This action might not be possible to undo. Are you sure you want to continue?

# Risk Analysis for Dam Safety

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document

Version: Delta July 2008

Reclamation Document: Corps of Engineers Document: URS Document: UNSW Document: Risk Analysis Methodology – Appendix E UFC 22238839 UNICIV R 446

THE UNIVERSITY OF NEW SOUTH WALES

-DisclaimerReference: “A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping” Draft Guidance Document dated July 31, 2008. The following report represents a draft working document for internal use by the Bureau of Reclamation and the Corps of Engineers. If a copy of this document has been received, it should be used as information only. Methodologies described in the document are under evaluation. The Bureau of Reclamation and the Corps of Engineers have not endorsed it for estimating the probability of failure of embankment dams by internal erosion and piping. No effort will be made by either agency to provide anyone holding a copy with updates or corrections.

Authors:

Robin Fell, School of Civil and Environmental Engineering, University of New South Wales, Sydney

Mark Foster, URS Australia. John Cyganiewicz, Bureau of Reclamation George Sills, ERDC, US Army Corps of Engineers. Noah Vroman, ERDC US Army Corps of Engineers. Richard Davidson, URS Corporation.

Date: Status:

31 July 2008 Delta Version

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008

Contents

1

Introduction ------------------------------------------------------------------------------------------------- 1-1 1.1 1.2 1.3 1.4 General Training Terminology to Describe Embankment Types Terminology 1-1 1-1 1-1 1-3

2

Methodology------------------------------------------------------------------------------------------------ 2-1 2.1 2.2 2.3 Introduction General Process Information Review 2-1 2-1 2-2

3

Failure Modes and Load Partitioning -------------------------------------------------------------- 3-1 3.1 3.2 Event Tree General Failure Modes 3.2.1 Internal Erosion Through the Embankment 3.2.2 Internal Erosion Through the Foundation 3.2.3 Internal Erosion of the Embankment into or at the Foundation Identification of Failure Paths 3.3.1 Overview 3.3.2 Examples Failure Path Screening 3.4.1 Overview 3.4.2 Internal Erosion Through the Embankment 3.4.3 Internal Erosion Through the Foundation 3.4.4 Internal Erosion of the Embankment into or at the Foundation Partitioning of the Reservoir Levels 3.5.1 “Normal” and “flood” loading levels 3.5.2 Pool of record level 3.5.3 Partitioning of reservoir levels 3.5.4 Assessing Frequencies of Reservoir Loading Earthquake Load Partitioning 3-1 3-3 3-3 3-5 3-5 3-9 3-9 3-10 3-13 3-13 3-14 3-21 3-22 3-23 3-23 3-23 3-23 3-25 3-25

3.3 3.4

3.5

3.6 4

Application of Tables for Estimating Conditional Probabilities--------------------------- 4-1 4.1 4.2 4.3 4.4 General Approach Historical Frequencies of Cracks and Poorly Compacted or High Permeability Zones in Embankments Historical Frequencies for Internal Erosion in and into the Foundation Estimating Conditional Probabilities 4.4.1 Estimating Conditional Probabilities Using Relative Importance Factors and Likelihood Factors 4.4.2 Estimating Conditional Probabilities Using Scenario Tables 4.4.3 Estimating Conditional Probabilities Using Probability Estimate Tables Length Effects 4-1 4-1 4-4 4-4 4-4 4-5 4-5 4-5

4.5

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008

i

Contents

4.6

4.7

4.8 4.9

Nature of the estimates of probabilities given by the Toolbox 4.6.1 The Toolbox gives “Best Estimate” Probabilities 4.6.2 Adjusting the Toolbox Best Estimates 4.6.3 Limitations of the methods used in the Toolbox 4.6.4 Assessment of probabilities of failure for failure modes which are not covered by the Toolbox Modelling uncertainty in the estimates of conditional probabilities 4.7.1 Purpose of this section 4.7.2 Sensitivity analysis 4.7.3 Uncertainty analysis Summarizing (Making the case) Combining Probabilities 4.9.1 Adding Probabilities for Static, Hydrological and Seismic Loads 4.9.2 Development of Fragility Curves

4-13 4-13 4-13 4-14 4-15 4-16 4-16 4-16 4-17 4-21 4-22 4-22 4-23

5

Probability of Initiation of Erosion in Transverse Cracks in the Embankment ------ 5-1 5.1 5.2 Overall Approach Estimating the Probability of Transverse Cracking (P C ) in the Upper Part of the Dam 5.2.1 Likelihood of a Transverse Crack Due to Cross Valley Differential Settlement (IM1) 5.2.2 Likelihood of Transverse Cracking Resultant on Cross Section Settlement due to Poorly Compacted Shoulders (IM4) 5.2.3 Likelihood Of Transverse Cracking Due To Differential Settlements In Soil In The Foundation Beneath The Core (IM5) 5.2.4 Likelihood Of Transverse Cracking Due To Differential Settlements Due To Embankment Staging (IM6) 5.2.5 Likelihood of Transverse Cracking due to Desiccation (IM7, IM8) Estimating The Probability Of Transverse Cracking Or Hydraulic Fracture (P C ) In The Middle And Lower Parts Of The Dam 5.3.1 Likelihood of Transverse Cracking or Hydraulic Fracture Due To Cross Valley Differential Settlement (IM9) 5.3.2 Likelihood Of Transverse Cracking Or Hydraulic Fracture Due To Differential Settlement Causing Arching Of The Core Onto The Shoulders Of The Embankment (IM10) 5.3.3 Likelihood Of Transverse Cracking or Hydraulic Fracture Due To Differential Settlement In The Foundation Under The Core (IM11) 5.3.4 Likelihood of Transverse Cracking Or Hydraulic Fracture At The Foundation Contact Due To Small Scale Irregularities In The Foundation Profile Under The Core (IM12) 5.3.5 Probability Of Transverse Cracking or Hydraulic Fracture - Factors To Account For Observations And Measured Settlements Estimation Of The Probability Erosion Will Initiate In A Crack Or Hydraulic Fracture In An Embankment (P IC ). 5.4.1 Overall Approach 5.4.2 Details Of The Method 5-1 5-2 5-2 5-8 5-11 5-13 5-13 5-18 5-18 5-19 5-20 5-20 5-22 5-25 5-25 5-27

5.3

5.4

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008

ii

Contents

5.5 6

Estimation Of The Probability Of Transverse Cracks In The Embankment Caused By Earthquake (IM13)

5-37

Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment ---------------------------------------------------------- 6-1 6.1 6.2 Overall Approach Estimation Of The Probability Of A Continuous Poorly Compacted or High Permeability Zone In The Embankment Or On The Core-Foundation Contact 6.2.1 Poorly Compacted or High Permeability zone within the core (IM14) 6.2.2 Poorly Compacted or High permeability layer on the core-foundation contact (IM15) 6.2.3 Poorly Compacted or High Permeability Layer in the Embankment due to Freezing (IM16, IM17) Probability Of A Poorly Compacted or High Permeability Zone Around A Conduit or Features Allowing Erosion Into the Conduit 6.3.1 Poorly Compacted or High Permeability Zone Around A Conduit Through The Embankment (IM18) 6.3.2 Features Allowing Erosion into a Conduit (IM19) Poorly Compacted or High Permeability Zone or Gap Associated With A Spillway Or Abutment Wall 6.4.1 Approach 6.4.2 Poorly Compacted or High Permeability Zone Associated with a Spillway Or Abutment Wall (IM20) 6.4.3 Crack/Gap Adjacent to a Spillway or Abutment Wall (IM21) 6.4.4 Differential Settlement Adjacent to a Spillway or Abutment Wall (IM22) 6.4.5 Wrap around details for connection of embankment dam to concrete gravity dam (IM23) Probability Of Poorly Compacted or High Permeability Zone - Factors To Account For Observations Estimation Of The Probability Of Initiation Of Erosion In A Poorly Compacted or High Permeability Layer In The Embankment, Adjacent A Wall Or Around A Conduit 6.6.1 Screening Check On Soil Classification 6.6.2 Assessment Of The Probability Of Initiation Of Backward Erosion In A Layer Of Cohesionless Soil or Soil with Plasticity Index ≤ 7 6.6.3 Probability Of Initiation Of Erosion By Suffusion In A Layer Of Cohesionless Soil or Soil with Plasticity Index ≤ 7 (PI ≤ 12 for seepage gradients >4) 6.6.4 Probability Of Initiation Of Erosion In A Poorly Compacted or High Permeability Cohesive Soil Layer and in silt-sand-gravel soils in which collapse settlement may form a crack or flaw 6.6.5 Probability Of Initiation Of Erosion In A Poorly Compacted or High Permeability Cohesive Soil Layer Around A Conduit 6.6.6 Probability Of Initiation Of Erosion In A High Permeability Soil Due to Frost Action Allowance for the Presence of Unknown and Unpredictable Flaws in the Core of the Embankment 6-1 6-2 6-2 6-7 6-9 6-14 6-14 6-16 6-19 6-19 6-20 6-21 6-23 6-24 6-26 6-28 6-28 6-28 6-32 6-35 6-37 6-37 6-38

6.3

6.4

6.5 6.6

6.7

iii

Contents

6.7.1 6.7.2 6.7.3 6.7.4 7

Background Recommended Procedure Suggested size and location of the flaw and the probability of occurrence Assessing the Probability of Failure if the Flaw Were to Exist and Reporting the Outcome of the Analysis

