Soil Model for Analysis ....................................................... 41

9.0 Development of Trial Dam Numerical Models ..........................................43

9.1 Model Set-up .......................................................................................... 449.2 Model Results ......................................................................................... 46

10.0 Assessment of Results ...............................................................................47

10.1 Assessment of Dynamic Global Stability ............................................... 4810.1.1 Dynamic Global Stability Analyzed as a Seismic Event ............. 4810.1.2 Dynamic Global Stability Analysis Based on Ground Motions .. 4910.1.3 Commentary on Dynamic Global Stability Assessments ........... 51

10.2 Localized Blast Impact ........................................................................... 5310.2.1 Development of Simplified Chart ................................................ 5510.2.2 Application and Limitations of the Simplified Chart ................... 57

10.3 Assessment of Post-Blast Global Stability ............................................. 58

11.0 Parametric Study .........................................................................................61

11.1 Results of Parametric Soil Material Model Assessment ........................ 6211.2 Results of Parametric Model with Concrete ........................................... 65

12.0 Mitigation Measures ....................................................................................69

12.1 Site Access ............................................................................................. 7012.2 Operational Protocols ............................................................................. 7012.3 Structural Improvements ........................................................................ 7212.4 Summary of Mitigation Measures ........................................................... 73

13.0 Areas for Further Research ........................................................................75

14.0 Conclusions .................................................................................................79

References ............................................................................................................83

v

Exhibits

2-1 Types of dams ..................................................................................................8

3-1 Summary of dams by primary type for the U.S. ............................................12

3-2 Summary of dams by height ..........................................................................14

3-3 Summary of typical small dam geometric configurations .............................15

3-4 Trial dam cross-section geometry .................................................................18

3-5 Trial dam global stability results for steady-state seepage ...........................18

4-1 Location of explosive load for trial dam analysis ..........................................21

5-1 Dynamic global stability failure (stability during the explosion) .....................24

5-2 Post-blast breach failure resulting from uncontrolled seepage through cratered/cracked zone ...................................................................................25

5-3 Post-blast global stability failure resulting from crater/cracked section .......25

7-1 Euler and Lagrange solvers ...........................................................................32

8-1(a) Soil skeleton consisting of solid soil particles and void space .....................37

8-1(b) Phase diagram of dry or partially saturated soil ............................................37

8-1(c) Phase diagram of fully saturated soil ............................................................37

8-2 Field apparatus for validation testing and validation model set-up ..............40

8-3 Comparison of published AUTODYN model and re-created AUTODYN model for saturated sand ...............................................................................41

8-4 Results of steel bucket validation models using sand-linear and clay-linear soil models .....................................................................................................41

9-1(a) 50-foot (15-meter) dam model, explosive over upstream side of crest ........45

9-1(b) 20-foot (6-meter) dam model, explosive over downstream side of crest, 3-D view .........................................................................................................45

9-2(a) Model results during blast showing propagation of pressure wave through the dam .............................................................................................46

9-2(b) Model results after blast showing effective strain contours at 3-D dam cross section ..................................................................................................46

10-1 Peak acceleration (g’s) of trial dam resulting from 5,000-pound (2,270-kilogram) TNT blast on upstream crest ..............................................49

10-2 Results of dynamic global stability assessment using gauge point data of 6 feet (1.8 meters) below crest .........................................................................50

vi

10-3 Model results for a blast over the upstream side of the crest showing effective strains ..............................................................................................53

10-4 Strain contours for explosions over upstream and downstream sides of the crest ................................................................................................................55

10-5 Simplified chart for estimating depths of potential crater and crack formation caused by explosive blasts on the dam crest ...............................................56

10-6 Comparison of pre- and post-blast global stability of trial dam section ......59

11-1 Results of parametric study of various soil material models ........................63

11-2(a) Trial dam composed of SC-SAT ....................................................................64

11-2(b) Trial dam composed of SP-SAT ....................................................................64

11-2(c) Trial dam composed of sand compaction .....................................................64

11-3 Partial dam model (upper 20 feet [6 meters] of embankment) with 2 feet (0.6 meters) of concrete on crest ..........................................................................66

11-4 Results of model (strains) with 2 feet (0.6 meters) of concrete on crest .......67

12-1 Moveable vehicle barrier with a very shallow foundation .............................71

vii

FOREWORD“Malicious acts intended to cause the failure of a major dam or dams are a threat to the nation and its citizens…Following the 9/11 attacks, the physical assurance of dams took on new importance. Now the owners and operators of dams find they need to change the ways they identify threats to and vulnerabilities of dams, manage risk, and implement measures to protect dams from security related failures.”

Assessment of the Bureau of Reclamation’s Security ProgramNational Research Council of the National Academies, 2008

The 9/11 attacks and the failure of levees during Hurricane Katrina have created serious concerns among agencies and owners responsible for water-retention structures that there exists a possibility for a malicious event that would combine the effects of these two disasters—an attack on a high-hazard dam or levee that could result in serious loss of life, property damage and economic devastation. Many agencies and owners are taking measures to reduce this possibility through risk analyses and security measures. However, this is a rapidly developing field and although security measures are being evaluated, there is not a substantive body of work with regard to how explosive blasts actually affect dams and levees or how well these structures can withstand such attacks. A clear and cost-effective method for analyzing explosive impacts on dams and levees has not yet been developed.

Many risk and vulnerability assessments focus on qualitative vulnerability assess-ments and security upgrades, but do not include actual numerical or computa-tional analyses of the impacts of explosive attack scenarios. A proper assessment of the impact of explosions on dams and levees requires sophisticated numerical analyses that account for the size, shape and materials of the structure as well as the location and intensity of the explosive load.

The primary objective of this research is to provide guidelines for analyzing explo-sive blast impacts on dams and levees. It presents a detailed research study on the impact of explosions on embankment dams and levees, including consider-ation of the behavior of soils under blast loads, the possible failure mechanisms of a dam or levee under such loads, an approach to performing blast-impact as-sessments and an approach for assessing the results of blast-impact models with regard to potential dam failure scenarios. Development of numerical blast mod-els for these unique loadings conditions are discussed in this monograph and a simplified chart for assessing localized blast impacts and the potential for erosive breach has been developed as well.

The monograph also includes results of a parametric study that has implications for possible structural protection measures. A discussion on possible mitigation measures to protect against or mitigate impacts of such threats is included.

Finally, areas for further research and study are identified which can advance the state of the practice in dam safety.

viii

ACKNOWLEDGEMENTSThis research was sponsored by the annual William Barclay Parsons Fellowship Program. I am grateful to the Parsons Brinckerhoff Career Development Com-mittee (CDC) for its support of this research program and the production of this monograph. In particular, the support of Greg Benz has been vital to the comple-tion of this research.

The technical guidance and support of my mentors, George Munfakh, Hugh Kelly and Sunghoon Choi, has been pivotal in making this innovative and exciting study a possibility. Special thanks to George for his encouragement and support throughout the program and to Hugh for his practical knowledge of the dam safety industry. I would also like to acknowledge my supervisor, John Wisniewski, for his support in managing this work among the other workload tasks during this period.

In addition, I would like to express my gratitude to Rodney Sedillo, who performed the slope stability analyses for this research, and to Carrie Nicholson who pre-pared many of the exhibits presented in the text.

Most important, I would like to acknowledge my wife, Trish, and our three daugh-ters, Emily, Mia and Caitlyn, whose encouragement, support and understanding made it possible to put the effort into completing this research, including many nights and weekends. Without their unconditional love and support, this work nev-er would have been possible.

James Parkes, P.E.Senior Professional AssociateGeotechnical and Tunneling Technical Excellence CenterParsons Brinckerhoff100 S. Charles StreetTower 1, 10th FloorBaltimore, MD 21201March 2014

1

1.0 INTRODUCTION

2

1.0 INTRODUCTION

1.1 BackgroundInfrastructure security is a growing area of engineering analysis with focus on high-risk structures such as buildings, bridges and tunnels. Dams and levees also need to be considered as part of these critical infrastructure assessments. Ac-cording to the National Inventory of Dams (NID) maintained by the United States Army Corps of Engineers (USACE), there are over 87,000 dams in the United States which serve many vital needs, such as municipal and industrial water sup-ply, flood control, hydroelectric power, irrigation and recreation (USACE 2013). In addition, there are over 100,000 miles (160,900 kilometers) of levees that protect communities from disastrous flooding (American Society of Civil Engineers [ASCE] 2013).

Many dams and levees contain public roadways or are accessible from nearby public roadways. Additionally, most dams and levees are built primarily out of earthfill and, as such, are not hardened structures or designed to resist high- intensity blast-impact loads. Therefore, these types of structures may be accessi-ble and vulnerable to the impacts of an explosive blast with potentially catastroph-ic consequences.

Agencies and dam owners are taking measures to reduce this possibility through risk analyses and security measures. Many risk and vulnerability assessments have focused on qualitative vulnerability assessments and security upgrades. These assessments often do not include computational analyses of the impacts of explosions, but rather focus on the accessibility of a structure, possible threat scenarios, increased security measures, etc. Computational analyses to assess the impacts of explosions involve complex, costly and time-consuming numeri-cal modeling and simulations. As a result, such computational analyses may be beyond a dam owner’s limited resources, or may only be done for the highest- priority structure within a particular owner’s inventory of dams. The overall num-ber, size, age and location of these types of structures can make such assess-ments a daunting task. Significant benefit to the profession could be realized if a consistent method of analysis and simplified tools could be developed to facilitate how such vulnerabilities are assessed.

Proper assessments of the risk that an explosive blast impact could pose to infra-structure typically involve a multistage process that consists of:

1. Threat/vulnerability/risk assessment.

2. Blast- and post-blast analysis.

3. Development of damage mitigation measures and preventive countermeasures.

3

The goal of this research is to provide tools that can be part of the Stage 2 (blast- and post-blast analysis) of the overall risk assessment to such structures. An as-sumed threat scenario (Stage 1) has been developed as the basis of the assessment and a limited discussion on mitigation measures (Stage 3) has been included.

1.2 Previous Analyses and ShortcomingsThe impacts of explosions on structures has been analyzed extensively since the 1960s. Much of the early research was based on empirical assessments as a re-sult of field experience or trials. Advances in computer technology and numerical modeling have allowed the development of methods to analyze explosions and their impacts on adjacent structures.

Manufactured materials, such as steel and concrete, have been widely studied. Geomaterials, such as soil and rock, have not been studied as much with regard to security analyses. In most cases, existing blast-security research has been fo-cused on structural design of facilities or vehicles and the potential impacts on oc-cupants. Where soils and geomaterials have been studied, it is oftenwith respect to crater formation, depth and diameter, or the use of explosives for excavation or ground improvement. Security analyses of the impact of explosions on structures built of geomaterials, such as embankments, has been limited.

1.3 Examples of Explosive Impacts on Dams A literature review indicates a limited number of actual explosive impacts on dams. This makes analyzing explosive impacts difficult as there is not a lot of precedent for comparison and because the characteristics of each individual dam (height, materials, construction, reservoir level, geometry, etc.) are different. Two previous attacks on embankment dams—the Sorpe Dam in Germany and the Pe-ruca Dam in Croatia—are discussed in this monograph.

Other types of dams, specifically concrete gravity dams, have also been attacked, most notably during World War II. However, the focus of this discussion and re-search will be embankment dams as will be described in detail in Section 3.

The attack on the Sorpe Dam was carried out in World War II as part of Operation Chastise, commonly referred to as the Dambusters Raid. The Sorpe Dam is an approximately 200-foot (60-meter) tall earth-embankment dam with a concrete core constructed between 1922-1933. The dam crest is approximately 33 feet (10 meters) wide with upstream and downstream slope configurations of 2.25H:1V and 2.5H:1V respectively. (Braimah and Rayhani 2012).

Sorpe Dam was attacked twice during WWII, in 1943 and 1944. Both attacks resulted in multiple direct hits to the dam crest with bombs ranging in size from 8,000 pounds (3,600 kilograms) to 12,000 pounds (5,500 kilograms). These hits resulted in craters approximately 40 feet (12 meters) deep and up to 80 to 100

4

feet (25 to 30 meters) in diameter, but did not result in dam failure (Braimah and Rayhani 2012). Although repairs were undertaken after the attacks, serious leak-age problems, including surges of muddy water, developed in 1951. Grouting was performed, but additional settlement of 4 to 5 feet (1.2 to 1.5 meters) occurred by 1956. Repairs to the outlet, concrete core and upstream face were performed by 1962 (Calcagno 2004).