6-38 6-39 6-39 6-40

Probability of Initiation of Erosion in a Soil Foundation------------------------------------- 7-1 7.1 7.2 Screening Check on Soil Classification Assessment Of The Probability Of Initiation Of Backward Erosion In A Layer Of Cohesionless Soil or Soil with Plasticity Index ≤ 7 In The Foundation (IM24) 7.2.1 Description of Method 7.2.2 Estimation of the probability of heave or reaching the critical gradient (P H ) from piezometer data and/or seepage flow net models. 7.2.3 An approximate method for assessing the probability of heave Estimating the probability of backward erosion given heave has occurred 7.3.1 Probability of initiation and progression of backward erosion if sand boils have been observed. 7.3.2 Estimating the probability of initiation and progression of backward erosion where heave or reaching the critical gradient is predicted or sand boils have been observed at higher reservoir levels (P IH ) 7.3.3 Estimation of the probability of initiation and progression of backward erosion (P INH ) for cases where heave is not predicted. 7.3.4 Estimation of the total probability of initiation and progression for this reservoir stage Estimation Of The Probability Of Initiation Of Suffusion In A Cohesionless Layer In The Foundation (IM25) Estimation Of The Probability Of Initiation Of Erosion In A Crack In Cohesive Soil In The Foundation (IM26) 7.5.1 Overall Approach 7.5.2 Some Factors To Consider In This Assessment And Suggested Method For Estimating The Probability Of A Continuous Crack 7-1 7-1 7-1 7-3 7-5 7-7 7-7 7-7 7-9 7-11 7-12 7-12 7-12 7-12

7.3

7.4 7.5

8

Probability of the Presence of Open or In Filled Defects in Rock Foundations ------------------------------------------------------------------------------------------------ 8-1 8.1 8.2 Overall Approach Probability of one or more continuous in filled or open defects in the rock in the foundation beneath the embankment (IM27) 8.2.1 Overview of method 8.2.2 Estimation of the probability of one or more continuous in filled or open defects in the rock foundation beneath the embankment based on geologic and topographic data 8.2.3 Estimation of the probability of one or more continuous in filled or open defects in the rock foundation beneath the embankment based on site investigations and construction data 8.2.4 Effects Of Blasting On The Foundation 8-1 8-4 8-4 8-5 8-8 8-8

iv

1 Calculating weighted averages of estimates A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-10 8-10 8-11 8-13 8-15 8-15 8-16 8-18 8-20 8-20 8-21 8-22 8-22 8-25 8-25 8-26 8-26 8-28 8-30 8-30 8-31 8-32 8-32 8-32 8-33 8-33 8-33 v .5 Probability of one or more continuous open or in filled features associated with other geological features such as landslides.2 Estimation of the probability of one or more continuous in filled or solution features in the rock foundation beneath the embankment based on geologic and topographic data 8.6.10.10.Contents 8.5.3 Width and extent of solution features 8.8 Likelihood of Grouting Not Being Effective in cutting off open or in filled defects or solution or other features 8.3 Estimation of the probability of erosion of infill initiating using first principles 8.1 Overview of method 8.3.7 Likelihood of Defects or Solution Features being in filled 8.6.12 Combining probabilities for a continuous open defect in rock and describing the defects and solution features 8.11.11 Probability that erosion of infill continues 8.12.9 Likelihood of Cut-off Walls Not Being Effective in cutting off open or in filled defects or solution or other features 8.4.5.2 Estimation of the probability of one or more continuous in filled or open valley bulge or rebound features in the rock foundation beneath the embankment based on geologic and topographic data 8.2 Width and extent of features associated with stress relief effects in the valley floor – valley bulge or rebound 8.1 Approach 8.10. construction and monitoring data 8. faults and shears 8.3 Estimation of the probability of one or more continuous in filled or open solution features in the rock foundation beneath the embankment based on site investigations.3.3.1 Extent of occurrence and width of defects associated with stress relief defects in the valley sides 8.1 Overview of Method 8.2 Probability of Filtered or Unfiltered Exit 8.3 Estimation of the probability of one or more continuous in filled valley bulge or rebound feature in the rock foundation beneath the embankment based on site investigations and construction data 8.1 Overview of method 8.11.1 Overview of method 8.6 Width and extent of open or in filled defects or solution features in the embankment foundation 8.4.2 Estimation of the probability of one or more continuous in filled or open other geological features in the rock foundation beneath the embankment based on site investigations and construction data 8.6.3 Probability of one or more continuous open or in filled valley bulge or rebound features in the rock in the foundation beneath the embankment 8.10 Probability that erosion of infill in the defects or solution feature initiates 8.4.2 Estimation of the probability of erosion of in fill initiating based on performance data 8.4 Probability of one or more continuous open or in filled solution features in the rock in the foundation beneath the embankment 8.

3.2 How to model scour into defects of varying width and persistence Probability of erosion of the core following hydraulic fracture due to arching in a narrow cut-off trench 9-1 9-1 9-2 9-2 9-2 9-3 9-4 9-5 9-6 9-6 9-6 9-10 9. joint or crack in the foundation.1 Probability of a continuous pathway of open defects and solution features in the rock foundation (PCR) 9.3 Probability for Continuation – Scenario 2 (Downstream shoulder can hold a crack or pipe) 10.13.6 9. in a wall or conduit) 10.3.3.1 9.1 Probability of Continuation for Internal Erosion in the Embankment 10.2.1.3 Probability of a continuous pathway for erosion of the core into a rock foundation (Ppath) Probability of a continuous pathway of coarse grained layers in soil foundations (IM29) Probability of the initiation of internal erosion by backward erosion or suffusion starting at the core-foundation contact.6.5 9.2 Probability for Continuation for Internal Erosion Through the Foundation 10.4 9.9-1 9.1 Approach 10.2 Likelihood treatment of the embankment cut-off foundation does not prevent contact of the core with open defects or solution or other features (PTI) 9.1 Internal Erosion Through the Embankment – Overall Approach 10.6 Probability for Continuation – Scenario 5 (Erosion into a toe drain) 10.3 Probability of Continuation for Internal erosion of the embankment at or into the foundation 10.13 Describing the defects and solution features and failure modes 8.1 Describing the defects and solution features in relation to the embankment details 8.1 Erosion into open joints in rock foundation 10-1 10-1 10-2 10-2 10-4 10-15 10-16 10-17 10-17 10-17 10-22 10-22 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 vi .4 Probability for Continuation – Scenario 3 (Filter/transition zone is present downstream of the core or a downstream shoulder zone which is not capable of holding a crack/pipe) 10.2 Describing the failure paths 9 8-34 8-35 8-35 8-36 Probability of Initiation of Erosion from Embankment into Foundation --------------.1.Contents 8.12.1. Probability of initiation of scour at the core-foundation contact 9.13.2 Probability of Filtered or Unfiltered Exit 10.6.7 10 Probability of Continuation ---------------------------------------------------------------------------10-1 10.2 9.1.2 Probability for Continuation – Scenario 1 (Homogeneous zoning) 10.1 The Steps to be followed 9.5 Probability for Continuation – Scenario 4 (Internal erosion into an open defect.1.2.2 Summing probabilities 8.3.3 General Principles Overall Approach Probability of a continuous pathway for erosion of the core of the embankment into a rock foundation (Ppath) IM28 9.1.

2.2. soil foundation and from embankment into foundation 13.4.3 Internal Erosion of the Embankment into or at the Foundation 11.4 Probability for Limitation of Flows 11.4 Estimation of the Probability of Breach by Sloughing or Unravelling 13.1 Screening of Breach Mechanisms 13.3.3 Probability of Crack Filling Action Not Being Effective 11.3 Estimation of the probability of Slope Instability of the embankment initiates for internal erosion in a rock foundation 13.2 Probability of Forming a Roof 11.1.4 Assessing the likelihood of Intervention and Repair 12. Intervention and Repair --------------------------------------------12-1 12. 13.3 Flow into jointed bedrock 11-1 11-1 11-1 11-1 11-4 11-4 11-4 11-4 11-6 11-6 11-6 11-6 12 Probability of Detection.3 Detection 12.1 Overall Approach and Screening 13.4 Assessment for internal erosion in a rock foundation 13.3.3. and embankment into foundation.1 Approach 13-1 13-1 13-3 13-3 13-4 13-4 13-5 13-6 13-6 13-6 13-8 13-12 13-13 13-13 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 vii . 13.2 Estimation of the Probability of Breach by Gross Enlargement 13.1 Internal Erosion the Embankment 11.1 Overall Approach 11.4.3.2.1 Flow limitation by upstream zone 11.3.1 Internal Erosion Through the Embankment 11.2 11 Erosion into coarse grained soil foundation 10-22 Probability of Progression----------------------------------------------------------------------------11-1 11.2.2 Some Information on the Rate of Internal Erosion and Piping 12.2.3 Estimation Of The Probability Of Breach By Slope Instability 13.3.2.1 Screening for internal erosion in the embankment.1 Approach 13.2 Screening for internal erosion in a rock foundation.4.3.2 Internal Erosion through a Soil Foundation 11.4 Loss of Freeboard due to Slope Instability 13.2 Assessing the Probability of Not Detecting Internal Erosion 12.3 Assessment for internal erosion in the embankment.1 General Principles 12.2 Estimation of the probability of slope instability initiates for internal erosion in the embankment.2 Internal Erosion Through the Foundation 11.2 Flow into/out of open joint in conduits 11.3.1 Some General Principles 12. soil foundation.4. and from embankment into foundation 13. soil foundation.Contents 10.5 Calculation of Probability of Not Detecting and Not Intervening 12-1 12-2 12-5 12-5 12-6 12-10 12-12 13 Probability of Breach -----------------------------------------------------------------------------------13-1 13.3.3.

Contents 13.5.5.3 Probability of Loss of Freeboard due to Sinkhole Formation 14 15 13-17 13-17 13-17 13-17 References -------------------------------------------------------------------------------------------------14-1 List of Acronyms & Symbols ------------------------------------------------------------------------15-1 Appendices Appendix A Appendix B Appendix C Appendix D Appendix E Navigation Table for Internal Erosion Through the Embankment Navigation Table for Internal Erosion Through a Soil Foundation Navigation Table for Internal Erosion Through a Rock Foundation Navigation Table for Internal Erosion of the Embankment into or at the Foundation Guidance for Failure Paths Not Covered by the Unified Method A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 viii .1 Approach 13.2 Probability of Sinkhole Formation 13.5.5 Estimation of the Probability of Breach by Sinkhole Development 13.

............ 4-9 Table 4.....................................................................Screening of failure paths – Internal erosion through the foundation.................... 5-6 Table 5....................... 4-10 Table 4................Historical frequencies for cracking or poorly compacted zone in the embankment dam body. 5-9 ∑ ∑ ∑ ∑ A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 ix .Factors influencing the likelihood of cracking or hydraulic fracturing in the upper part of embankment dams-Differential settlement adjacent a cliff at the top of the embankment (IM2)..........3 ..........Screening of failure paths – Internal erosion due to concentrated leaks in poorly compacted or high permeability zones in the embankment ................................................Example of earthquake load partitions........... likely high and likely low equivalence table............ 3-18 Table 3.....Factors influencing the likelihood of cracking or hydraulic fracturing in the upper part of embankment dams ..Example of reservoir level partitions .......Screening of failure paths – Internal erosion due to concentrated leaks in transverse cracks in the middle and lower parts of embankment dams .....................Screening of initiating mechanisms – Internal erosion due to concentrated leaks in transverse cracks in the upper part of embankment dams ..................................Estimated historical frequencies of cracking.................................................................................. 3-24 Table 3...... 3-21 Table 3....................6 – Length Effects – Internal erosion in the foundation and into the foundation.......................................List of Tables & Figures Tables Table 3..................2 ...................... 5-3 Table 5.............Probability of cracking or hydraulic fracturing in embankment due to cross valley arching (Relative importance factor (RF)) x (Likelihood factor(LF)) .................................................................................................... 5-2 Table 5........7 – Factors influencing the likelihood of cracking or hydraulic fracturing in the upper part of embankment ...7 ......8 ......1 ........................4 .....................Screening of failure paths – Internal erosion in the embankment into or at the foundation ...................................................5 – Length Effects – Poorly compacted or high permeability zones in the embankment ..1 ..Factors influencing the likelihood of cracking or hydraulic fracturing in embankment due to cross valley arching (IM3)................................................. 4-19 Table 5. 5-4 Table 5.................3 – Length Effects – Transverse cracking in the upper part of the embankment ..................................................................... 5-4 Table 5................. 4-11 Table 4....cross section settlement resulting from poorly compacted shoulders (IM4)............5 .4 – Length Effects – Transverse cracking in the middle and lower parts of the embankment...................Cross Valley Differential Settlement (IM1) ..........................................................................4 .5 ........ 3-14 Table 3...................................1 ...............................2 . 4-3 Table 4...... 3-22 Table 3.......... 5-6 versus Table 5............................ hydraulic fracture or poorly compacted or high permeability zones in embankment dams. 4-2 Table 4...6 ..................... .......................... 3-17 Table 3...........................6 ..............Probability of a cracking or hydraulic fracture in the upper part of embankment-Differential (Relative settlement adjacent a cliff at the top of the embankment versus importance factor (RF)) x (Likelihood factor(LF)) .......... 5-8 Table 5...Probability of a transverse crack or hydraulic fracture resultant on cross section settlement (Relative importance factor resulting from poorly compacted shoulders versus (RF)) x (Likelihood factor (LF))...................7 – Best estimate.Probability of a cracking or hydraulic fracture in the upper part of embankment dams-Cross (Relative importance factor (RF) x Valley Differential Settlement versus (Likelihood factor(LF)) ............................................2 ......................................... 4-7 Table 4.3 ............. 3-25 Table 4...........................

................... 5-16 Table 5...... 5-23 Table 5............................. 5-20 Table 5.........................................................12 .Probability of a crack or hydraulic fracture due to differential settlement in the foundation (Relative importance factor (RF)) x (Likelihood factor (LF)) ....................................................... 5-28 Table 5........................ 5-28 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 ∑ ∑ ∑ ∑ ∑ ∑ ∑ x ......11 – Factors influencing the likelihood of cracking in the upper part of embankment damscracking in the crest due to desiccation by drying (IM7) ........................................Maximum likely width of cracking at the dam crest versus (Relative importance factor) x (Likelihood factor) for cracking in the upper part of the dam .......... 5-27 Table 5......................................................................................( Depths in feet and meters) ..............16 ........9 – Factors influencing the likelihood of cracking or hydraulic fracturing in the upper part of embankment dams...........Probability of a transverse crack or hydraulic fracture in the middle and lower parts of embankment dams due to small scale irregularities in the foundation profile under the core versus (Relative importance factor (RF)) x (Likelihood factor (LF)) ..............................................26 – Examples of estimated maximum depths below the dam crest and widths of cracks formed by potential hydraulic fracture for cracking in the upper part of the dam ................Settlement multiplication factors versus observed settlements............ 5-11 Table 5......... 5-24 Table 5.......List of Tables & Figures Table 5. Maximum likely depth of desiccation cracking for a gravel surface layer with no road pavement cover based on climate..................15 ... 5-19 Table 5.............10 .......... 5-21 Table 5........................................25 .....22 .settlement resulting from arching of the core onto the shoulders (IM10) ...............Factors influencing the likelihood of cracking on seasonal shutdown layers during construction and staged construction surfaces due to desiccation by drying (IM8) ......................19 ......Probability of a transverse crack or hydraulic fracture in the middle and lower parts of (Relative embankment dams due to cross valley differential settlements versus importance factor (RF)) x (Likelihood factor (LF)) .......................13 – Screening Tool................................................................................Factors influencing the likelihood of cracking or hydraulic fracture in the middle and lower parts of embankment dams-cross valley differential settlements (IM9)................................23 ..............18 – Factors influencing the likelihood of cracking or hydraulic fracturing in the middle and lower parts of embankments .......24 ..............settlement resulting from arching of the core onto the shoulders of the embankment versus (Relative importance factor (RF)) x (Likelihood factor (LF)) .............17 ........Probability of a transverse crack.................14 .......settlement resulting from differential settlements in soil in the foundation (IM5) ................ cracking in the crest due to desiccation by drying versus (Relative importance factor (RF)) x (Likelihood factor(LF)) ..... 5-14 Table 5...........Likely crack width at the depth shown versus maximum crack width at the dam crest determined from Table 5........ 5-15 Table 5......... 5-21 Table 5..21 . 5-17 Table 5...................Probability of a transverse crack or hydraulic fracture in the middle and lower parts of embankment dams.............Cracking observation factors (applies to upper embankment only).24 for cracking in the upper part of the dam....................... 5-18 Table 5........... 5-18 Table 5..20 – Factors influencing the likelihood of cracking or hydraulic fracturing in the middle and lower parts of embankment dams due to small scale irregularities in the foundation profile under the core (IM12) .................. 5-12 versus Table 5....Probability of a transverse crack on seasonal shutdown layers during construction and staged construction surfaces due to desiccation by drying versus (Relative importance factor (RF)) x (Likelihood factor(LF)) .... 5-14 Table 5....................................................

......................... 6-11 Table 6..............................................................30 .............. 5-38 Table 5............ 5-35 Table 5........... 6-10 Table 6..............28 ......... or SM with >30% fines soil types............... 6-13 Table 6.... 6-12 Table 6...........32 . 5-31 Table 5.......................... or CL-ML soil types....................................................... 2003) .... middle and lower parts of the embankment) ........................ 2004) .............5 ..............................................12 .... 5-39 Table 6......................33 .............................................35.34 ...3 .............................................Estimation of probability of initiation in a crack for CL-CH or CH with LL<65% soil types... 5-33 Table 5........... 6-3 Table 6..27 Representative erosion rate index (IHET) versus soil classification for non dispersive soils based on Wan and Fell (2002.......... 5-30 Table 5...........Probability of cracking or poorly compacted zones in the crest due to freezing versus (Relative importance factor (RF)) x (Likelihood factor (LF)) ......................... 6-14 Table 6........... 6-8 versus Table 6........................11 – Factors influencing the likelihood of poorly compacted or high permeability zones along outside of a conduit................37 – Initial Shear Stress assumed for Tables 5......10 ...9 ..........29 ........ versus crack width and flow gradient ................................. 5-36 Table 5..................... CL-CH).....Maximum likely width of cracking in the dam versus (Relative importance factor) x (Likelihood factor) for cracking in the middle and lower parts of the dam........................... 6-11 Table 6..38 – Damage classification system (Pells and Fell....1 – Factors influencing the likelihood of poorly compacted or high permeability zones in the embankment-cohesive soils (IM14) ..........................................Probability of a high permeability layer on seasonal shutdown layers during construction (Relative importance and staged construction surfaces due to freezing versus factor (RF)) x (Likelihood factor(LF)) ....Estimation of probability of initiation in a crack for ML or SM with <30% fines soil types5-33 Table 5......Factors influencing the likelihood of high permeability layer on seasonal shutdown layers during construction and staged construction surfaces due to freezing (IM17) ...........8– Maximum likely depth of freezing induced flaws based on climate....7 ...............29 to 5....39 – Estimation of the probability of transverse cracking from the damage class ....... 5-34 Table 5.......................................2 – Factors influencing the likelihood of poorly compacted or high permeability zones in the embankment-non cohesive soils (IM14) ................................. 6-7 Table 6................................................................................Estimation of probability of initiation in a crack for CH with LL>65% soil types ........List of Tables & Figures Table 5...................................................Estimation of probability of initiation in a crack for dispersive soils (CL......................Probability of a poorly compacted or high permeability layer in the embankment (Relative importance factor (RF)) x (Likelihood factor (LF)) (applies to versus upper..35 ........... 5-34 Table 5...................Estimation of probability of initiation in a crack for SC with >40% fines.Estimation of probability of initiation in a crack for CL or MH soil types...............Probability of a poorly compacted or high permeability layer on the core foundation contact (Relative importance factor (RF)) x (Likelihood factor (LF)) ........................................... CH............Probability of a poorly compacted or high permeability zone associated with a conduit (Relative importance factor (RF)) x (Likelihood factor(LF)) ............ (IM18) ........................4 – Factors influencing the likelihood of poorly compacted or high permeability zones on the core-foundation/abutment contact (IM15). 5-35 Table 5... 6-6 Table 6......31 .............................6 – Factors influencing the likelihood of cracking and poorly compacted zones in the upper part of embankment dams due to freezing (IM16) ..........36 – Estimated hydraulic shear stress (N/m2) from water flowing in an open crack.......Estimation of probability of initiation in a crack for SC with <40% fines................................ 6-15 versus ∑ ∑ ∑ ∑ ∑ ∑ A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 xi ........................................ 2002............................. 5-37 Table 5............................. 6-5 Table 6........... 5-36 Table 5...................