The Peruca Dam in Croatia was attacked during the Balkan Wars on January 28, 1993. The Peruca Dam is an approximately 200-foot (60-meter) high rockfill embankment dam that was constructed between 1955 and 1960 (Braimah and Rayhani 2012). Serbian Armed Forces detonated 35 to 37 tons (35,500 to 37,600 kilograms) of explosives in seven locations within the dam (Darnton 1993). The explosives were planted in the spillway structure and inspection gallery. The explosions failed to breach the dam. It is noteworthy that the reservoir level was about 15 feet (4.6 meters) lower than anticipated, the rims of blast craters were 6 to 7 feet (1.8 to 2.1 meters) above the reservoir and the gallery entrances were left open, allowing venting of the explosive gases (Calcagno 2004).

The crest road at Peruca Dam deformed and settled as much as 10 feet (3 me-ters). The clay core of the dam was severely damaged and raised concerns about internal erosion. The reservoir level was lowered as quickly as possible (Darnton 1993). The dam was rehabilitated, including construction of a plastic concrete cut-off wall to replace the compromised clay core (Bauer 2013).

In the case of the Sorpe Dam, very powerful explosives dropped at high speed directly onto the crest of the dam did not lead to its failure. In the case of the Pe-ruca Dam, explosions within internal structures of the dam failed to compromise the global stability of the dam. However, although global stability failure did not occur in these cases, such a failure mechanism is a significant concern for any blast impact in the vicinity of a dam or slope. There are case histories of embank-ment-dam or slope failures resulting from nearby construction blasting activities (Braimah and Rayhani 2012). Controlled blasting and its potentially adverse effect on slope stability are widely recognized issues and the subject of much analysis in mining, quarrying and excavation work near slopes, embankments and dams. Therefore, the direct impact of an explosive blast on the global stability of a dam cannot be dismissed.

The examples above also illustrate that the direct impact of the explosion on overall stability is not the only concern. Cratering and the potential for an erosive breach, as well as damage to internal seepage control measures and the potential for internal erosion or piping, are also concerns for explosive impact effects on dams.

5

1.4 Research ObjectivesThe objectives of this research are to develop a proper numerical analysis that accounts for the unique loading conditions, apply that analysis to develop models for a trial dam/levee section, evaluate the possible modes of failures and assess the impacts of the explosion on overall stability of the dam/levee. The research will provide a framework for performing such analyses and a simplified chart that can be used to assess a particular failure mode.

The guidelines and simplified chart developed in this monograph can be used by engineers as part of their risk-management programs and can improve vulnerabil-ity assessments. The chart provides a simplified tool that can be used to quickly evaluate a dam or levee to:

• Establish the need for more in-depth analysis.

• Estimate the extent of potential damages under the given threat scenarios.

• Evaluate the need for and develop conceptual mitigation alternatives.

When used as part of an overall risk assessment program, the results of this research can be used to develop a more meaningful risk-assessment score. This can aid in evaluating where resources are prioritized for in-depth analyses and mitigation measures. The final result will be a better overall dam safety assess-ment that has a quantitative basis for blast vulnerabilities.

The discussions in this monograph are focused primarily on geotechnical engi-neering aspects of dam engineering and assessments, with particular focus on the dam structure itself. Hydraulic engineering considerations and considerations rel-ative to appurtenant or ancillary structures, such as spillways, gates, outlet works, etc., are not part of this research.

1.5 ApproachAssessment of the impact of explosions on dams and levees requires a rational approach and the use of appropriate computational methods. The following ap-proach was used in this research program:

1. Develop a trial dam section for analysis.

2. Determine the blast load and location for analysis.

3. Determine possible failure modes.

4. Develop an analytical procedure.

5. Develop numerical models for analysis.

6. Analyze the trial section for impacts of explosions.

6

7. Assess the results of blast impacts on potential failure modes.

8. Develop a simplified chart for analyzing similar dams.

9. Perform a limited parametric study.

10. Develop conceptual mitigation measures.

7

2.0 GENERAL BACKGROUND ON DAMS AND LEVEES

8

2.0 GENERAL BACKGROUND ON DAMS AND LEVEESSimplified background information is provided in this section with regard to the development of the blast-assessment analysis. It is presumed that the reader has some basic knowledge and understanding of dam and levee engineering. The terminology used in this section will be used throughout the remainder of this monograph. Because both dams and levees are water-retention structures and may have many similar characteristics, uses and loading conditions, this research may be applicable to both types of structures. For simplicity, the term “dam” will be used to refer to both dams and levees for the rest of the discussion. Addition-ally, dams tend to be designed and regulated to a higher standard than levees and so considerations for dam design and assessment have been used in this mono-graph to develop the models and assess the results.

2.1 Dam TypeThe type of dam refers to the type of material used in the construction of the structure. Most dams fit into one of three major categories: earth, rockfill, or con-crete. Concrete dams include gravity, arch and slab-and-buttress dams. Earth and rockfill dams consist of embankments constructed primarily out of readily avail-able natural materials. Earth dams are built of soils and can be either homoge-neous or zoned. Rockfill dams are generally constructed in areas where crushed rock is readily available, but soils are relatively scarce. Rockfill dams generally consist of a relatively impervious thin core section supported by highly pervious rockfill outer sections or shells. Both earth and rockfill dams may include internal filters and drains to improve the internal seepage design of the structures. Pre-dominant dam types are illustrated in Exhibit 2-1. Other types of dams also exist, such as masonry dams, but are much less common that those discussed above.

Exhibit 2-1: Types of dams.

2.2 Dam Size

The size of a dam generally refers to its height above the stream elevation (maxi-mum height of the dam). In general, dams are considered small if they are less than 50 feet (15 meters) in height (United States Bureau of Reclamation [USBR] 1973).

9

2.3 Hazard Classification The hazard classification for a dam is defined as either low, significant, or high. Hazard classification is a function of a dam’s size and potential downstream im-pact. Dams are classified as high hazard if failure of the dam is likely to result in loss of human life. Significant hazard applies where loss of life is possible and sig-nificant property damage or environmental destruction is likely (USACE 2013). Low hazard indicates that loss of life is unlikely and that property damage is not exten-sive and may be within the financial capacity of the owner to repair. Major impacts to critical infrastructure are often included in the high-hazard category rating, such as loss of a bridge or interstate highway section, loss of major power lines, etc. There are over 14,700 high-hazard dams and over 12,400 significant-hazard dams in the United States (USACE 2013).

It is important to note that the hazard classification of a dam is a function of the consequences of failure of the structure. These consequences may not necessarily be related to the size of the structure. In general, large dams are considered high- hazard dams since the volume of water impounded will undoubtedly have signifi-cant risk potential if released in a sudden and uncontrolled manner. However, a dam may be considered a small dam, less than 50 feet (15 meters) in height, but may still be classified as significant or high hazard depending on land use and population at risk downstream of the dam. The hazard classification of a dam may change over time owing to development of property or infrastructure downstream of the dam. Comparison of the 2010 and 2013 NID reports published by USACE indicates that the number of high-hazard dams is increasing, with over 700 additional dams classi-fied as high hazard in 2013 compared to 2010 (USACE 2010 and 2013).

Many small yet significant or high-hazard dams have been designed with simpli-fied methods, such as the USBR Design of Small Dams, or published guidelines by various agencies. These dams are unlikely to have been assessed for explosive load impacts. These are the types of dams to which research is anticipated to be largely applicable.

11

3.0 DEVELOPMENT OF THE

TRIAL DAM CROSS SECTION

12

3.0 DEVELOPMENT OF THE TRIAL DAM CROSS SECTION A trial dam section was developed based on widely used industry guidelines. The trial dam section was developed to be a “common” or reasonably applicable dam configuration and loading condition, with some simplifying assumptions to facilitate the analysis. The following sections outline the development of the geometry for the trial dam section for analysis.

3.1 Type of DamEarth dams are the most common type of dam. They are built of readily available lo-cal soil materials that can be incorporated into the final construction of the structure with a minimal amount of processing. Foundation requirements are often less strin-gent for earth dams than for other types of dams. As a result, a minimum earth-dam section is often the most economical dam section that can be built for a particular site. According to the NID (USACE 2013) and as indicated in Exhibit 3-1, the over-whelming majority of dams in the U.S. are earth dams. This research effort focuses on earth dams in order to have the widest possible applicability to the industry.

Exhibit 3-1: Summary of dams by primary type for the U.S. (USACE 2013).

13

Note that the term “earth” or “earthen” dams generally refers to embankment dams made of soil materials, such as sand, silt, clay and gravel as defined by the Amer-ican Society for Testing and Materials (ASTM International) and the Unified Soils Classification System (USCS). Rockfill dams are considered a separate type of dam because they are built primarily of larger gradation materials, including cobbles and boulder-size fragments. Lift thickness, material size, gradation, material placement and compaction of rockfill differ significantly from soil and, as such, special con-siderations apply to the design and construction of rockfill dams. For the purposes of this research, rockfill dams are considered to be a distinctly separate category of structure from earthfill dams. This is consistent with the industry approach as a whole, as earth dams and rockfill dams are distinguished as different categories in Exhibit 3-1 from the NID. Although some aspects of analyzing earthfill and rockfill dams are similar, the focus of this research is on earthfill dams and therefore may not be entirely applicable to rockfill dams.

Earth dams are constructed by placement of compacted layers or lifts of fill soils, referred to as rolled fill. Other types of construction, most notably hydraulic fill or semi-hydraulic fill, have been used in the past to construct large dams or levee systems. However, modern construction using rolled-fill practices is more econom-ical, offers better construction quality control and does not produce the seismic failure risk potential that hydraulic fill does (loose saturated sandy soils susceptible to liquefaction). Hydraulic and semi-hydraulic construction methods are generally no longer used in the U.S. This research will focus on compacted rolled-fill embank-ments; liquefaction potential is not considered.

Earth dams can be homogeneous fill, zoned, or diaphragm. Homogeneous dams are comprised of a single material, not including slope protection, internal filters, or berms. Homogeneous earth dams are the most common type of embankment dam. These types of dams are built from readily available nearby soils and require a mini-mum borrow volume of specified materials (filters or core). Construction operations are also facilitated by the homogeneous section.

The majority of earth dams tend to be small, defined as less than 50 feet (15 meters) in height, as indicated in Exhibit 3-2. Therefore, the trial dam will be a homogeneous earth dam (or levee) with a maximum height of 50 feet (15 meters) in order to have wide applicability.

14

Exhibit 3-2: Summary of dams by height (USACE 2013).

3.2 Geometry of Trial DamLarger dams are often designed with project-specific detailed analyses, whereas small dams are more likely to have been designed following general industry stan-dards and guidelines. For example, many small dams have been designed using Design of Small Dams by the USBR (1973) or by using published guidelines by the National Resource Conservation Service (NRCS) or USACE. The geometry of the trial dam is developed primarily from USBR and USACE guidelines.

3.2.1 Dam SlopesEmbankment dams and levees typically have slopes ranging from as steep as 2H:1V to as flat as 5H:1V depending on the type of material used in the embank-ment, the foundation materials, the construction methods and the available space. In general, upstream slopes are usually slightly flatter than downstream slopes. Special loading conditions, such as rapid drawdown, are not considered in this geometry of the trial dam.

For homogeneous dams, USBR (1973) indicates that downstream slopes of 2H:1V to 3H:1V and upstream slopes of 2.5H:1V to 4H:1V are applicable. This is consis-

15

tent with other industry guidelines, such as Design and Construction of Levees (USACE 2000), which indicate that levee slopes may vary from 2H:1V to 5H:1V, although the 5H:1V configuration is generally not typical. The slope configuration is a function of the material used for construction (fine-grained and higher- plasticity soils require flatter slopes for stability).

Common slope configurations from Design of Small Dams (USBR 1973) are sum-marized in Exhibit 3-3. For consideration in analysis, a reasonable, yet minimal, cross section is desired. A minimal cross section is considered to be more vulner-able to the impacts of an explosion based on the assumption that additional mass will provide additional inertia and resistance to the explosion impacts. A minimal cross section is based on the steepest possible slope configuration, which is applicable for silty or clayey sands and gravels. For the trial dam, silty or clayey sand—classified as SM or SC in accordance with the Unified Soils Classification System (USCS)—is assumed for the dam materials. For simplicity in the models as well as wide applicability, a homogeneous earth dam is assumed. Therefore, the upstream and downstream slope configurations for the trial dam are 2.5H:1V and 2H:1V, respectively.