.......2 – Estimation of the probability of heave from embankment geometry and the foundation permeability ratio kh/ kv for situations where there is no confining layer in the foundation..........Time to develop seepage gradient in cohesionless soils ..............22 ...........Probability of initiation of erosion into a conduit versus ∑ (Relative importance factor (RF)) x (Likelihood factor(LF))................ 6-20 Table 6.............................................Factors influencing the likelihood of cracking or hydraulic fracturing due to differential settlement adjacent a spillway or abutment wall (IM22)............................... 6-34 Table 6.... 7-5 Table 7...........................Probability of the development of piping along the conduit given erosion initiates into the (Relative importance factor (RF)) x (Likelihood factor(LF)) from conduit versus Table 6............................... 6-27 Table 6.11 ..............16 ......................................... 7-4 Table 7.....Estimation of the probability of initiation and progression of backward erosion P IP in cohesionless soils and soils with PI ≤ 7 for uncompacted layers ...................................................... 6-38 Table 7....................... 6-16 Table 6...............27 .... 6-23 Table 6..Probability of a high permeability zone associated with a spillway or abutment wall (Relative importance factor (RF)) x (Likelihood factor(LF)) ............... 6-18 Table 6............................................................13 – Factors influencing the likelihood of initiation of erosion into a non-pressurized conduit when internal condition is known.... 6-32 Table 6...30 ...................14 – Factors influencing the likelihood of initiation of erosion into a non-pressurized conduit when the internal condition is not known (IM19) ...........17 – Factors influencing the likelihood of a poorly compacted or high permeability zone associated with a spillway or abutment wall (IM20).............................. 6-23 Table 6..........................................29 ...............15 ..Probability of cracking or hydraulic fracture due to differential settlement adjacent a (Relative importance factor (RF)) x (Likelihood spillway or abutment wall versus factor(LF)) .Width of frost induced flaw versus (RFxLF)............................................................26 ................3 – Estimation of the probability of heave from embankment geometry and the foundation permeability ratio kh/ kv for situations where there is a confining layer in the foundation7-5 Table 7............................................Probability of a gap or crack associated with a spillway or abutment wall (Relative importance factor (RF)) x (Likelihood factor(LF)) ........... 6-21 versus Table 6............. ........................................................................................ 6-21 Table 6...........................20 ......... 6-36 Table 6.18 .................4 – Estimation of the probability of initiation and progression of backward erosion in the foundation given heave is predicted (P IH ) ................................List of Tables & Figures Table 6.................... 6-20 versus Table 6.............................Seepage observation factors......................25 .................................................Seepage gradients at which suffusion may occur ....Amount of collapse settlement which may occur on saturation versus compaction properties ............... 7-9 ∑ ∑ ∑ ∑ A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 xii ........................................................................................................... 6-30 Table 6....................................23 ......................24 .................................................21 ...................................................................... 6-18 Table 6.............Estimation of the probability of initiation and progression of backward erosion P IP in cohesionless soils and soils with PI ≤ 7 for well compacted layers .1 – Estimation of the probability of heave or reaching the critical gradient (P H ) from the calculated factor of safety against heave.19 – Factors influencing the likelihood of a crack or gap adjacent to a spillway or abutment wall (IM21)..........................28 ......................................Factors to be considered in assessing seepage gradients on wrap-around .. 6-32 Table 6.................................................... 6-24 Table 6................... 6-17 Table 6...........................................

................6 ......................................Probability of a continuous in filled or open valley bulge feature in rock in the foundation (Relative importance factor (RF) x (Likelihood beneath the embankment versus factor (LF)) .....................2 ........Probability of grouting not being effective for continuous open defects or solution features (Relative importance factor (RF) x (Likelihood factor (LF))................................................................................................................Factors influencing the likelihood of a continuous in filled or open geological feature in the rock in the foundation beneath the embankment .............Factors influencing the likelihood of a continuous in filled or open valley bulge or rebound feature in the rock in the foundation beneath the embankment .............13 ........................Probability of a continuous open or in filled features in the rock foundation beneath the (Relative importance factor (RF) x (Likelihood factor (LF))8-22 embankment versus Table 8.Factors influencing the likelihood of a continuous in filled or open defects in the rock in the foundation beneath the embankment ..... 8-20 Table 8........... 7-11 Table 8.....Factors influencing the likelihood of grouting not being effective for continuous open defects and solution features.....................4 .............Factors influencing the likelihood of a continuous in filled or open solution feature in the rock in the foundation beneath the embankment ..........................Probability of a continuous in filled or open solution features in rock in the foundation (Relative importance factor (RF) x (Likelihood beneath the embankment versus factor (LF)) ...................................................Factors influencing the likelihood of a continuous in filled or open solution feature in the rock in the foundation beneath the embankment ...............8 ............... 8-14 Table 8...............19 ...................3 ...............5 ...............................................Factors influencing the likelihood of a continuous in filled or open defect in the rock in the foundation beneath the embankment ............... 8-26 Table 8... 8-27 Table 8........................... 8-21 Table 8...Factors influencing the likelihood of a cut-off in the foundation not being effective for continuous open defects and solution features ..................................... 8-17 Table 8...............18 ......................... 8-6 Table 8.................Factors influencing the likelihood of a continuous in filled or open valley bulge or rebound feature in the rock in the foundation beneath the embankment ............. 8-12 Table 8........................10 ............ 8-9 Table 8..................... 8-19 Table 8.............. 8-28 versus Table 8..................................Probability of a continuous in filled or open valley bulge or rebound feature in rock in the (Relative importance factor (RF) x foundation beneath the embankment versus (Likelihood factor (LF)) ......Factors influencing likelihood for topography...............................12 ................... 8-7 Table 8....... 8-12 Table 8......5 – Estimation of the probability of initiation of backward erosion for cases where heave is not predicted (P INH ) ............... 8-29 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 ∑ ∑ ∑ ∑ ∑ ∑ ∑ ∑ xiii ........................................................................................................... 8-7 embankment versus Table 8..............................9 .......17 .......... 8-18 Table 8..........................................1 ....................... 8-15 Table 8.14 ....................16 – Probability of Defects or Solution Features being open or in filled.............Probability of a continuous in filled or open feature in the rock foundation beneath the (Relative importance factor (RF) x (Likelihood factor (LF)) .........Probability of a continuous in filled or open features in rock in the foundation beneath the (Relative importance factor (RF) x (Likelihood factor (LF))8-10 embankment versus Table 8...............15 .........................................................11 ........................Probability of a continuous in filled or open solution features in rock in the foundation beneath the embankment versus (Relative importance factor (RF) x (Likelihood factor (LF)) ..................7 ....................................List of Tables & Figures Table 7...............