Type Facility Purpose Soil TypeUpstream

SlopeDownstream

Slope

Homogeneous Dam Detention or Storage

GC, GM, SC, SM 2.5:1 2:1CL, ML 3:1 2.5:1

CH, MH 3.5:1 2.5:1Zoned Earthfill, Min. Core Width

Detention or StorageGC, GM, SC, SM, CL, ML, CH, MH

2:1 2:1

Zoned Earthfill, Max. Core Width

Detention or Storage

GC, GM 2:1 2:1SC, SM 2.25:1 2.25:1CL, ML 2.5:1 2.5:1CH, MH 3:1 3:1

Minimum 2:1 2:1Maximum 3.5:1 3:1For analysis 2.5:1 2:1

Exhibit 3-3: Summary of typical small dam geometric configurations (modified from USBR 1973).

Note: For dams not subject to rapid drawdown.

3.2.2 Crest WidthThe width of the crest can vary significantly. Generally, for construction, inspection and maintenance access, the minimum width for a levee is 10 to 12 feet [3.0 to 3.7 meters] (USACE 2000). Wider crest widths are used if the embankment incorpo-

16

rates a roadway on the crest. For dams, a crest width of 25 to 40 feet (7.6 to 12.2 meters) is typical (USACE 2004). USBR (1973) indicates a minimum crest width of 10 feet (3 meters) for construction and access and provides a formula for relating the crest width to the dam height, which results in a minimum width of 20 feet (6 meters) for a 50-foot (15-meter) high embankment. USBR also notes that 20 to 25 feet (6 to 7.6 meters) is a likely minimum width if a public roadway is to be includ-ed on the crest.

For this research, it is assumed that there is a public two-lane roadway on the crest. Minimum travel-lane widths of 10 to 12 feet (3.0 to 3.7 meters) are assumed with no shoulders and minimal area for fencing or guardrails. Therefore, a crest width of 25 feet (7.6 meters) is used for the trial dam section.

3.2.3 Dam HeightThe height of the embankment is considered to be 50 feet (15 meters) or less, based on the geometry assumptions discussed in the narrative on dam slopes (see Section 3.2.1) and the number of similar-height structures discussed in Sec-tion 3.1. For the model, the maximum height of 50 feet (15 meters) is used. No berms or slope benches are included in order to maintain a simple cross section.

3.2.4 FreeboardThe amount of freeboard (vertical distance between the water surface and the crest of the dam) can vary considerably depending on the dam purpose and the loading conditions. USBR (1973) indicates that a minimum freeboard of 3 to 7 feet (0.9 to 2.1 meters) above the maximum water level should be used. For the trial dam section, a minimum freeboard of 5 feet (1.5 meters) has been used.

The amount of freeboard at a particular structure will vary depending on precipita-tion levels, flood levels, flood frequency and dam operations. Reservoir levels may vary by a few feet to tens of feet over a period of months. A reasonable freeboard level is necessary for the trial dam analysis; a parametric study on the effects of the freeboard level would be useful, but is beyond the scope of this research.

3.2.5 Internal Dam GeometryThe trial dam section is assumed to be homogeneous with uniform internal ge-ometry. A toe drain is assumed solely for the purpose of maintaining the phreatic line within the embankment and for controlled discharge; this drain will not be modeled in the numerical model for the explosion. All soil within the embankment is assumed to be saturated or nearly saturated; soil below the phreatic surface would be saturated as a result of seepage, while soil above the phreatic surface may be saturated as a result of capillary rise, precipitation infiltration, or from be-ing placed wet of optimum moisture content during construction.

17

3.2.6 Summary of Trial Dam SectionIn summary, the trial dam section is developed based on the following considerations:

1. The dam is a homogeneous small dam (a toe or blanket drain may be present for seepage control, but is not considered in the blast-impact model).

2. Dam height is 50 feet (15 meters) or less.

3. The dam is built of a mixture of clayey or silty sand (SC or SM material). The upstream and downstream slopes are 2.5H:1V and 2H:1V, respectively.

4. The trial dam section is not subject to rapid drawdown.

5. Freeboard is 5 feet (1.5 meters).

6. The dam crest is 25 feet (7.6 meters) wide and contains a public, two-lane roadway.

The trial dam section is reasonable, yet minimal, in mass. The trial dam section is illustrated in Exhibit 3-4, while Exhibit 3-5 presents the results of a steady-state seepage and slope stability analysis. The minimum factor of safety (FS) against global stability of 1.5 is within acceptable industry guidelines for dam safety. Therefore, the trial dam section is reasonable as well as consistent with industry practice for global stability.

18

Exhibit 3-4: Trial dam cross-section geometry.

Exhibit 3-5: Trial dam global stability results for steady-state seepage.

19

4.0 DEVELOPMENT OF THE EXPLOSIVE LOAD AND LOCATION

20

4.0 DEVELOPMENT OF THE EXPLOSIVE LOAD AND LOCATION An appropriate explosive load and location have been developed for the trial dam mod-el. The explosive load developed for this monograph is based on an effective charge weight of TNT. Other explosive types can be related to TNT by using effective charge-weight ratios (for examples, refer to Choi 2009). The explosive is modeled solely as a mass of TNT. The delivery platform, casing, etc., is not included in the model.

4.1 Size of Explosive Load Infrastructure security assessments typically consider several sizes of explosive loads, ranging from a small hand-carried load such as a bag or briefcase; an inter-mediate explosive load such as a car trunk; or a large explosive such as a truck or van (Parsons Brinckerhoff 2010). For this research, the assumed threat scenario is the explosive delivered using a vehicle on the public crest roadway.

An explosive load consistent with a truck or van is used. Examples of similar events include the 1993 attack on the World Trade Center and the 1995 attack on the Okla-homa City Federal Building. The threat scenario for this monograph uses an equiv-alent explosive size of 5,000 pounds (2,270 kilograms) of TNT. This is the upper end of the possible range for an explosive delivered by a van or light truck (Parsons Brinckerhoff 2010). This size of the explosive is consistent with analyses that have been performed for other transportation-related infrastructure. The impact of a smaller load, equivalent to 2,500 pounds (1,135 kilograms) of TNT, is also assessed as part of the parametric study presented in Section 11 of this monograph.

4.2 Explosive Location The explosive location is evaluated for two locations: the approximate center of the upstream half of the crest and the approximate center of the downstream half of the crest. This is intended to model cases with the vehicle in the upstream lane or in the downstream lane. Since the crest on the trial dam is 25 feet (7.6 meters) wide, the explosive will be centered approximately 6 to 6.5 feet (1.8 to 2.0 meters) from either the top of the upstream slope or the top of the downstream slope.

The explosive is centered above the ground surface because the vehicle platform will have ground clearance and the explosive itself will have mass. The height of the explo-sive above the ground surface is referred to as the “standoff distance” because it is the minimum distance from the center of the explosive to the structure being assessed. A standoff distance of 4 feet (1.2 meters) is used. The location of the explosive load on the trial dam is shown in Exhibit 4-1. The same location and standoff distance is used for the smaller 2,500-pound (1,130-kilogram) load used in the parametric study.

21

Exhibit 4-1: Location of explosive load for trial dam analysis.

23

5.0 SIMPLIFIED FAILURE-MODE ANALYSIS

24

5.0 SIMPLIFIED FAILURE-MODE ANALYSIS

5.1 Possible Modes of FailureA simplified failure-mode analysis was performed to determine possible failure mechan- isms that could result from the impact of an explosive blast on the crest of an embank-ment dam. The blast wave from the explosion will produce a crater as well as ground vibrations/shock. The following failure modes were considered to be possibilities:

• Global stability failure resulting from the dynamic impact of the explosion (fail-ure during the explosion). The ground shock and vibration from the explosion may result in internal shearing and/or an increase in pore pressure, which may reduce or compromise the global stability of the dam.

• Localized failure resulting from an erosive breach following the blast. Water inflow into a crater or cracked section can lead to erosion, overtopping and/or piping, which could become progressively worse and develop into an uncontrolled breach.

• Global stability failure resulting from post-blast geometry. The post-blast ge-ometry of the dam will include a crater and/or a cracked section which could adversely affect the global stability.

These three failure modes are illustrated in Exhibits 5-1, 5-2 and 5-3. The failure modes in Exhibits 5-2 and 5-3 may take hours to weeks to develop. Because repairs to a dam, such as filling in craters, take time to accomplish, it is possible for the dam to be vulnerable to these failure modes before repairs have been completed despite the fact that there was no apparent failure in the immediate aftermath of the explosive event. External loading conditions may change during the event aftermath and repair process. Precipitation events and rising reservoir levels may increase the external reservoir load, the internal pore pressures and the susceptibility for erosion.

Exhibit 5-1: Dynamic global stability failure (stability during the explosion).

25

Exhibit 5-2: Post-blast breach failure resulting from uncontrolled seepage through cratered/cracked zone.

Exhibit 5-3: Post-blast global stability failure resulting from crater/cracked section.

5.2 Considerations Regarding Other Failure ModesThe failure modes described above are generalized failure modes applicable to an explosion on the crest of a homogeneous embankment dam. Additional failure modes, or modifications to these failure modes, may be applicable to other types of dams, other parts of a dam, or other explosive locations.

Failure modes related to the dam foundation and abutment—for example, a failure along pre-existing planes, such as rock joints—have not been considered in this monograph. Also beyond the scope of the research presented here are damages to spillways or other ancillary or appurtenant structures.

27

6.0 BLAST-IMPACT ASSESSMENT APPROACH

28

6.0 BLAST-IMPACT ASSESSMENT APPROACHThe ultimate goal of a blast-impact assessment is to determine whether an explo-sive blast will compromise the stability of the dam. Multiple possible failure modes exist, as indicated in Section 5. A rational multistage approach is needed for an effective blast-impact assessment. This approach is similar in principle to conven-tional geotechnical dam assessments.

6.1 Conventional Approach for an Embankment Dam Stability Assessment

Conventional dam engineering assessments for both new dams and existing dams require multiple analyses and software programs. Construction of a new dam embankment or modifications to an existing embankment may include placement of fill to build or raise the dam, the addition of stabilization berms, the installation of drainage measures, the installation of filter measures, construction of cut-off measures, etc. Assessments for conventional design will include:

• Deformation analyses: evaluate settlements, lateral spreading, development and dissipation of excess pore pressures, etc.

• Seepage analyses: evaluate steady-state seepage pressures, uplift pressures and the potential for heave and piping.

• Slope stability analyses: evaluate global stability and seismic stability.

• Multiple loading conditions: evaluate conservation pool and probable maxi-mum flood (PMF) conditions, seismic events, rapid drawdown, etc.

In addition to the assessments listed above, conventional assessments also include foundation assessments for settlement, bearing capacity, under seepage, etc. How-ever, this blast-impact assessment focuses on the dam itself and does not include an assessment of impacts to the foundation or abutments. For consistency, the conventional geotechnical analyses discussed also focus only on the dam.

Performing conventional geotechnical dam engineering assessments requires mul-tiple types of assessments, evaluation of multiple loading conditions and the use of several different software programs. In some cases, simplified methods—such as published charts or equations based on simplified geometry and assumptions—can be used as tools to aid in these assessments. Such simplified methods exist for evaluating the potential for piping or heave, relief well design, simplified slope stabil-ity charts, etc. The simplified methods can be used as tools in overall assessments, such as providing an initial simplified assessment, an empirical check of more com-plicated numerical models, or a final design for low-hazard structures.

29

6.2 Proposed Approach for Blast-Impact Assessments of Embankment Dams

Similar to conventional assessments, blast-impact assessments require multiple types of analyses using several software programs to evaluate multiple failure modes and loading conditions. The following multistage approach is proposed for blast-impact assessments of new or existing embankment dams:

1. Global stability during the blast:

a. Evaluate the dynamic loads on the dam (ground acceleration, peak-particle velocity, ground displacements, stress increases, etc.).

b. Assess global stability under the dynamic loads.

2. Localized impact and erosive breach potential:

a. Evaluate the formation of a cratered or damaged section under the blast impact.

b. Assess the potential for erosion/piping through the damaged section.

3. Global stability following the blast:

a. Assess static steady-state global stability of altered dam cross section re-sulting from the blast impact (cratered/damaged section).

For all of the above cases, multiple loading conditions are possible and include variations in the explosive size and location and the reservoir level.

Performing the above assessment requires specialized software for analyzing the impact of the explosive blast (Steps 1.a and 2.a), conventional geotechnical engineering software for analyzing the global stability impacts (Steps 1.b and 3) and engineering judgment for assessing the potential for formation of an erosive breach (Step 2.b).

Of critical importance is the modeling and analysis of the explosive blast impact; the other assessments build on the results of that analysis. The software for per-formance of the blast-impact analysis, the required material models and the trial dam numerical models are discussed in Sections 7, 8 and 9 of this monograph. Assessment of the model results with regard to the approach outlined above fol-lows in Section 10.

31

7.0 SOFTWARE FOR BLAST- IMPACT ANALYSIS

32

7.0 SOFTWARE FOR BLAST-IMPACT ANALYSISModeling the impacts of explosions requires specialized software not commonly used in geotechnical engineering. Conventional geotechnical analyses are based on relatively slow loading conditions that occur over time periods ranging from days to years. The loads are generally static and of low to moderate intensity, with applied pressures generally within the range of conventional soil-strength param-eters. Deformations are typically small compared to the size of the structure or thickness of the soil deposit.

Explosion-impact analyses, by contrast, involve high-intensity and very short duration loading conditions. Loading occurs over time periods of fractions of a second. Deformations and strains can be large and load intensity (pressure) may be several times the material strengths. Transient analyses and nonlinear material behavior are significant aspects of these types of analyses, especially compared to conventional geotechnical analyses. Therefore, it is necessary to use special-ized software specifically designed for such analyses.

Analysis of explosions and blast impacts is typically performed using explicit- dynamics software which enables modeling of very rapid, high-intensity loads on structures. Such analyses require complex numerical modeling software that can be used to analyze high-intensity rapid loading conditions using different solvers for fluids and solids in a coupled analysis. This is known as Euler-Lagrange cou-pling. The Euler processor is used to analyze large distortions of fluids and gases, including explosives and detonation products. In a Euler analysis, the fluids pass through the individual elements representing an area or space. The Lagrange pro-cessor is used for modeling solid structures. In a Lagrange analysis, the elements representing the solid body deform under the loading conditions. The Euler and Lagrange analyses are depicted in Exhibit 7-1. Coupled together, the impact of an explosive blast on a solid structure can be effectively modeled.

Exhibit 7-1: Euler and Lagrange solvers (Parsons Brinckerhoff 2010).

33

One commonly used software package for performing Euler-Lagrange coupled analysis is AUTODYN by ANSYS, Inc. AUTODYN Release 14 has been used for the research in this monograph; therefore, aspects of the research presented here may only be applicable to AUTODYN. Other similar programs exist and can be used to perform similar analyses. AUTODYN was selected for the analysis in this monograph because it is widely used in practice for performing blast-impact analyses and because available research that was reviewed for this work was also based on AUTODYN.

The results of the AUTODYN analysis have been assessed using conventional geotechnical software for slope stability as well as engineering judgment. The programs Slope/W and Quake/W by GeoStudio were used in these assessments. These programs are widely used in conventional dam engineering assessments and are considered appropriate for this work. The assessments are discussed in detail in Section 10.

35

8.0 SOIL MATERIAL MODELS FOR BLAST ANALYSIS

36

8.0 SOIL MATERIAL MODELS FOR BLAST ANALYSIS

8.1 Suitability of Soil Models in Software LibrariesMost software for explicit-dynamics analyses include material libraries of models for fluids and structural materials for use in analysis. Most materials have a com-plex response to dynamic, high-intensity loading and these material libraries have been developed over years of practice based on evolving research. However, the majority of the material models tend to be for manufactured materials, such as steel, concrete, glass, masonry, etc. In most blast-impact analyses, soils are not involved, or are involved as secondary aspects of the analyses, such as confining media around underground structures. This type of loading is not typical for geo-technical applications and there are relatively few structures built of soils that have required such assessments.

The default library in AUTODYN includes one material for soil, Sand-Compaction. This material is a compaction model that simulates deformation of dry sand in response to an explosive load.

In this research, embankment soils for a homogeneous dam have been modeled. Materials for dams are generally placed wet of optimum moisture content in order to improve ductility and minimize post-construction settlements upon impound-ing of the reservoir and seepage through the dam. Seepage through the dam and capillary rise above the phreatic surface will lead to saturation of the dam soils if they are not already saturated as a result of the construction process. Therefore, soils within a homogeneous embankment dam are likely to either be at or near full saturation.

The default material in the AUTODYN library (dry-sand compaction model) may not be the most appropriate for modeling embankment dam soils. It is necessary to consider the behavior of saturated soils under explosive loads and develop a material model that is appropriate to such behavior.

8.2 Soil Behavior Under Explosive Loading and Soil Models

Research has been performed by Clemson University and the Army Research Laboratory (CU-ARL) on the impact of the degree of saturation on sands and clayey sands with regard to landmine applications (Grujicic et al. October 2005, January 2007, March 2007, April 2007, January 2008). Some of the findings of that research are considered relevant to this work and are discussed in this section.

37

Sands consist of a soil skeleton of solid particles with void space between, as indicated in Exhibit 8-1[a]. The void space may be filled almost entirely with air (dry), entirely with water (fully saturated), or with some voids filled with air and some with water (partially saturated). Volume of voids, air, water and the degree of saturation are all related through phase relationships and visualized using phase diagrams in geotechnical engineering. Phase diagrams of partially saturated and fully saturated soils are shown in Exhibits 8-1[b] and [c], respectively.

[a] [b] [c]

Exhibit 8-1: [a] Soil skeleton consisting of solid soil particles and void space.

[b] Phase diagram of dry or partially saturated soil.

[c] Phase diagram of fully saturated soil.

Deformation of a mass of sand under an explosive blast load is believed to be a function of two mechanisms (Grujicic et al. January 2007):

1. Elastic deformations and/or fracture of the interparticle bonds/forces between the sand grains: Elastic deformations occur at low pressure levels, while frac-ture of the bonds occurs at higher pressure levels. The interparticle bonds/forces are a function of surface roughness and binders within the soil.

2. Elastic and plastic deformation of the constituent materials in the sand mass: The sand, air and water within the sand mass may compress in response to load. (Note: Actual fracture of sand particles is not considered. Because these particles are so small, this is a reasonable assumption for an earthfill dam.)

As noted in Section 3, rockfill dams are considered a separate category of dam from earth dams. Strength of rockfill in conventional dam engineering may be affected by the breakdown of larger particles under high stresses. Similarly, it is theorized that larger rockfill particles with inherent weaknesses could break down under a blast load. Such particle breakdown would introduce another mechanism of failure and deformation for the material as a whole. This mechanism has not been considered in this research or in research reviewed for this work; instead, it is meant to serve as an example as to why rockfill dams must be considered separate from earthfill dams within this research.

38

For relatively slow loading conditions (conventional geotechnical loading condi-tions), the air or water within the pore spaces is able to drain and the deformation of the sand mass is dependent on the response of the solid particles. However, for very rapid loading conditions (explosive blasts), pore fluids do not have the op-portunity to drain; therefore, the response of the soil mass is a combination of the response of the soil particles and the pore fluid (air and/or water).

For low levels of saturation, the soil mass is relatively dry and contains a relatively high volume of air relative to the volume of void space (Exhibit 8-1[b]). Therefore, interparticle bonds (friction and/or binder material strength) are relatively high and the mass contains a relatively high volume of compressible fluid (air) within the pore space. The compression of the air within the void space provides an ener-gy-dissipating effect on the overall blast impact for dry or partially saturated soil.

For completely saturated sand, interparticle forces are lower than for the dry case (water reduces friction, buoyancy reduces normal forces) and the void space is filled with incompressible fluid (water) as shown on Exhibit 8-1[c]. Therefore, the two mechanisms of deformation for the overall sand mass are both adversely af-fected for the saturated case: the interparticle bonds are reduced and the degree of compressibility is reduced.The compressibility effect is the most significant. The saturated sand lacks the energy-dissipating effect of the air within the pore spac-es and therefore behaves as an incompressible material (Grujicic et al. October 2005). As a result, blast impacts will be expected to be more significant for satu-rated soils than for dry or partially saturated soils.

Research on the effect of clay binder indicates that the clay may coat the sand particles and also serve to reduce interparticle bonds/forces. However, published work indicates that the addition of up to 15% clay binder has a less significant ef-fect than the degree of saturation (Grujicic et al. April 2007). This is consistent with other research findings (Grujicic et al. October 2005) that indicate that the particle size, particle distribution and presence of silt and/or clay on the blast response of sand is small compared to the degree of saturation.

There are several important implications in the published landmine research (Grujicic et al. October 2005, January 2007, March 2007, April 2007, January 2008) with regard to this research pertaining to dams. The degree of saturation is the most significant aspect of the soil model to be accounted for. As indicated previously, soil for a homogeneous dam is expected to be at or near saturation. Therefore, a soil model for a saturated soil is considered necessary for performing the blast-impact analyses. Second, is that the effect of clay or other binders is not significant compared to the degree of saturation. Therefore, for this research, it is less important whether the embankment material is an SC, SM, or some other sand-silt-clay mixture and more important whether the soil is dry or saturated.

39

8.3 Sources of Soil Material Models for Analyses

Based on the discussion above in Section 8.1, the default Sand-Compaction mod-el in the AUTODYN material library is not appropriate for blast-impact analyses of dams. Additional soil material models were developed based on research from two sources:

1. Modified compaction models developed by CU-ARL based on Grujicic et al. (October 2005, January 2007, March 2007, April 2007, January 2008): As described above in Section 8.2, CU-ARL performed research on saturated sand and clayey sand with regard to analyzing buried landmine impacts. This research focused on modification of the AUTODYN compaction model; the compaction model is the most widely used soil material model in military ap-plications (Grujicic et al. January 2008).

These soil material models were developed for similar soils (saturated sand and clayey sand) using the same software (AUTODYN) for use in blast-impact analyses (landmine research) and have been validated using field testing. Therefore, these models are considered appropriate for use in this analysis.

The models from the published CU-ARL research were re-created by approx-imating the piecewise functions presented in the published research. Minor changes were made to the tensile function in the soil models. The models were designated SC-SAT and SP-SAT for saturated clayey sand (SC) and sat-urated clean sand (SP) respectively.

2. Linear-failure models, as published by Parsons Brinckerhoff based on Choi (2009): Research and modeling has been performed to assess the impact of explosions in tunnels. This research includes two soil models, one for sand and one for clay. These models are linear-failure-mode models, not compac-tion models. These were also developed for use in AUTODYN and have been previously validated through Parsons Brinckerhoff’s research and applica-tion in practice. For this analysis, they were designated as Sand-Linear and Clay-Linear.

However, the main goal of Parsons Brinckerhoff’s research and subsequent analyses in practice has been to analyze the impacts of explosions on tunnels. Those analyses were focused on the impacts on the tunnel liner, which gen-erally consists of a concrete or steel structure surrounded by soil; the impact of explosions directly on soil was not of primary concern in those analyses. Since the soil plays a secondary role in those analyses, the soil models do not necessarily have to be highly accurate for direct impacts of explosive loading on soils. Therefore, although these models exist and can be used, they may not be the most applicable to the conditions being analyzed in this research.

40

8.4 Validation ModelsTo verify that the published soil models developed by CU-ARL (discussed in Section 8.3) had been re-created to a reasonable degree of accuracy, a valida-tion model presented in the published work (Grujicic et al. January 2007 and April 2007) was re-created. The published work includes a model of a buried explosive in a steel bucket backfilled with soil. This model and the field testing apparatus are shown in Exhibit 8-2.

The results of the analysis, the shape of the explosion, crater formation, etc. are reasonably consistent with the published reference for SP-SAT and SC-Sat as indicated in Exhibit 8-3. Therefore, the material models for SC-Sat and SP-SAT appear to have been reasonably re-created and are considered valid for further analyses.

The same model was also run using the Sand-Linear and Clay-Linear materi-al models (discussed in Section 8.3) based on Choi (2009) as indicated in item 2 above (linear-failure models). Results of these models, shown in Exhibit 8-4, indicate a less reasonable result in terms of the blast pattern and cratering effect. These models were therefore not considered further for this work.

Exhibit 8-2: Field apparatus for validation testing and validation model set-up from Grujicic et al. (January 2007).

41

Exhibit 8-3: Comparison of published AUTODYN model (left) [Grujicic et al January 2007] and re-created AUTODYN model (right) for saturated sand (SP-SAT).

Exhibit 8-4: Results of steel bucket validation models using sand-linear (left) and clay-linear (right)

soil models.

8.5 Selection of Soil Model for AnalysisBased on the anticipated material properties of the dam soils, review of published soil models and re-creation of the soil and validation models, the saturated clayey-sand model (SC-SAT) was selected for use in the analyses.

43

9.0 DEVELOPMENT OF TRIAL DAM NUMERICAL MODELS

44

9.0 DEVELOPMENT OF TRIAL DAM NUMERICAL MODELS

9.1 Model Set-upNumerical models were developed in AUTODYN using the trial dam size and geometry, the explosive size and location and the SC-Sat soil model described in Section 8. A three-dimensional model is necessary to capture the limited extent of the explosive. A two-dimensional cross-sectional model, as is typical of many geotechnical analyses, results in a model in which the explosive extends infinitely into the page; as a result, it is unrealistically large and the impacts of the explosion are greatly exaggerated.