................Weighting factors for assessing probabilities of open or in filled defects and solution features............... 10-16 Table 10. US Corps of Engineers 1994)............................................... 8-31 Table 8.......2 – Probability for crack filling action not stopping pipe enlargement – internal erosion through the embankment ....... .................................... 12-3 Table 12.... 9-3 Table 9........3 ...........................Probability of treatment of the cutoff foundation not preventing contact of the core with open defects or solution features ...... for breach by gross enlargement................Probability of hydraulic fracturing in cut off trench due to arching versus (Relative importance factor (RF)) x (Likelihood factor(LF)) ................ 10-3 Table 10..........................1 – A method for the approximation estimation of the time for progression of piping and development of a breach.......... 2003)......................Factors influencing the likelihood of hydraulic fracturing within the cutoff trench due to arching ....................................................5 – Factors influencing the likelihood of not observing a concentrated leak (Pnol)... 10-12 Table 10............................................................................................... 10-14 Table 10...... 10-15 Table 10.............4 – Qualitative terms for times of development of internal erosion......................Rate of Erosion of the core or soil in the foundation ........ 12-8 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 ∑ ∑ xiv ..........9 Example of Estimating Probabilities for No........ ..............................4 .........................................6 – No erosion boundary for the assessment of filters of existing dams (after Foster and Fell 2001)........................5 – Susceptibility of filter/transition zones to segregation versus weighted score (Relative importance factor (RF)) x (Likelihood factor(LF)) ......... 11-7 Table 12................. 2003).... 10-18 Table 11... 10-9 Table 10............ cut-off wall or a concrete element in the erosion path............21 Probability of erosion initiating based on performance data................... ................................. and slope instability linked to development of a pipe (Fell et al 2001.................. 10-10 Table 10...................................................... 12-4 Table 12.................2 .......... 12-4 Table 12...........................1 .............................. 10-11 Table 10.................. 9-11 Table 9..............................10 . 11-3 Table 11........Likelihood for Filters with Excessive Fines Holding a Crack ...............7 – Excessive and Continuing erosion criteria (Foster 1999............20 ....... 10-11 Table 10..............11 – Aid to judgement for estimation of probability for continuation for open defects/joints/cracks................................................................... 8-34 Table 9.......13 – Probability of by-passing the foundation filter for piping through the foundation or piping from the embankment into the foundation ..........2 .........................................................................................................4............................. 9-4 Table 9............ Some....................... 2001)..............................4 – Gradation Limits to Prevent Segregation (USDA SCS 1994....... Excessive and Continuing Erosion for the example shown in Figure 10....3 – Influence of the material in the downstream zone of the embankment on the likely time for development of a breach............... piping and breach (Fell et al 2001......................... ............... 9-11 Table 10..............................1 – Probability of a soil being able to support a roof to an erosion pipe.........................3 .......... USBR 1987..Continuing Erosion criteria for erosion into an open defect.........................................List of Tables & Figures Table 8........................................ 12-4 Table 12.............1 Conditional Probability Ranges for Continuation (Scenario 2) .....Probability of a cut-off not being effective for continuous open defects or solution features (Relative importance factor (RF) x (Likelihood factor (LF)).......................................2 – Probability of a continuous pathway for erosion into soil foundation (IM29)........Potential for Segregation of Filtering Materials...................................................................................................................................................... Foster and Fell 1999................... 10-10 Table 10......................... 10-8 Table 10... 10-15 Table 10.............................. 11-5 Table 11......... 8-30 versus Table 8................................3 – Probability that flow in the developing pipe will not be restricted by an upstream zone..................8 – Aid to judgement for estimation of probability for Continuing Erosion (PCE) when the actual filter grading is finer than the Continuing Erosion Boundary.......12 – Probability of continuation for erosion into toe drains..................................22 ..........

.....Estimation of the probability of breach by slope instability – slide initiates (Psi-i) for (Relative importance factor (RF)) x internal erosion in rock foundations versus (Likelihood factor (LF)) ..................... 13-12 Table 13................................. 12-8 Table 12........... 13-5 Table 13................................11 .............Estimation of the probability of breach by sloughing (earthfill) versus (Relative importance factor (RF)) x (Likelihood factor (LF)) for internal erosion in rock foundation.........8 – Assessment of the probability that given the concentrated leak is detected.................................................................... and from embankment into foundation......... 13-14 Table 13..................................... in soil foundations..........7 – Assessment of size of leak in defect or solution feature in a rock foundation relative to discharge capacity of foundation drains and downstream shell ..13 ....................15 – Factors which influence likelihood of breaching by unravelling – dams with a rockfill downstream zone .Estimation of the probability of breach by slope instability – slide initiates for internal erosion in the embankment............................. and from embankment into foundation ...........................Estimation of the probability of breach by loss of freeboard versus (Relative importance factor (RF)) x (Likelihood factor(LF)) ..................... 12-9 Table 12.............................................................................................5 ...................................................................................................... 13-11 Table 13................ 13-15 Table 13.......................7 – Probability that given the leak is observable it is not detected given the time between the first appearance of the concentrated leak................. and the frequency of inspections and/or reading of monitoring instruments................... in soil foundations....List of Tables & Figures Table 12....................... 13-2 Table 13..........................2 – Probability of breach by gross enlargement of the pipe – Ability to support a pipe (Screening)....12 – Factors which influence likelihood of breaching by unravelling – dams with an earthfill downstream zone .............................................................. and from embankment into foundation ...................................................... 13-7 Table 13......3 – Probability of breach by gross enlargement of the pipe ................ and from embankment into foundation (Relative importance factor (RF)) x (Likelihood factor(LF)) ...6 – Probability of seepage exits from defects or solution features in a rock foundation into the downstream shell (PS)..........10 – Factors which influence likelihood of breaching by instability of the downstream slope – loss of freeboard .............9 ..................................................4 – Factors which influence likelihood of breaching by instability of the downstream slope – Slide Initiates for internal erosion in the embankment..... 13-12 Table 13..................................... 13-8 versus Table 13...... ....................... in soil foundations............................................................................................1 – Screening of breach mechanisms for internal erosion through the embankment...6 ........Estimation of the probability of breach by sloughing (earthfill) versus (Relative importance factor (RF)) x (Likelihood factor(LF)) for internal erosion in the embankment... 13-9 Table 13...................Probability of not observing a concentrated leak (P nol ) versus factor (RF)) x (Likelihood factor(LF)) for internal erosion in an embankment ............................................................... 12-11 Table 13.................................................... 13-3 Table 13......8 – Factors which influence likelihood of breaching by instability of the downstream slope –Slide Initiates for internal erosion in rock foundation ...........14 .... internal erosion in soil foundations.................................. 13-11 Table 13.... 13-14 Table 13.......... 13-15 ∑ (Relative importance ∑ ∑ ∑ ∑ ∑ A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 xv ................ intervention and repair is not successful (P ui ) .... 13-10 Table 13.......

......... and from embankment into foundation .................................List of Tables & Figures Table 13...1 ..................................................... 5-10 Figure 5..................................Longitudinal section through staged embankment .Flowchart for Internal Erosion of the Embankment into or at the Foundation............................2 .......................Cracking or hydraulic fracture adjacent cliffs due to differential settlement of the embankment........................ 3-7 Figure 3................. 13-16 Table 13.....................2 .................... 13-17 Table 13................................................... 5-3 Figure 5...................................Gradation of Broadly Graded Soils with Poor Self-Filtering Characteristics (Sherard 1979)1-4 Figure 3................... 5-13 Figure 5.7 .......4 ...Flowchart for Internal Erosion through a Rock Foundation.........Flowchart for Internal Erosion in the Embankment .................................... Note that this mechanism only applies for Wb/Hw< 2.... 3-4 Figure 3..................... 13-18 ∑ (Relative ∑ ∑ Figures Figure 1..........4 ............... 1-4 Figure 1............ 5-32 Figure 5............................... 3-11 Figure 3... 3-12 Figure 3..................2 ....... 5-38 Figure 5............ 6-15 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 xvi ........................Flowchart for Internal Erosion through a Soil Foundation...........Sloping core dam (a) Definitions of terms......................................1 .................................................. 4-24 Figure 5...............5 – Typical scenarios which may lead to differential settlement in the foundation............5 .......................Dam zoning categories ..Definition of terms used to describe cross valley geometry ...........................................................................................Estimation of the probability of breach by unravelling (rockfill) versus (Relative importance factor (RF)) x (Likelihood factor (LF)) for internal erosion in rock foundation......................................1 – Example fragility curve....... ...............................................6 ...................................... 3-8 Figure 3...19 – Factors which influence likelihood of breaching by sinkhole development – loss of freeboard given sinkhole develops ......3 ................20 . 5-39 Figure 6... (b) Limit of what constitutes a sloping core dam.........................................1 ......................Estimation of the probability of breach by sinkhole development – loss of freeboard (Relative importance factor (RF)) x (Likelihood given sinkhole develops versus factor (LF)) ...................Incidence of transverse cracking versus seismic intensity and damage class contours for earthfill dams (Pells and Fell 2002............................ 3-13 Figure 4...................... 2003) .................................3 – Longitudinal profiles of the dam showing the definition of terms for cross valley arching....Examples of embankment crest details which may result in relatively high likelihood of internal erosion......................... in soil foundations.............................5............. 5-12 Figure 5. 5-5 Figure 5................. ......17 ...............8 .......7 Example of the estimation of crack width and flow gradient in the crack......Example of an embankment with significantly different probabilities of internal erosion above and below the top of the downstream berm ......9 ......................1 Example of poor detailing of seepage collars around a conduit (from FEMA 2005)....6 .... 13-16 Table 13.Soils types which are subject to internal instability and suffusion ................................. 5-7 Figure 5.......Typical embankment dam showing some key features associated with potential internal erosion failure paths.................... 13-18 Table 13....................................................18 – Probability of a sinkhole or crest settlement developing (Ps-f) ...................Incidence of transverse cracking versus seismic intensity and damage class contours for earthfill and rockfill dams (Pells and Fell 2002.......... 1-2 Figure 1...Estimation of the probability of breach by unravelling (rockfill) versus importance factor (RF)) x (Likelihood factor(LF)) for internal erosion in the embankment....................................... 2003)........ ...16 ..........................................3 ......... 3-6 Figure 3..........

10-12 Figure 10.................Seepage model C where foundation is overlain by low permeability clay (Maniam 2004).......................... 6-25 Figure 6.1 – Flow chart for estimating the probability... ....... 6-34 Figure 7......................... ........ 7-6 Figure 7... . 7-2 Figure 7........7 ..Example of an embankment where much of the seepage flow will be to an unfiltered exit.... 8-35 Figure 9..................... 8-24 Figure 8.......Example of plot showing filter/transition gradings compared to Filter Erosion Boundaries...4 ................. average and coarse gradings of the core material....... 6-22 Figure 6..............2 – Approximate method for estimating DF15 after washout of the erodible fraction from a suffusive soil or for soils susceptible to segregation.................................. needed for complete piping (initiation and progression for an unfiltered exit) versus uniformity coefficient of soil (Schmertmann 2000)....................Contours of the probability of internal instability for sand-gravel soils with less than 10% non-plastic fines passing 0.................. ............4 ..............................3 – Criteria for Excessive Erosion Boundary................................ 8-28 Figure 8................... Plasticity Index ≤ 12........ .... average and coarse) for the assessment of filter compatibility..2 ................................ 10-21 Figure 12......List of Tables & Figures Figure 6................. 7-8 Figure 8............. 6-33 Figure 6...............................5 – Computation of probability of a continuous open defect or solution feature below the embankment....5 ..............................................................................075 mm (Wan and Fell 2004)... 10-7 Figure 10.....................2 ......................................................4 – Potential mechanisms for erosion of infill within a defect or solution feature..................6 ......1 ..............Seepage model D used to represent the cracked downstream clay (Maniam 2004)............3 – Wrap around details for connection of embankment dam to concrete gravity dam........................ 8-30 Figure 8...................Cross section of an embankment and dam foundation showing seepage flow net and definition of terms .... ............................................. 9-10 Figure 10.......10-21 Figure 10...Seepage model B of embankment dam and foundation (Maniam 2004)............. ....... 12-12 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 xvii ........... Evaluate the filter erosion boundaries for the representative fine............................. ................5 ............3 ................... and (c) for massive rocks (e...2 – Assumed distribution of defect depths for defects related to stress relief effects in the valley sides – (b) Jointed rocks such as thinly bedded sandstones...g...........Situations where a gap may form between the dam fill and spillway wall (a) Steep foundation adjacent spillway wall.......................1 – Sub-event tree for calculating the probability of not intervening.................Example of an embankment where there is an unfiltered exit due to day lighting of the foundation sand layer downstream of the dam.......... 10-8 Figure 10................. 10-13 Figure 10....................... 7-3 Figure 7............Seepage model A of embankment dam and foundation (Maniam 2004).................... ........... Figures from Fell et al 2004)....... (b) Change in slope of the retaining wall (Fell et al 2004)............2 ......... ......... ipmt.............6 ........ some granite)............... ...Maximum point gradient..... 6-31 Figure 6...........................................................3 – Definition of Δhp and hp..............................1 – Example of the selection of representative grading curves (fine..............Section through embankment and foundation showing definition of terms to estimate the average gradient in the foundation sand................................... (Wan and Fell 2004).............................. granite or basalt.. ...... 10-20 Figure 10...........6 ...............................Definition of terms for arching across a cut-off trench.......Contours of the probability of internal instability for silt-sand-gravel soils and clay-siltsand-gravel soils of limited clay content and plasticity................. depth and spatial distribution of continuous open defects and solution features in rock foundations........An example of a situation where there is no continuous layer of cohesionless soil in the foundation and backward erosion cannot occur..5 – Examples of scenarios of fully penetrating and partially penetrating foundation filter drains............. ...............7 .................................. 7-6 Figure 7. width............................. ...............1 – Examples of foundations with continuous open defects of varying width ......... 7-6 Figure 7............................. 9-7 Figure 9... 8-2 Figure 8.....................................4 ........ 7-6 Figure 7.

Piping Toolbox) for embankment dams.1 General The United States Department of the Interior. Sydney. and to guide the analyst through a detailed. Australia and used in practice in Australia by URS and other consultants. the probability is linked to embankment type. The key goals of the training would be to provide an understanding of all features and components of the methodology. real life example use of the methodology. Bureau of Reclamation (Reclamation) has developed a methodology for performing quantitative risk analysis for piping and internal erosion failure modes (i. URS were commissioned by Reclamation and USACE to work with them to do this further development.e.Introduction SECTION 1 1 Introduction 1. This report presents the results of the study and the methods which were developed for estimating the probability of internal erosion and piping in embankment dams and their foundations.2 Training The Development Team strongly recommends that all risk analysts that will use this toolbox be trained in its use.3 Terminology to Describe Embankment Types In several of the tables provided to assist in assessing probabilities. The terms shown in Figure 1. URS and Professor Fell it was recognised that some further development of the methods in the Toolbox was necessary to facilitate uniformity of application by risk analysis teams in Reclamation and USACE. In discussions held in Denver in July 2005 between Reclamation. 1. Recently. under a cooperative agreement. The complexity of the issues and importance of the end product demands that all analysts fully understand the methodology. USACE.. These methods are closely related to those developed at the University of New South Wales. 1.1 have been adopted. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 1-1 . to outline for the analyst the supporting information and background that was used in the development of the methodology. the United States Army Corps of Engineers (USACE) has assisted in a revision of this toolbox.

Zoned earthfill 7. Central core earth and rockfill 9. Hydraulic fill Figure 1.Dam zoning categories A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 1-2 . Earthfill with corewall Rock toe Max 0.Introduction SECTION 1 core downstream zone of rockfill Puddle core 0. Concrete face earthfill 10. Earthfill with rock toe 6. Rockfill with corewall core downstream zone of sand/gravel concrete facing rockfill hydraulic fill core 3. Puddle core earthfill Embankment filter and/or Foundation filter rockfill core rockfill concrete corewall earthfill 1.2H concrete facing earthfill concrete corewall rockfill 2. Concrete face rockfill 11. Homogeneous earthfill 4. Zoned earth and rockfill 8. Earthfill with filter 5.1 .

A continuous crack. or it may occur in a continuous permeable zone containing coarse and/or poorly compacted materials which form an interconnecting voids system. Flaw. Occurs when soil particles within an embankment dam or its foundation. such as the ground surface downstream of a soil foundation or the downstream face of a homogeneous embankment or a coarse rockfill zone immediately downstream from the fine grained core. The concentration of flow causes erosion (sometimes called scour) of the walls of the crack or interconnected voids. Internal erosion. The detached particles are carried away by the seepage flow and the process gradually works its way towards the upstream side of the embankment or its foundation until a continuous pipe is formed. Internal erosion can initiate by concentrated leak erosion. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 1-3 . freezing and thawing. Coarse graded and gap graded soils. Erosion in a concentrated leak may occur in a crack in an embankment or its foundation. are susceptible to suffusion. and by hydraulic fracture. are carried downstream by seepage flow. Soils which are susceptible to suffusion are internally unstable. such as those shown schematically in Figure 1. and results in the formation of a continuous tunnel called a ‘pipe’ between the upstream and the downstream side of the embankment or its foundation. leaving behind an intact soil skeleton formed by the coarser particles. In these soils the volume of fines is less than the volume of voids between the coarse particles.Introduction SECTION 1 1. Suffusion and internal instability. backward erosion. Piping. suffusion and soil contact erosion. The fine particles are removed through the constrictions between the larger particles by seepage flow. Piping is the form of internal erosion which initiates by backward erosion.2.4 Terminology The Unified Method uses the following terminology which should be adopted. Suffusion is a form of internal erosion which involves selective erosion of fine particles from the matrix of coarser particles (coarse particles are not floating in the fine particles). desiccation. Backward erosion involves the detachment of soils particles when the seepage exits to a free unfiltered surface. or erosion in a crack or high permeability zone. high permeability or poorly compacted layer in which a concentrated leak may form. caused by differential settlement. Backward erosion. Concentrated leak erosion.

The figure is not meant to define the boundary of such soils. which in turn prevent erosion of the fine particles.01 0. only examples.Introduction SECTION 1 100 90 80 70 % Passing 60 50 40 30 20 10 0 0. and very broadly graded soils such as those which fall into the grading envelope shown in Figure 1. In soils which self-filter.Soils types which are subject to internal instability and suffusion Self-filtering. and the dams from which the soils were taken had all exhibited signs of internal erosion. the coarse particles prevent the internal erosion of the medium particles. The soils had particle size distributions which plotted nearly as a straight line. Figure 1. were typically of glacial origin.Gradation of Broadly Graded Soils with Poor Self-Filtering Characteristics (Sherard 1979) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 1-4 .001 0.3 .1 1 10 100 1000 Particle size (mm) COA RSELY GRA DED SOIL WITH A FLA T TA IL OF FINES GA P GRA DED SOIL CLAY TO SILT SAND GRAVEL Figure 1. The soils have a volume of fine particles greater than the volume of voids between the coarse sand and gravel fraction and the coarser particles are “floating” in the finer particles.2 .3 (Sherard 1979). Soils which potentially will not self-filter include those which are susceptible to suffusion (internally unstable).

g. where hydraulic shear stresses within the eroding soil may or may not lead to the enlargement of the pipe. whether upstream zones may control the erosion process by flow limitation and whether a pipe will extend through the low permeability zones of the embankment. • • • Annual Exceedance Probability (AEP): The estimated probability that an event of specified magnitude will be exceeded in any year. Some Erosion. The increase in leakage flows is so small that it is unlikely to be detectable. Progression is the third phase of internal erosion. The main issues are the likelihood of and rate of pipe enlargement and whether the pipe will collapse. Overtopping (e. The filtering material stops erosion with no or very little erosion of the material it is protecting. assumptions and information. Foster and Fell (1999. Breach. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 1-5 . It may occur by one of the following four phenomena (listed below in order of their observed frequency of occurrence). Likelihood: Conditional probability of an outcome given a set of data. Also used as a qualitative description of probability and frequency. but the flows are self healing. The filter material allows erosion from the material it is protecting. Unravelling of the downstream face. 2001) and Foster (1999) define four levels of severity of continuation from ‘no erosion’ to ‘continuing erosion. due to settlement of the crest from suffusion and/or due to the formation of a sinkhole from a pipe in the embankment). and in the process permits large increases in leakage flow. Continuation is the phase where the relationship of the particle size distribution between the base (core) material and the filter controls whether or not erosion will continue. Breach is the final phase of internal erosion. The filtering material is too coarse to stop erosion of the material it is protecting and continuing erosion is permitted. but it eventually seals up and stops erosion. • Gross enlargement of the pipe (which may include the development of a sinkhole from the pipe to the crest of the embankment). The extent of erosion is sufficient to cause sinkholes on the crest and erosion tunnels through the core. Increases of pore pressure and seepage occur. Slope instability of the downstream slope. Unlimited erosion and leakage flows are likely. Frequency: A measure of likelihood expressed as the number of occurrences of an event in a given time or in a given number of trials (see also likelihood and probability).Introduction SECTION 1 Continuation. Excessive Erosion. Progression. No Erosion. The filtering material initially allows erosion from the soil it is protecting. Continuing Erosion.