The analysis was performed using a full-size, 50-foot (15-meter) high embankment model for primary assessment of blast impacts. A smaller, limited height model of only the upper 20 feet (6 meters) was also developed for investigating localized impacts. The crest width and slope configurations were the same as indicated in Section 3.2 for the trial dam: crest width of 25 feet (7.6 meters) and slope configu-rations of 2.5H:1V upstream and 2H:1V downstream. The length of the dam along the dam centerline was varied from 20 to 40 feet (6 to 12 meters) in the analysis. The final models used a length of 40 feet (12 meters), including symmetry around the blast center. This length was necessary so that the blast impacts were not causing distortion at the model boundaries. The dam model included 183,000 elements, with smaller elements located near the crest in the area of the blast and then becoming progressively larger with distance from the blast.

Euler space for the air, reservoir water and explosive was created around the dam. The element mesh size for the Euler space was comparable to the dam. The explosive was modeled in two ways: as a mass of TNT within the numerical model and as a separate blast file remapped into the dam model. Results were compara-ble for both approaches to modeling the blast.

Boundary conditions around the edges of the model and base of the dam or dam section were set to allow transmission of the pressure waves out of the model and prevent reflection of the wave back into the embankment. In effect, this assumes the dam is on a soil foundation because a hard-rock foundation would reflect a portion of the pressure wave back into the embankment.

Arrays of data collection points were placed under the crest of the dam. This al-lowed data on pressure, particle velocity, etc. to be collected during the analysis for subsequent evaluation, including exporting the data to spreadsheets and other analytical programs.

45

The models are shown in Exhibits 9-1(a) and (b). Based on the model develop-ment, the following aspects of blast-impact models should be noted for conduct-ing such assessments:

• 3-D models are necessary to account for the limited extent of the blast.

• Sufficient length along the dam centerline is necessary to avoid distortions at the model limits.

• Boundary conditions should be carefully considered to reflect the type of foun-dation expected in the field.

Exhibit 9-1(a): 50-foot (15-meter) dam model, explosive over upstream side of crest.

Exhibit 9-1(b): 20-foot (6-meter) dam model, explosive over downstream side of crest, 3-D view.

46

9.2 Model ResultsThe models can be run to examine a variety of parameters. For these assess-ments, effective strain and pressure were reviewed. Screen shots of model results are presented in Exhibits 9-2(a) and (b). Additional data was collected at each of the gauge points for ground motions and displacements; the data was then exported to Microsoft Excel for plotting, review and input into other programs. Acceleration of the dam part as a whole was also assessed.

Exhibit 9-2(a): Model results during blast showing propagation of pressure wave through the dam.

Exhibit 9-2(b): Model results after blast showing effective strain contours at 3-D dam cross section.

47

10.0 ASSESSMENT OF RESULTS

48

10.0 ASSESSMENT OF RESULTSAs discussed in Section 5.1: Possible Modes of Failure, the impact of the blast is assessed with regard to three potential failure modes: dynamic stability during the blast, localized breach potential following the blast and post-blast global stability impact based on the altered cross-sectional geometry.

10.1 Assessment of Dynamic Global StabilityThe global stability during the blast is one of the most difficult aspects of the analysis to evaluate. The explicit-dynamics software is not designed to perform slope stability anal-yses. Therefore, the stability analyses must be performed with conventional geotechni-cal slope stability software. The input to the stability analyses must include output from the explicit-dynamics software to capture the effect of the blast impact. For the research in this monograph, stability analyses were performed using a 2-D critical cross section.This is consistent with typical slope stability analyses, but it requires reducing the results of the 3-D explicit-dynamics model to a representative input for a 2-D section.

In this research, two methods of evaluating the global stability impact of the blast were used. The first approach involved modeling the blast as a seismic event. The second approach involved using ground motion data recorded at gauge points in the explicit-dynamics model for input in the stability analyses.

10.1.1 Dynamic Global Stability Analyzed as a Seismic EventIn order to analyze the global stability as a seismic event, the overall acceleration of the dam is determined from AUTODYN. AUTODYN can calculate the accel-eration of a model part resulting from the dynamic impact or dynamic effect of a blast; in this case, the part is the entire dam.

A full 50-foot (15-meter) model section with a length along the centerline of 40 feet (12 meters) was analyzed under a 5,000-pound (2,270-kilogram) equivalent of a TNT explosive blast over the crest, as previously described in the discussion of the set-up for the trial dam numerical model (Section 9.1). The overall acceleration of the dam in each direction was determined (x, y and z directions) as shown in Exhibit 10-1.

The accelerations were input into the program Slope/W by GeoStudio using a cross section based on the trial dam (see Exhibit 3.4: Trial Dam Cross-Section Ge-ometry for the steady-state seepage Slope/W model). The peak accelerations in the horizontal and vertical directions relative to the dam cross section were input as seismic accelerations in the slope stability analysis. These peak accelerations are approximately -3x10-4(g) for vertical acceleration and 9x10-6(g) for horizontal acceleration. The acceleration along the centerline of the dam was not included as the GeoStudio models are 2-D cross-sectional models.

49

The results of this analysis showed no significant drop in the global stability FS as a result of the explosion. The FS during the blast was determined to be 1.499 compared to 1.500 for steady-state global stability. These results are discussed in Section 10.1.3.

10.1.2 Dynamic Global Stability Analysis Based on Ground MotionsAnalyzing the impact of the explosion as a seismic event does not account for high ground motions in the immediate vicinity of the blast. Therefore, an attempt was made to analyze the dynamic blast impact using ground motions recorded at a series of gauge points in the crest of the dam.

An array of gauge points was created across the crest of the dam within the AU-TODYN model as shown in Exhibit 9-1(b). These gauge points were located at a depth below the crest of 6 feet (1.8 meters). The gauge points were used to record data, including force and ground motions. The data was exported to Microsoft Excel and averaged over the area of the blast impact. This was done to develop an approximation of average-time history data for applied pressures and ground motions to be used in a 2-D cross-sectional model of the area impacted by the blast. The time histories for these average data sets were input into a 2-D model in Quake/W at the depth of the gauge points.

Exhibit 10-1: Peak acceleration (g’s) of trial dam resulting from 5,000-pound (2,270-kilogram) TNT blast on upstream crest.

50

The velocity vs. time data was entered into Quake/W, which then generated ground displacement vs. time functions. Quake/W modeled the effect of this displace-ment function and developed internal stresses within the embankment based on the ground motions. The stresses from Quake/W were imported into Slope/W and the global stability was analyzed. This Slope/W analysis was performed using the Quake/W stresses as well as a Newmark analysis. Both resulted in severe drops in the overall FS value, with resulting FS values below 0.1, as shown in Exhibit 10-2.

The analysis was repeated with another series of gauge points added at a depth of 21 feet (6.4 meters) below the crest. The intent of this analysis was to collect gauge point data that would be below the area of localized impacts and there-fore perhaps more indicative of the internal stability of the overall structure. The gauge point data was averaged as before and then exported and used as be-fore in Quake/W and Slope/W. The results were similar, with a severe drop in FS, although in this case the FS value reached a minimum of about 0.2 instead of approaching 0.0 as with the data from a depth of 6 feet (1.8 meters). These results are discussed below in Section 10.1.3.

Exhibit 10-2: Results of dynamic global stability assessment using gauge point data of 6 feet (1.8

meters) below crest.

51

10.1.3 Commentary on Dynamic Global Stability AssessmentsThe dynamic global stability methods used in this assessment provided drastically different results. One approach showed virtually no impact to the overall FS value, while the second approach showed complete loss of strength with a FS value of approximately zero. These two assessments and considerations regarding these and other approaches are discussed below.

Analyzing the blast event as a seismic event seems appropriate because the ground motions are similar. In both cases, a sudden acceleration of the dam oc-curs and rapidly dissipates. However, in an earthquake, the accelerations come from the ground underneath the dam as the entire site experiences the motion. In the blast scenario, the acceleration of the dam part results from a concentrat-ed load at the crest, which is then applied as an average site acceleration. It is not clear how much this averaging of the focused blast load as an approximate ground motion has on the overall analysis results. This analysis also does not cap-ture the impact of the localized ground motions directly under the blast, which can be several orders-of-magnitude higher than earthquake-induced ground motions.

In addition, the dam part in the explicit-dynamics numerical model is a 3-D object. The mass of the dam will have a direct impact on the average acceleration that is reported from the analysis, since a greater mass will reduce the overall average acceleration for the same explosive blast (greater mass results in more inertia, which will, under the same applied force, result in a smaller average acceleration).There-fore, the average acceleration value is a function of the length of the numerical mod-el section. The length of the section should be minimized to the local area affected by the blast to ensure that the acceleration value is indicative of the blast section and does not include the effects of extraneous dam length outside of the blast area. However, as noted in the discussion of model set-up (Section 9.1), the length of the dam in the model must be sufficient so that distortions and errors do not occur as a result of the boundary conditions being within the area of the blast impact.

Even with these considerations, the results may not be unreasonable. As discussed in Section 1.3, the Sorpe Dam and Peruca Dam were affected by high explosives at closer standoff distances, yet they did not suffer global stability failures. However, without field validation data, definitive conclusions cannot be drawn regarding the accuracy of the seismic-analysis approach.

In a similar fashion, there are issues with regard to the approach based on ground motion input.The motions input into this analysis were based on gauge points at a variety of depths ranging from the crest surface to a depth of 6 feet (1.8 meters) below the crest. Data from these points (AUTODYN output) indicates extreme-ly high applied pressures—pressures on the order of over 1 million pounds per square foot (psf) (47,900 kilopascals [kPA]) even at a depth of 6 feet (1.8 meters) below the crest. The output of the Quake/W model shows consistent stresses calculated based on the ground motions.

52

An issue arises in that these stresses exceed the strength of the materials. The soil would fail before those stresses could be transferred deeper into the structure. In effect, this is exactly what happens, since in reality a crater forms as a result of the applied stresses exceeding the bonds between the soil particles. Howev-er, in the numerical models in both Quake/W and AUTODYN, the models contain finite element meshes which deform, but do not fail. The embankment structure is assumed to remain intact within the numerical modeling programs and therefore these very high stresses are transferred throughout the embankment, resulting in very high internal stresses that lead to FS values near zero. This explains why even when the gauge points were placed deeper within the embankment, at a depth of 21 feet (6.4 meters) below the crest, the stresses were still relatively high and the FS values dropped to near zero. It would be expected that in the field, such stresses would not transfer throughout the embankment because of the fail-ure of soil bonds once the internal soil strengths are exceeded.

The general approach of using the ground motions from AUTODYN as input for seismic analysis remains a reasonable approach. However, it appears that a modification to the soil model or development of a new soil model is needed to adequately capture the internal failure mechanisms that may occur during blast loading for the purpose of analyzing dynamic global stability. The soil models used in these analyses are modified compaction models, which have been used suc-cessfully for assessing localized blast and deformation patterns. However, these models may not adequately capture the internal failure mechanisms that affect stress propagation through a large soil structure.

Further exploration using field trials would be needed to collect and evaluate data from various depths throughout an embankment. This data could then be used to modify or create soil material models that would be applicable to global stability assessments. However, this would be a time-consuming and labor-intensive pro-cess requiring field trials, data collection, evaluation and iterations to develop and adjust the soil models to fit the field data.

It is possible that separate soil models will be needed for assessing global stability impacts than for assessing localized impacts, such as crater formation and de-formation. This is consistent with other material applications in AUTODYN. Many materials within the AUTODYN materials library have multiple types of models that are used for different applications and failure criteria.

Another consideration regarding the results of the global stability analysis using the gauge point data is that the blast loading occurs over a period of milliseconds. Although the FS drops well below 1.0, it is uncertain how meaningful this is given the time frame of the blast loading. Actual sliding failure would be expected to take several minutes at least, which is several orders-of-magnitude longer than the actual blast. This may be enough time for the internal pressures to dissipate and reestablish a higher FS value. Slope stability analyses do not generally include

53

time considerations; however, experience has indicated that slopes with FS values at or below unity may continue to stand for periods of days to years until internal pressures equalize. Therefore, it is uncertain how meaningful the drop in FS is given the time frame of the blast duration.

At the time when this monograph was written, both methods presented in this section—the seismic-analysis approach (Section 10.1.1) and the approach based on ground motion (Section 10.1.2)—appear to have a reasonable basis, yet pro-duce significantly different results, which demonstrates that field validation, in-cluding instrumentation data, is needed.

10.2 Localized Blast ImpactThe blast results in a localized impact on the crest which will produce a crater and local damage as a result of high localized forces and strains. This localized dam-age results in the potential for an erosive breach and the formation of a different post-blast geometry. Therefore, assessing the localized blast impact is critical to assessing the potential for dam breach and for post-blast global stability impacts.

The analyzed models show a cratering effect which is consistent with expecta-tions. However, it is possible that blast impacts extend below the depth of the crater itself. Reviewing the model results in terms of effective strains indicates that the impacts of the explosion extend well below the immediate cratering depth, as shown in Exhibit 10-3.

Exhibit 10-3: Model results for a blast over the upstream side of the crest, showing effective strains.

54

Based on the strains that extend below the crater, it is possible that the damage resulting from the blast may penetrate significantly deeper into the embankment. Most dams are made of soils that include some degree of cohesion and therefore may be capable of developing cracks.

Crack formation in dams is a widely recognized concern and may result from ten-sion, differential movements or shrinkage (USACE 2004). Cracked soil sections are often included in slope stability assessments and in simplified stability charts (USACE 2003). In a similar fashion, it is considered a possibility for cracks to devel-op as a result of shearing and deformations caused by the explosive impact. The shearing and deformation effects may extend below the depth of the surface crater; therefore, the possibility exists for internal damage to the dam that may adversely affect the overall stability or integrity of the structure. Uncontrolled seepage or inflow of water into a crater or cracked section may lead to erosion or piping which could eventually result in an uncontrolled breach of the dam. It is important to be able to assess this damage potential because it may not be readily observable or apparent and repairs for a blast impact may need to extend beyond filling of the crater.

Criteria for assessing a depth to which cracking may occur must be developed in order to assess the localized blast impact. In conventional slope stability, potential crack depth is based on soil cohesion. In effect, soils with cohesion are cable of exhibiting tension and forming vertical cracks. However, blast-loading conditions are of such high intensity and short duration that conventional shear strengths are not directly considered in the material models and analyses. Another criterion that is often considered in dam engineering is limiting strain. Limiting strain values can be used as a criterion for the interpretation of shear-strength data. Strain compat-ibility and strength-reduction factors are considerations for dam and foundation soils of differing strength types (USACE 2003).

The use of a limiting strain criterion for developing an estimated depth of possible cracking is proposed for assessing localized impacts. In conventional geotech-nical engineering practice, limiting strains for interpretation of shear-strength and strain-compatibility assessments are often in the range of 5% to 15%. For this application, strains in the range of 10% to 30% are considered. A strain of 10% is consistent with the range of conventional assessments. Blast-model results indicate that above 30%, the strain contours are relatively tightly spaced, indicat-ing high deformations that are likely indicators of soil-structure failure, excessive displacement or cratering. The range of 10% to 30% is considered a range over which the soil remains primarily intact, but may still be subject to strains large enough to produce damage or cracks.

Based on the above, the localized impacts consist of the development of a crater and a depth or zone of potential crack formation below the crater. These localized effects are assessed using strain data generated in the blast model.

55

10.2.1 Development of Simplified ChartEuler-Lagrange nonlinear dynamic analyses involve complex and time-intensive ana-lytical skills. A significant effort is needed to prepare, perform and interpret an analyti-cal model for blast impacts. A significant benefit in terms of time and resources could be realized from the development of simplified tools to aid in assessing the impacts of explosions on dams. A simplified chart for assessing localized explosive blast impacts was developed for this purpose and is presented in this monograph.

Analyses were run with the blast centered on both the upstream and downstream lanes to facilitate the assessment of a blast anywhere within reason on the crest roadway. (It is assumed that the delivery platform/vehicle could not be centered at the edge of the roadway because a guardrail is likely to be there which would restrict that placement; in addition, the width of the vehicle would preclude center-ing the blast on the edge of the crest.)

The localized impact of the blast results in an irregular pattern of deformation resulting from the slopes of the dam and the different confining and overbur-den conditions that exist. Contours of strain are developed from the blast model results. As shown in Exhibits 10-3 and 10-4, the maximum depth of strain does not occur directly under the blast center; instead, it is offset towards the slope because of the lack of external confining pressure, which causes the maximum depth of strain to occur between the center of the explosion and the edge of the crest. Strains extend laterally and upwards from the maximum depth to the exter-nal slope. This pattern is evident for blasts on both the upstream and downstream sides of the crest, as shown in Exhibit 10-4. Depth of maximum strains below the crest and below the slope is comparable for blasts on the upstream and down-stream sides of the crest.

Exhibit 10-4: Strain contours for explosions over upstream and downstream sides of the crest.

The depth of strain propagation is comparable for the blast over the upstream and downstream sides of the crest. Slight differences are observed, possibly as the result of the differences in slopes (2.5H:1V vs.2.0H:1V). However, the differ-ences in the depth of strains on the upstream and downstream cases are gen-erally less than 10% of the total strain depth; therefore, the depth of strains are considered comparable.

56

The results of both the upstream and downstream blast analyses were superim-posed on each other to develop a composite pattern of strain propagation for a blast anywhere across the crest. A simplified “zone of potential cracking,” based on strain data, was developed that encompasses the results of both analyses.

For simplicity and conservatism, the point of deepest strain is moved to directly beneath the outer edge of the crest on both the upstream side and downstream side. The deepest point of strain propagation beneath the crest is used as a lower horizontal boundary. The point of deepest strain propagation below the upstream and downstream slopes is used for both slopes. The result is a simplified “zone of potential cracking” that allows an estimate to be made of potential damage result-ing from an explosive blast on the crest roadway. A similar procedure is used for developing a depth of potential crater formation. The depth of crater formation is taken from the 100% strain contour.

Exhibit 10-5 provides the explosive sizes of 5,000 pounds (2,270 kilograms) of TNT equivalent and 2,500 pounds (1,135 kilograms) of TNT equivalent. Data is presented for strains of 10%, 20% and 30% for assessing the zones of potential cracking as well as areas of potential crater formation. The chart is based on the trial dam section using the SC-SAT soil material model.

Exhibit 10-5: Simplified chart for estimating depths of potential crater and crack formation caused

by explosive blasts on the dam crest.

57

10.2.2 Application and Limitations of the Simplified ChartThe simplified chart can be used as a starting point for a quick assessment of po-tential blast impacts. The determination of which strain to use for an assessment is a matter of judgment based on the risk tolerance of the dam owner, regulatory agency and/or engineer as well as considerations regarding the dam materials, geometry and consequences of failure. Once the limiting strain criteria have been determined, the depth of potential cracking can be determined from the chart for a given explosive size.

The numerical values presented in the chart have been rounded up to the nearest whole number. Based on the application described above and the limitations indi-cated below, whole numbers were used so as to not imply an unreasonable level of accuracy in the data presented.

It is important to stress that the simplified chart is a useful tool for performing a quick initial assessment; it is not intended for final assessment or detailed design.

The following limitations apply to the chart:

1. Depths indicated in the chart do not include any consideration for freeboard or any safety margin. An allowance for freeboard above the reservoir or internal phreatic surface is recommended.

2. Data presented in the chart have not been validated at the time when this monograph was written. The chart has been developed using numerical model-ing techniques based on soil material models developed for landmine analyses. Although the numerical models described in this monograph have followed general industry standards and a rational approach, the results of these models, including the simplified chart presented above, have not been field validated.

3. The chart is based on strain data and the possibility of crack formation resulting from strains exceeding certain threshold values. The dam’s vulnerability to a breach is based on the premise that once cratering and cracking form, the dam will be more susceptible to the possibility of uncontrolled seepage paths and the potential for erosion/piping through the areas that have experienced high strains.

4. Estimated crater depth is based on 100% strain contour. The crater depth has not been verified or validated with other analyses or field trials. The strain levels in Ex-hibit 10-5 that are higher than approximately 30% are tightly grouped in the immedi-ate depths below the crater. Lower strain levels propagate significantly deeper.

5. The chart is intended to be used as part of an overall dam safety and securi-ty-risk assessment that includes an evaluation of structure vulnerability, risks and consequences of failure.

6. The chart has been developed for the trial dam and threat scenario described in this monograph. Deviations from these conditions must be assessed using engineering judgment.

58

10.3 Assessment of Post-Blast Global StabilityThe impact of the explosion results in an altered post-blast geometry that includes a crater and possible embankment cracking resulting from high strains. The post-blast global stability can be assessed by using conventional slope stability assess-ment methods and the estimated post-blast stability.

As discussed in Section 10.2, the high strains resulting from the blast impact may lead to cracking of the embankment soils. Conventional slope stability assess-ments often include considerations for possible tension cracks. Similarly, post-blast global stability assessments should include potential cracks resulting from the blast impact. As with conventional stability assessments, cracks should be assumed to be filled with water resulting from seepage or precipitation. The depth of potential cratering and cracking can be estimated from either the blast model results or the simplified chart.

The formation of a crater and cracked zone is expected to reduce the overall global stability of the dam. The formation of a crater/cracked section may result in a portion of the critical failure surface with no shear strength. The presence of wa-ter within a cracked section will provide an additional driving force on the critical failure section. The impact of the post-blast geometry on the trial dam is illustrated in Exhibit 10-6. The magnitude of the overall impact of these effects on the global stability will be a function of the size, geometry and material strengths of the dam.

For the trial dam, the overall impact on static, steady-state global stability is mi-nor; the difference in global steady-state stability FS between the pre- and post-blast geometry is less than a 10% drop. For a smaller dam height, the relative size of the crater and cracked section will affect a larger portion of the critical failure surface and, as a result, the impacts may be more significant.

59

Exhibit 10-6: Comparison of pre- and post-blast global stability of trial dam section.

61

11.0 PARAMETRIC STUDY

62

11.0 PARAMETRIC STUDYA limited parametric study was performed to assess the effects of differing materi-als and explosive sizes on the results of the blast model. Explosive loads of 5,000 pounds (2,270 kilograms) and 2,500 pounds (1,135 kilograms) equivalent TNT were modeled. The dam was assessed using the soil material models discussed in Section 8). Parametric analyses included the following conditions:

1. Trial dam made of SC-SAT, saturated clayey-sand model, base case (modified from Grujicic et al. January 2007).

2. Trial dam made of SP-SAT, saturated-sand model (modified from Grujicic et al. January 2007).

3. Trial dam made of Sand Compaction, default AUTODYN library compaction model for dry sand.

4. Trial dam made of SC-SAT, base case model, modified to include 2 feet (0.6 meters) of structural concrete on the crest.

11.1 Results of Parametric Soil Material Model Assessment

The parametric models for assessing the soil material models (items 1 to 3 above) were evaluated for explosive sizes of 2,500 and 5,000 pounds (1,135 and 2,270 kilograms) of TNT. The intent of this parametric assessment was to evaluate the impact of the different soil material models on the overall results. The strain depths for assessing the localized blast impact were used as the basis for com-parison. Depths of strains from the parametric models are shown in Exhibit 11-1. The results of the parametric study are illustrated in Exhibit 11-2(a), (b) and (c) showing sections in 2-D and 3-D.

63

Strain Level (%)

2,500 lb. (1,135 kg) TNT Equivalent 5,000 lb. (2,270 kg) TNT EquivalentDepth at

Slope (Y in Exh.

10-5)ft (m)

Max. Depth Below Crest

(X in Exh. 10-5)ft(m)

Crater Depth

(Z in Exh. 10-5)ft (m)

Depth at Slope

(Y in Exh. 10-5)ft (m)

Max. Depth Below Crest (X in Exh. 10-5)

ft(m)

Crater Depth

(Z in Exh. 10-5)ft (m)

SC-Sat (Base Case, Saturated Clayey Sand)10 % 9 (2.7) 23 (7.0)

13 (4.0)13 (4.0) 28 (8.5)

17 (5.2)20 % 8 (2.4) 19 (5.8) 11 (3.4) 23 (7.0)30 % 7 (2.1) 17 (5.2) 9 (2.7) 21 (6.4)

SP-Sat (Modified Compaction Model, Saturated Clean Sand)10 % 5 (1.5) 13 (4.0)

5 (1.5)7 (2.1) 17 (5.2)

8 (2.4)20 % 3 (0.9) 11 (3.4) 5 (1.5) 14 (4.3)30 % 2 (0.6) 9 (2.7) 4 (1.2) 12 (3.7)

Sand-Comp. (AUTODYN Default Compaction Model, Dry Sand)10 % 0 (0) 6 (1.8)

2 (0.6)0 (0) 7 (2.1)

3 (0.9)20 % 0 (0) 2 (0.6) 0 (0) 3 (0.9)30 % 0 (0) 2 (0.6) 0 (0) 3 (0.9)

Exhibit 11-1: Results of parametric study of various soil-material models.