obtained by considering all available information honestly. Failure Path. judgement regarding an evaluation. Subjective probability (degree of belief) – Quantified measure of belief.Introduction SECTION 1 Probability: A measure of the degree of certainty. or the quality and quantity of information. It is an estimate of the likelihood of the magnitude of the uncertain quantity. and with a minimum of bias. and which may lead to an uncontrolled release of the reservoir. This measure has a value between zero (impossibility) and 1. or the likelihood of the occurrence of the uncertain future event. Subjective probability is affected by the state of understanding of a process. It may change over time as the state of knowledge changes. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 1-6 .0 (certainty). fairly. There are two main interpretations: Statistical – frequency or fraction – The outcome of a repetitive experiment of some kind like flipping coins. Such a number is called an “objective” or relative frequentist probability because it exists in the real world and is in principle measurable by doing the experiment. or confidence in the likelihood of an outcome. such as a defect. It includes also the idea of population variability. judgement. A sequence of potential events starting from an initiating mechanism. flaw or seepage path in the dam or its foundation.

Generic event trees have been developed for each general failure mode the navigation tables in Appendices A to D. seismic and hydrological) as described in Sections 3. then develop a different event tree using the guidance given in Appendix E. • • • Internal erosion through the embankment. produce the Annual Probability of Failure (APF) and are the main topic of this document.3 as an aid). Internal erosion through the foundation. The first two components of this equation.4.Methodology SECTION 2 2 Methodology 2. Estimate the conditional probabilities for each node on the event tree.5 and 3.3 and 3.6. Risk = [Probability of the loading] × [Probability of adverse response given the loading] × [Adverse consequence given the failure]. Specific guidance is given for estimating the conditional probabilities for various Step 4: Step 5: A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 2-1 . Develop detailed descriptions and sketches of all realistic failure paths. fully documenting the rationale. Select the loading partitions and estimate loading probabilities for each of the load conditions (static.2 General Process The process for estimating the annual probability of failure by piping and internal erosion includes the following general steps: Step 1: Step 2: Review all information pertinent to piping and internal erosion (refer to Section 2. Screen those failure paths that are assessed to have negligible contribution to the annual probability of failure and document the reasons for their exclusion. when multiplied. and Internal erosion of the embankment into or at the foundation. 2.1 Introduction Risk is defined as the probability of a loss occurring in a given time period (annually) The equation for risk is. Guidance is given in Sections 3. Select the event tree and associated navigation table that best fits the failure path being considered. If the failure path can not be matched with one of the generic event trees provided. Step 3: Decompose each of the potential failure paths into event trees. considering each of the failure locations. Identify all potential failure paths associated with internal erosion and piping.

however. Additional site information is provided by documented performance history. and additional analyses may be warranted.3 Information Review The quality and credibility of a risk estimate will suffer unless the analyst is fully aware of all pertinent information about piping and internal erosion. and characteristics of the setting. and foundation material characteristics. for the project is indicated. and post-construction investigations need to be reviewed. laboratory testing. it may be less expensive to simply fix the structure for the failure mode of concern. Guidance is also given in Appendix E for estimating probabilities for situations which are not covered by the method. and seepage cut-off and control features. Use the annual probability of failure estimates in follow-on risk analysis and assessment. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 2-2 . Navigation tables are provided in Appendices A to D to assist the user to find the location of the guidance tables for each node of the event tree. Step 6: Step 7: Calculate the probability of failure by internal erosion and piping for each failure path and review for consistency between failure paths (Section 4.9). The site visit allows a better understanding of the layout of the dam and helps in the process of identifying potential failure paths. field investigations. performance history. To the maximum extent practicable. If significant risk. In some cases. existing data from design and construction records. such as embankment geometry. Typical information to review includes characteristics of the constructed project. materials. existing data supplemented by engineering judgment provide a sufficient basis for evaluation. such as site geology and stratigraphy. A site visit to the dam has been found to provide valuable input into the risk analysis process.Methodology SECTION 2 initiating mechanisms and failure locations in Sections 5 to 13. At preliminary stages of risk analysis. 2. construction methods. including consideration of uncertainty. zoning.

• Concentrated leak erosion. Reservoir Rises Initiation – Flaw exists (1) (2) Initiation – Erosion starts Continuation – Unfiltered or inadequately filtered exit exists (consider: no erosion/some erosion/excessive erosion/continuing erosion) Progression – Roof forms to support a pipe Progression – Upstream zone fails to fill crack Progression – Upstream zone fails to limit flows Intervention fails Dam breaches (consider all likely breach mechanisms) Consequences occur (1) A ‘flaw” is a continuous crack. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-1 . For Backward Erosion Piping (BEP) no flaw is required but a continuous zone of cohesionless soil in the embankment or foundation is required.1 Event Tree The following generic sequence of events has been developed for internal erosion failure modes. high permeability or poorly compacted zone in which a concentrated leak may form. erosion must start to initiate for internal erosion to develop.Failure Modes and Load Partitioning SECTION 3 3 Failure Modes and Load Partitioning 3. Initiation is the first phase and considers the existence of a flaw such as a continuous crack or poorly compacted layer in which a concentrated leak may form. If a flaw exists. Erosion can commence from the walls of a crack within the soil or within a poorly compacted layer. There are several processes by which erosion can initiate in the embankment or foundation as follows. (2) Generic event tree structures have been developed based on this sequence of events and these are presented in the Guidance Tables in Appendices A to D for each of the general failure modes.

Slope instability of the downstream slope. The main issues are whether the pipe will collapse and whether upstream zones may control the erosion process by flow limitation or crack filling. where hydraulic shear stresses within the eroding soil may or may not lead to the enlargement of the pipe. This is a form of internal erosion which involves selective erosion of fine particles from the matrix of coarser particles (coarse particles are not floating in the fine particles). Overtopping (e. Some Erosion. Unravelling of the downstream face. Erosion of the soil may occur where it is in contact with seepage passing through the foundation either through a coarse grained soil or open joints in rock. Backward erosion. Progression is the third phase of internal erosion. leaving behind an intact soil skeleton formed by the coarser particles.g. Dam Breaches is the final phase of internal erosion. Increases of pore pressure and seepage occur. It may occur by one of the following four phenomena (listed below in order of their observed frequency of occurrence). The detached particles are carried away by the seepage flow and the process gradually works its way towards the upstream side of the embankment or its foundation until a continuous pipe is formed. • • Continuation is the phase where the relationship of the particle size distribution between the base (core) material and the filter controls whether or not erosion will continue. The fine particles are removed through the constrictions between the larger particles by seepage flow. • Gross enlargement of the pipe (which may include the development of a sinkhole from the pipe to the crest of the embankment). due to settlement of the crest from suffusion and/or due to the formation of a sinkhole from a pipe in the embankment). 2001) and Foster (1999) define four levels of severity of continuation. Excessive Erosion and Continuing Erosion.Failure Modes and Load Partitioning SECTION 3 • Scour at the embankment –foundation contact. and this considers whether the internal erosion failure mechanism will be detected and whether intervention and repair will successfully stop the failure process. Backward erosion involves the detachment of soils particles when the seepage exits to a free unfiltered surface. Suffusion. • • • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-2 . Foster and Fell (1999. Intervention fails is the fourth phase of the event tree. No Erosion.

internal erosion is occurs solely within the embankment. separate event trees should be developed for the following cases. The general process involved in assessing the probability of internal erosion through the embankment is summarized in the flowchart shown in Figure 3.1.2.Failure Modes and Load Partitioning SECTION 3 3. Internal erosion through a high permeability zone in the embankment. Internal erosion through the foundation. Internal erosion through cracks in the middle/lower parts of the embankment. which are. and Internal erosion of the embankment into or at the foundation. such as an outlet works. • • • • • Internal erosion through cracks in the upper part of the embankment. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-3 . Internal erosion associated with a conduit/culvert penetrating through the embankment. • • • Internal erosion through the embankment. Where applicable for a dam.2 General Failure Modes Failure by internal erosion of embankment dams is categorized into three general failure modes. and Internal erosion along the outside of a spillway wall or other concrete structure. A more detailed description of the process and a generic event tree is given in the navigation tables given in Appendix A.1 Internal Erosion Through the Embankment In this failure mode. spillway or adjoining a concrete gravity structure. This includes internal erosion associated with through-penetrating structures. 3.