64

Exhibit 11-2: 2-D and 3-D sections showing comparison of model results (strains) for saturated clay-ey sand (SC-SAT), saturated clean sand (SP-SAT) and dry sand (Sand Compaction)

Exhibit 11-2(a): Trial dam composed of SC-SAT.

Exhibit 11-2(b): Trial dam composed of SP-SAT.

Exhibit 11-2(c): Trial dam composed of sand compaction.

As Exhibits 11-1 and 11-2 show, the selection of the soil material model has a sig-nificant effect on the results. The saturated clayey-sand model results in the most significant impact, followed by the saturated-sand and dry-sand models. This is consistent with the discussion in Section 8.2, which indicated that the effect of saturation and clay fines would be to reduce the energy dissipation effect of the soil material.

Although the blast-impact models presented in this monograph have not been validated in the field, the results are considered conceptually accurate. Further-more, the parametric study results clearly indicate the need to use appropriate soil material models that adequately consider the effects of saturation and clay fines for blast-impact assessments of dams and levees.

The results of the model for dry sand indicate that the localized impacts are great-ly reduced compared to the impacts for saturated sand models. This is unlikely to be a realistic scenario in the field because a dam or levee will have some degree

65

of saturation resulting from construction procedures, seepage and precipitation. However, it is worthwhile to note how significant the energy-absorbing effect of the dry sand is in mitigating the impact of the blast. As discussed in Section 8.2, this effect results primarily from compaction of air within the pore spaces. While it is not possible to include dry sands in the construction of a dam or roadbed, the capability of dry sand to absorb energy does raise the possibility that material with entrained air—such as air-entrained flowable fill or foamed concrete—could be used as an energy-dissipating layer within the upper portion of the embankment, possibly within the pavement or subgrade section, for this blast-impact threat scenario. However, further analysis of this possibility requires the application of an explicit-dynamics material model capable of evaluating entrained-air materials.

11.2 Results of Parametric Model with ConcreteA partial dam model consisting of the upper 20 feet (6 meters) of the embankment was analyzed with the inclusion of 1 to 2 feet (0.3 to 0.6 meters) of concrete as a crest roadway section or structural-hardening measure. The concrete extends across the width of the crest except for a 2-foot (0.6-meter) space at each edge to accommodate the installation of guardrails, fencing, etc. as shown in Exhibit 11-3. The concrete was modeled with a structural concrete material model for 5,000 pounds per square inch (psi) [35,000 kPa] concrete found in the AUTODYN materials library.

The intent of analyzing 1 to 2 feet (0.3 to 0.6 meters) of concrete on the crest sur-face was to conceptually explore the potential for mitigation measures related to the crest roadway. Many roadways are built of asphalt rather than concrete and many pavement sections are much thinner than 2 feet (0.6 meters); therefore, this partial dam model would be more indicative of an upgraded or hardened structure. The ob-jective of the analysis was to determine what impact the structural concrete would have on the overall blast impact. As with the other parametric models, the results were evaluated in terms of the strain propagation within the embankment.

The results indicate that a crater still forms and strains still propagate into the embankment. However, the size (depth and diameter) of the crater and the pattern of strains is different than without the concrete section. When the concrete is in place, the high strains in the embankment are concentrated immediately below the concrete section, while the strains above 10% are relatively tightly spaced, as can be seen in Exhibit 11-4. The strains and deformations within the concrete are intended for evaluation; as a structural material model, this material has to be assessed with regard to the structural damage criteria established in the material model. Because the intent is to assess the impact to the dam in comparison to sections without the concrete, the focus here is placed on the pattern of strains within the embankment, rather than on the performance of the concrete section.

This model result indicates that the inclusion of 2 feet (0.6 meters) of concrete on the crest surface may help to mitigate the impacts deeper into the embank-

66

ment. Specifically, the presence of the concrete reduces the size of the crater and the depths to which strains—especially the higher strains—propagate within the embankment. However, it must be stressed that this model is meant to be considered illustrative and should not be considered highly accurate. The model uses a structural concrete material model together with a compaction model for the embankment soil. But without further research and/or validation, it cannot be definitely stated that these two different types of material models can be used together in a single structural model and still be capable of producing valid results. For example, the strain compatibility of the two materials has not been specifical-ly accounted for in the models. Additional research is needed to assess whether these materials can be used together and/or what modifications must be made in the model set-up or what considerations may be needed in evaluating the results.

Exhibit 11-3: Partial dam model (upper 20 feet [6 meters] of embankment) with 2 feet (0.6 meters) of concrete on crest.

67

Exhibit 11-4: Results of model (strains) with 2 feet (0.6 meters) of concrete on crest.

69

12.0 MITIGATION MEASURES

70

12.0 MITIGATION MEASURESIn the event that a dam is found to be vulnerable to the impacts of an explosive blast, mitigation measures should be implemented to reduce the risk of adverse conse-quences to acceptable levels. This is consistent with general geotechnical dam safety assessments in which mitigation measures would be designed and implemented if the risk over global stability failure, piping or seismic stability were concerns.

For the threat scenario described in this monograph, several types of mitigation measures are possible. These include:

1. Site access.

2. Operational protocols.

3. Structural improvements.

These measures are discussed in the following sections.

12.1 Site AccessOne of the key premises of the threat scenario for the models described in this monograph is that the dam crest contains a public roadway or is easily accessible. One possible mitigation technique is to restrict access to the crest of the dam. Access restrictions can be variable, depending on the details of the threat scenar-io and the risks involved.

The simplest solution may be to close the roadway on the dam crest to prevent ac-cess. The threat scenario described in this monograph pertains to relatively large ex-plosives, which would require a delivery platform consisting of a vehicle of some type (truck bomb, etc.). The restrictions, therefore, pertain to restricting vehicular access.

Physical barricades would be needed to prevent unauthorized access, while still making it possible for emergency personnel to access the crest or opposite abutment if necessary. If the roadway crest is easily accessible, but is not a pub-lic roadway—as is the case for many levees—then restricting access to the crest may not have significant impacts to the public. If the roadway on the crest is public, this may require construction of an alternative roadway location to bypass the dam crest. This will require coordination with other agencies and the public as well. This mitigation technique has been implemented on several high-profile dams, but it is costly because it requires construction of a bypass roadway.

12.2 Operational ProtocolsOperational protocols refer to how the operations at the dam site are conducted. This term applies to a number of aspects of dam operation. A few examples are

71

discussed in this section. However, operational protocols are variable depending on the size, type, location and purpose of a dam; therefore a comprehensive dis-cussion is beyond the scope of this monograph.

An example of an operational protocol is the management of the reservoir level and management of dam or levee access based on reservoir levels. If a dam is found to be susceptible to a possible breach resulting from the reservoir level being above the depth of possible cratering or cracking, then one possible mitigation measure would be to maintain the reservoir level below the depth of cratering/cracking.

In many cases, dams accommodate variable reservoir levels up to a high PMF event. If the dam is only at risk for a breach during high-water events, then one possible mitigation measure would be to restrict access to the crest of the dam during high-water events that are above a certain elevation. That elevation would be based on the depth of cratering/cracking plus considerations for freeboard.This would result in temporary road closures during certain high-water events. Simi-larly, access to the crest of a levee could be restricted only during flood events. Access restrictions under such conditions can be achieved with either police barricades, temporary barriers such as concrete barriers, removable bollards or moveable barriers installed within the roadway section. An example of a movable vehicle barrier is shown in Exhibit 12-1.

Exhibit 12-1: Moveable vehicle barrier with a very shallow foundation (by Delta Scientific Corp.).

Another operational protocol risk-mitigation measure would be to add or increase the presence of security personnel and/or patrols. This technique is used to reduce threat risks for many other types of high-risk infrastructure, such as bridges and tunnels. As with access restrictions, the use of security or police patrols could also be linked to reservoir levels, with more patrols or personnel onsite during high-water levels.

72

12.3 Structural ImprovementsStructural improvements involve construction modifications to the dam itself with the intent of improving the structure’s resistance to blast impacts. For the threat scenario in this monograph (a truck bomb on the dam crest), structural improve-ments would involve improving the dam crest so that the blast impacts are below some threshold levels of impact. This involves determining what the critical modes of failure are and designing improvement measures around them.

A few examples of structural improvements to improve the design of a dam (exist-ing or new) could include:

• Raising the dam crest to increase the clearance between the crest roadway and the reservoir as a possible structural improvement measure. This could involve a significant amount of earthwork and construction in order to provide enough material to dissipate the blast impact between the roadway and reservoir level.

• Adding more blast resistant materials to the crest. This could include several types of materials based on the results of this analyses and parametric study.

– Structural strengthening by providing higher strength materials on the crest of the dam. For example, increasing the thickness and strength of the pavement section on the crest to include a thicker section of structural concrete. Stronger materials will dissipate the initial blast energy more and help reduce the depth of the crater and strain propagation. This is illustrat-ed conceptually in the parametric study with concrete on the dam crest.

– Inclusion of air-entrained materials within the upper portion of the crest may help mitigate the blast impacts through dissipation of the blast energy through the compaction of the void space. This is discussed in Section 11.1 (Results of Parametric Soil-Material Model Assessment) with regard to the dry-sand parametric model results. This could involve the use of air-entrained materials, such as foamed concrete, air-entrained flowable fill, and geofoam or some similar material that would have a high air content and yet be struc-turally strong enough to support the roadway and would not become saturat-ed over time from precipitation or capillary rise. These materials would likely have to be within the crest freeboard section, above the highest possible reservoir elevation, to avoid potential seepage issues or concerns with these materials. Because dams, levees and roadway subgrades typically require compacted soils near or wet of optimum moisture content—and because soils can become saturated over time from seepage through the embank-ment or along the roadway subbase—it is unlikely that placing dry soil in the upper portion of the crest would be a feasible long-term measure.

– Inclusion of clean granular materials within the upper portion of the dam. The parametric study results indicate that the localized blast impact is less for clean sand than for clayey sand. The use of clean granular material may

73

help mitigate the localized blast impacts. Similar to the discussion above for air-entrained materials, clean granular materials would be most appro-priate above the maximum water level in order to prevent seepage-related issues because these materials would provide a preferential flow path. Al-ternatively, seepage prevention measures, such as an upstream or internal cut-off or core wall, could be included with clean granular materials and the depth of these materials could be increased.

– A combination of the options listed here. For example, the upper portion of an embankment dam, between conservation pool or flood level and the crest, could be built with clean granular materials, capped with flowable fill, and the crest road constructed out of concrete.

• The inclusion of ductile materials, such as fibers, geogrids or other types of reinforcement, may provide some benefit as well. Conceptually, such materials could provide additional strength and ductility for resisting the blast impacts by increasing the ability of the embankment materials to withstand the impacts of the blast, similar to reinforcements in concrete and fiberglass. The result could be that strains do not propagate as deep or that the embankment materials can withstand higher strains without cracking or formation of preferential flow paths. Such materials could be incorporated into the materials as they are placed, either as layers, such as geogrids, or as randomly oriented elements dispersed within the soil matrix, similar to fiber-reinforced concrete. However, at this stage of development, the inclusion of ductile materials is only a concept. Material models of the composite embankment materials would have to be developed for analysis and field trials with and without the ductile materials would need to be conducted to assess whether there is a benefit to this concept.

• In regard to potential crack formation, graded mineral filters (i.e., sand filters) are commonly used in conventional dam design to prevent internal erosion or pip-ing. However, it is unclear if such filters could be expected to survive the blast impact with enough confidence to be relied upon to prevent internal erosion following a blast impact.

Detailed structural modifications have not been explored as part of this monograph. Structural modifications will involve detailed assessments and modeling to establish that the proposed improvements will provide the desired mitigation effect. Where other construction materials are considered, such as concrete, roller-compacted concrete (RCC), foamed concrete, etc., the analyses will be dependent on adequate material models. Material models for materials that are not widely used, such as RCC or foamed concrete, may have to be developed as part of these analyses.

12.4 Summary of Mitigation MeasuresThe mitigation measures discussed in this section are intended to be conceptual examples of possible measures that can be used to reduce the vulnerability of a dam to an explosive impact on the crest. A site-specific risk assessment must be performed to identify all the possible vulnerabilities and mitigation measures that would be applicable to a specific structure.

These examples also illustrate how multiple risk-mitigation measures can be com-bined. For example, using security personnel to restrict access to the dam crest during high-water events would be an example of changing operational protocols and restricting site access. It may also be necessary to implement certain mea-sures temporarily until more significant measures can be implemented. An exam-ple would be the addition of security personnel/patrols until a public roadway can be rerouted off the dam crest.

The discussion presented here has pertained only to the particular threat sce-nario—an explosive blast on the dam crest—analyzed in this monograph. Other threat scenarios may exist. In some cases, a particular mitigation measure may be adaptable to multiple threat scenarios, such as security patrols. In other cases, separate mitigation measures address different threat scenarios (water barriers to restrict boat access, instead of roadway barriers to restrict vehicle access).

75

13.0 AREAS FOR FURTHER RESEARCH

76

13.0 AREAS FOR FURTHER RESEARCHThe research presented in this monograph is intended to serve as an initiative to expand the state of the practice. A number of areas for further research have be-come apparent through the preparation of this work. These include:

1. Validation of or development of soil material models for structures built of soil and subjected to explosive blast loads. Most work to date has been to ana-lyze soils in terms of crater formation or in terms of landmine analyses (ejecta, impacts on overlying vehicles, etc.). Very limited research has been performed on structures built of geomaterials. Such research should include:

a. Analysis of multiple types of soils, including sand, clays, rockfill and soil mixtures.

b. Analysis of soils under different degrees of saturation.

c. Consideration of different failure criteria, such as deformations, shear failure, etc. As previously discussed, a different failure criterion may be needed for assessing internal stresses for global stability than for localized deformations.

2. Validation of models for explosive impacts to embankments. Most previous research involving soil has focused on level ground conditions. The effect of reduced overburden resulting from sloped ground should be assessed and validated.

3. Additional analysis and validation of the impacts of the explosive blast on the global stability of an embankment during the blast itself. As indicated previous-ly, the effects of the dynamic impact throughout the embankment are not well understood. The dynamic loads are of such high impact and short duration that the conventional application of the available software methods does not appear to adequately model the impact of the blast loads. Explicit-dynamics software is not designed to model slope stability assessments; software for analyzing slope stability under dynamic loads is generally developed for either seismic stabili-ty—in which the whole site is shaking—or for relatively low-magnitude ground surface movements resulting from controlled blasting or construction vibrations, not high-intensity explosive loads that apply instantaneously.

4. Research regarding different foundation conditions. In this trial dam model, the base of the model (foundation) was set as a flow-out boundary, allowing the pressure waves to pass out of the base of the model. This implies a soil foun-dation comparable to or weaker than the embankment material. A different boundary condition may be appropriate for a different foundation condition, such as a reflective boundary for a rock foundation that may reflect part of the blast wave back into the embankment.

77

5. Research with regard to blast impacts on other types of embankment con-struction, such as zoned embankments, rockfill, or hydraulic fill structures, would be of benefit to the industry.

6. Development of material models and validation for other types of materials used in dam construction, or for materials that may be used as part of struc-tural improvements. This could include models for RCC, foamed concrete, flowable fill, etc.

7. Blast-impact analyses for other locations of explosive loads that may be applicable. Other locations will depend on site access. Such locations could include the toe of the dam (roadway at the toe for access) and at the reservoir/upstream slope interface (explosive delivered by boat).

8. Blast-impact analyses for appurtenant or ancillary structures, if accessible, such as spillways, etc.

9. Blast-impact analyses for other types of dams, such as gravity, arch, slab-and-buttress, etc.

79

14.0 CONCLUSIONS

80

14.0 CONCLUSIONSAnalyzing the impacts of explosions on embankment dams and levees requires a complex, multistage analysis. The following key conclusions have been reached as a result of this research:

1. Several modes of failure must be considered, including global stability during the blast impact, global stability following the blast, and localized blast im-pacts and the potential for erosion or piping to form a breach. Evaluating these modes of failure requires several types of analyses using multiple software programs. However, the analyses can be performed using commercially avail-able software.

2. The blast impacts should be evaluated using explicit-dynamics software to per-form Euler-Lagrange coupled analysis. This is consistent with industry practice for other types of blast-impact assessments. In this work, the impacts of explosions on the crest of dams were explored. However, these types of software assess-ments can be performed for explosions located anywhere on the dam or in the reservoir, as the Euler-Lagrange coupling allows the interaction of fluids (air, water and detonation products) and solids (dam materials) to be analyzed together in a single numerical model. Dams of different material construction, such as concrete, can also be analyzed. The basic requirements of the analyses are:

a. Blast-impact models should be 3-D to address finite explosive size.

b. Length of blast-impact models should be minimal, yet still long enough to prevent distortion or errors at the model boundaries.

c. Boundary conditions—especially foundation boundaries—should be care-fully evaluated and compared to actual field conditions.

3. Appropriate material models are necessary for the dam materials in the ex-plicit-dynamics numerical models. The results from these models are highly dependent on the material models. For analysis of dams, soil material models must account for the degree of saturation of the soil and the fines content.

4. The degree of saturation of embankment soils is a key aspect of the soil material models. Blast impacts will be more significant in fully saturated soils because of the incompressible nature of the pore fluid (water). Blast models and field validation experiments that do not reflect the anticipated degree of saturation within an embankment dam or levee may produce misleading results. It is considered conservative to assume that the dam or levee soils are fully saturated, since the fully saturated condition results in a higher degree of blast impact (crater depth, strain propagation, etc.).

5. Localized blast impacts may extend beyond and below the blast-impact crater and may lead to preferential seepage paths and internal erosion or piping. Lo-

81

calized blast impacts below the crater may not be immediately apparent in the aftermath of the explosion and seepage issues may take time to develop.

6. Criteria should be established for evaluating potential localized impacts that extend beyond the crater depth. In this work, limiting strain value has been proposed as the criterion to be used based on the assumption that some minimum level of strain/deformation will be needed to form cracks, preferential flow paths, or compromise the integrity of internal seepage control measures (clay core, filter materials, etc.).

7. A simplified chart was developed based on trial models that can be used to estimate the depth of strains below two sizes of explosions on the crest of a trial dam section. The simplified chart can be used as part of a larger safety and security risk-assessment evaluation. It should be noted that the simplified chart in this monograph has not been validated with field experiments as of the time the monograph was written.

8. Dynamic global stability requires careful assessment of the results of the blast-impact analysis. Field experimentation data is needed to further the understanding and development of models for this mode of failure. Dynamic stability was analysed using 2-D models and two procedures: 1) analyzing the overall acceleration of the dam similar to a seismic assessment and 2) ana-lyzing the effect of ground motions from an array of points directly under the blast. Both methods have shortcomings and can produce dramatically differ-ent results. Therefore, field validation testing is required to obtain actual data to validate or modify the modeling procedure.

9. Post-blast stability requires assessing the impact of the crater and possible crack formation on the overall stability of the dam. Conventional slope stability software can be used to perform this assessment. The simplified chart pre-sented in this monograph (see Section 10.2.1) can be used to estimate crater and crack depth formation in an initial evaluation if site-specific blast-impact analyses are not available.

10. Areas for further analysis, experimentation and research have been identified in Section 13.

11. A number of mitigation measures exist for reducing the risk and impacts of explosions on dams and levees. These measures can be focused on site ac-cess, facility operation, structural improvements, or hardening. A site-specific evaluation is necessary to assess which measures should be considered and implemented for a particular structure.

12. This state-of-the-art analytical research can best be validated by a field trial involving a blast test of an instrumented soil embankment with similar features to those considered in this monograph and taking into account scale effects. A laboratory experimental program with a centrifuge test may also be beneficial.

83

REFERENCES

84

REFERENCES1. American Society of Civil Engineers (ASCE). 2013 Report Card for America’s

Infrastructure. http://www.infrastructurereportcard.org/dams (accessed 2013).

2. BAUER Dam Rehabilitation Company FZE, “Cutter-Excavated Cut-off Wall to Restore the Peruca Dam in Croatia.” http://www.bauerdamrehabilitation.com/en/dam_projects/rehabilitation_upgrade_dams/peruca_dam/ (accessed May 2013).

3. Braimah, Abass, and Mohammad Rayhani, “How Explosives Affect Embank-ment Dams,” HydroWorld 20, no. 2 (March 1, 2012), http://www.hydroworld.com/articles/print/volume-20/issue-2/articles/dam-safety/how-explosives-af-fect.html (accessed May 2013).

4. Calcagno, Frank. “Historical Overview – Licensee Assessments (Lessons Learned).” Presentation to the Federal Energy Regulatory Commission (FERC), San Francisco, CA, January 27, 2004.

5. Choi, Sunghoon and William Pitard. “Blast Protective Design Seminar.” Pre-sentation to Parsons Brinckerhoff, Baltimore, MD, May 22, 2010.

6. Choi, Sunghoon. “Tunnel Stability Under Explosion.” Parsons Brinckerhoff 2003 William Barclay Parsons Fellowship, Monograph 19, Parsons Brincker-hoff, NY, July 2009.

7. Darnton, John. “Croats Rush Work on Crumbling Dam.” The New York Times, January 30, 1993. http://www.nytimes.com/1993/01/30/world/croats-rush-work-on-crumbling-dam.html/ (accessed May 2013).

8. Grujicic, M., B. Pandurangan, and B.A.Cheeseman, “The Effect of Degree of Saturation of Sand on Detonation Phenomena Associated with Shallow Buried and Ground Laid Mines.” Shock and Vibration 12 (October 2005): 1-21.

9. Grujicic, M., B. Pandurangan, B.A. Cheeseman, W. N. Roy, and R.R. Skaggs, “Application of the Modified Compaction Material Model to the Analysis of Landmine Detonation in Soil with Various Degrees of Water Saturation.” Shock and Vibration 14 (January 2007): 1-21.

10. Grujicic, M., B. Pandurangan, G.M. Mocko, S.T. Hung, B.A. Cheeseman, W.N. Roy, and R.R. Skaggs, “A Combined Multi-material Euler/Lagrange Computational Analysis of Blast Loading Resulting from Detonation of Buried Landmines.” Multi-discipline Modeling in Materials and Structures 4, no. 2 (March 2007): 105-124.

11. Grujicic, M., B. Pandurangan, N. Coutris, B.A. Cheeseman, W. N. Roy, and R.R. Skaggs, “Derivation and Validation of a Material Model for Clayey Sand for Use In Landmine Detonation Computational Analyses.” Multidisci-pline Modeling in Materials and Structures 5, no. 4 (April 2007): 311–344.

85

12. Grujicic, M., B. Pandurangan, R. Qiao, B.A. Cheeseman, W.N. Roy, R.R. Skaggs, and R. Gupta,“Parameterization of the Porous-material Model for Sand with Different Levels of Water Saturation.” Soil Dynamics and Earthquake Engineering 28, no. 1 (January 2008): 20-35.

13. United States Army Corps of Engineers (USACE). 2013 National Inventory of Dams. http://geo.usace.army.mil/pgis/f?p=397:5:0::NO (accessed May 2013).

14. —. “2010 National Inventory of Dams.” http://140.194.76.129/publications/eng-pamphlets/EP_360-1-23/EP_360-1-23.pdf (accessed January 2013).

15. —. Design and Construction of Levees. Engineer Manual, no. EM 1110-2-1913. Department of the Army. Washington, DC, April 30, 2000.

16. —. Slope Stability. Engineer Manual, no. EM 1110-2-1902. Department of the Army, Washington, DC, October 31, 2003.

17. —. General Design and Construction Considerations for Earth and Rock-fill Dams. Engineer Manual, no. EM 1110-2-2300. Department of the Army, Washington, DC, July 30, 2004.

18. United States Bureau of Reclamation (USBR). Design of Small Dams. 2nd ed. Pre-pared by the United States Department of the Interior. Washington, DC, 1973.

86

2009 WILLIAM BARCLAY PARSONS FELLOWSHIP FINALIST

James Parkes is a Senior Professional Associate with Parsons Brinckerhoff. He graduated summa cum laude in civil engineering from Virginia Tech in 1998, and earned a master’s degree in civil engineering from Virginia Tech in 1999. He began his career with Parsons Brinckerhoff in the firm’s New York office before becoming the geotechnical discipline lead in the Baltimore, Maryland office. He has worked on a range of infrastructure projects, including tunnels, underground stations, bridges, highways, and dams. His experience includes high-profile projects such

as East Side Access in New York, the Cooper River Bridge in South Carolina, the John J. Audubon Bridge in Louisiana, the Intercounty Connector in Maryland, and the Baltimore Red Line. His experience with dams and levees includes inspections, investigations, assessments, and design of new structures as well as upgrades and improvements to existing structures. He has worked on numerous embankment dams and levees of all sizes throughout the US, including several high-hazard dams, as well as design of a roller compacted concrete dam.