Flowchart for Internal Erosion in the Embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-4 . AVAILABILITY OF EQUIPMENT/MATERIALS) PROBABILITY OF BREACH GROSS ENLARGEMENT OF THE PIPE SLOPE INSTABILITY OF THE DOWNSTREAM SLOPE UNRAVELLING OR SLOUGHING OF THE DOWNSTREAM SLOPE SINKHOLE DEVELOPMENT OR CREST SETTLEMENT PROBABILITY OF FAILURE Figure 3. CLIFF/WALL. NO FILTER SCENARIO 2: DOWNSTREAM SHOULDER OF COHESIVE MATERIAL SCENARIO 3: FILTER/TRANSITION ZONE DOWNSTREAM OF CORE SCENARIO 4: EROSION INTO OPEN CRACK/JOINT SCENARIO 5: EROSION INTO TOE DRAIN PROBABILITY OF PROGRESSION WILL THE SOIL HOLD A ROOF? WILL CRACK FILLING ACTION STOP EROSION? WILL FLOW BE RESTRICTED BY AN UPSTREAM ZONE? PROBABILITY OF INTERVENTION FAILS WILL IT BE DETECTED (IS IT VISIBLE/DETECTABLE.COHESIVE SOIL LAYER PROBABILITY OF CONTINUATION SCENARIO 1: HOMOGENEOUS ZONING.) TIME TO FAILURE INTERVENTION ACTIONS (RELEASE. CROSS VALLEY ARCHING) (b) CROSS SECTION SETTLEMENT DUE TO POORLY COMPACTED SHELLS (c) DIFFERENTIAL SETTLEMENTS IN FOUNDATION (d) DIFFERENTIAL SETTLEMENT DUE TO EMBANKMENT STAGING (e) DESICCATION (AT CREST OR DURING CONSTRUCTION) (a) CROSS VALLEY DIFFERENTIAL SETTLEMENT (b) DIFFERENTIAL SETTLEMENT LEADING TO ARCHING OF CORE (c) DIFFERENTIAL SETTLEMENTS IN FOUNDATION (d) FOUNDATION CONTACT DUE TO SMALL SCALE IRREGULARITIES IN FOUNDATION PROFILE (e) DESICCATION DURING CONSTRUCTION (b) ON CORE-FOUNDATION CONTACT (c) ASSOCIATED WITH A CONDUIT (d) ASSOCIATED WITH A SPILLWAY WALL (e) WRAP AROUND DETAILS FOR CONNECTION TO CONCRETE GRAVITY DAM ADJUSTMENT FACTORS TO ACCOUNT FOR OBSERVATIONS AND SETTLEMENT ADJUSTMENT FACTORS TO ACCOUNT FOR OBSERVATIONS WILL EROSION INITIATE WITHIN THE CRACK? (a) (b) (c) (d) ESTIMATE CRACK WIDTH ESTIMATE INITIAL SHEAR STRESS TO INITIATE EROSION ESTIMATE CRACK WIDTH REQUIRED TO INITIATE EROSION COMPARE (a) AND (c) WILL EROSION EROSION INITIATE IN THE HIGH PERMEABILITY ZONE? .1 .Failure Modes and Load Partitioning SECTION 3 LOADING CONDITION PROBABILITY OF INITIATION OF EROSION WILL A TRANSVERSE CRACK OR HYDRAULIC FRACTURE DEVELOP IN THE DAM BODY? IS A POORLY COMPACTED OR HIGH PERMEABILITY ZONE PRESENT IN DAM? CRACKING MECHANISMS: UPPER PART: MIDDLE/LOWER PART: FAILURE PATHS: (a) WITHIN THE CORE (a) CROSS VALLEY DIFFERENTIAL SETTLEMENT (BENCHES. HOW OFTEN INSPECTED.BACKWARD EROSION .SUFFUSION .

Failure Modes and Load Partitioning SECTION 3 3.2.4.3 for a rock foundation. This is associated with internal erosion of an element within the dam foundation that is erodible. eroding the embankment material.2. and Figure 3. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-5 . 3.2. More detailed descriptions of the process are given in the navigation tables given in Appendix B for internal erosion in a soil foundation and Appendix C for internal erosion in a rock foundation. The general process involved in assessing the probability of internal erosion through a soil foundation is summarized in the flowchart shown in Figure 3. A more detailed description of the process is given in the navigation tables given in Appendix D. or ii) seepage in the foundation at the embankment contact. The general process involved in assessing the probability of internal erosion of the embankment into or at the foundation is summarized in the flowchart shown in Figure 3.3 Internal Erosion of the Embankment into or at the Foundation Internal erosion initiates at the contact between the embankment and foundation owing to i) seepage through the embankment eroding material into the foundation.2 Internal Erosion Through the Foundation This general failure mode involves internal erosion occurring solely within the foundation until the later stages of the breach process where the embankment starts to collapse.

WILL EROSION INITIATE THROUGH A CRACK IN THE SOIL? IS THERE A LAYER OF SOIL IN WHICH A CONTINUOUS CRACK OR INTERCONNECTED PATTERN OF CRACKS MAY EXIST ? IS THERE A CONTINUOUS LAYER OF COHESIONLESS SOIL ? PROBABILITY OF HEAVE PROBABILITY OF INITIATION AND PROGRESSION OF BACKWARD EROSION PIPING GIVEN HEAVE HAS OCCURRED PROBABILITY OF NO HEAVE PROBABILITY OF INITIATION AND PROGRESSION OF BACKWARD EROSION PIPING GIVEN HEAVE HAS NOT OCCURRED PROBABILITY OF INITIATION AND PROGRESSION OF BACKWARD EROSION PIPING OR. ASSESS IN TERMS OF FILTER EROSION CRITERIA IF UNFILTERED .2 .Failure Modes and Load Partitioning SECTION 3 LOADING CONDITION PROBABILITY OF INITIATION OF EROSION SCREENING CHECK ON SOIL CLASSIFICATION: IS THE FOUNDATION SOIL COHESIONLESS OR COHESIVE? IF SOIL IS COHESIONLESS.PROBABILITY = 1. WILL SUFFUSION INITIATE? IS THERE A CONTINUOUS LAYER OF COHESIONLESS SOIL ? ESTIMATE THE PROBABILITY OF SUFFUSION (a) IS SOIL POTENTIALLY INTERNALLY UNSTABLE? (b) PROBABILITY SOIL IS INTERNALLY UNSTABLE (C) PROBABILITY SUFFUSION WILL BEGIN WILL EROSION INITIATE WITHIN THE CRACK? (a) (b) (c) (d) ESTIMATE CRACK WIDTH ESTIMATE INITIAL SHEAR STRESS TO INITIATE ESTIMATE CRACK WIDTH REQUIRED TO INITIATE EROSION COMPARE (a) AND (c) PROBABILITY OF CONTINUATION PROBABILITY THAT THE EXIT WILL BE A FILTERED OR UNFILTERED EXIT IF FILTERED EXIT.Flowchart for Internal Erosion through a Soil Foundation A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-6 . WILL BACKWARD EROSION INITIATE? IF SOIL IS COHESIVE. AVAILABILITY OF EQUIPMENT/MATERIALS) PROBABILITY OF BREACH GROSS ENLARGEMENT OF THE PIPE SLOPE INSTABILITY OF THE DOWNSTREAM SLOPE UNRAVELLING OR SLOUGHING OF THE DOWNSTREAM SLOPE SINKHOLE DEVELOPMENT OR CREST SETTLEMENT PROBABILITY OF FAILURE Figure 3.) TIME TO FAILURE INTERVENTION ACTIONS (RELEASE. HOW OFTEN INSPECTED.0 PROBABILITY OF PROGRESSION WILL THE SOIL HOLD A ROOF? WILL CRACK FILLING ACTION STOP EROSION? WILL FLOW BE RESTRICTED BY AN UPSTREAM ZONE? PROBABILITY OF INTERVENTION FAILS WILL IT BE DETECTED (IS IT VISIBLE/DETECTABLE.

AVAILABILITY OF EQUIPMENT/MATERIALS) PROBABILITY OF BREACH GROSS ENLARGEMENT OF THE PIPE SLOPE INSTABILITY OF THE DOWNSTREAM SLOPE UNRAVELLING OR SLOUGHING OF THE DOWNSTREAM SLOPE SINKHOLE DEVELOPMENT OR CREST SETTLEMENT PROBABILITY OF FAILURE Figure 3.) TIME TO FAILURE INTERVENTION ACTIONS (RELEASE.Flowchart for Internal Erosion through a Rock Foundation A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-7 . HOW OFTEN INSPECTED.Failure Modes and Load Partitioning SECTION 3 LOADING CONDITION PROBABILITY OF INITIATION OF EROSION PROBABILITY OF ONE OR MORE CONTINUOUS OPEN OR IN FILLED DEFECTS (STRESS RELIEF VALLEY SIDES. IS GROUTING EFFECTIVE? FOR IN FILLED DEFECTS. SOLUTION FEATURES AND OTHER GEOLOGICAL FEATURES) IS THE DEFECT OPEN OF INFILLED? FOR OPEN DEFECTS.3 . GROUTING IS ASSUMED TO BE INEFFECTIVE? DOES EROSION OF THE IN FILL INITIATE? DOES EROSION OF THE IN FILL CONTINUE? COMBINE THE PROBABILITY ESTIMATES FOR OPEN DEFECTS AND IN FILLED DEFECTS THAT HAVE ERODED IDENTIFY THE FAILURE MODES AND BREACH MECHANISMS WHICH MAY DEVELOP FROM THE PRESENCE OF THESE FEATURES PROBABILITY OF CONTINUATION PROBABILITY THAT THE EXIT WILL BE A FILTERED OR UNFILTERED EXIT IF FILTERED EXIT. VALLEY BULGE.PROBABILITY = 1.0 PROBABILITY OF PROGRESSION WILL THE SOIL HOLD A ROOF? WILL CRACK FILLING ACTION STOP EROSION? WILL FLOW BE RESTRICTED BY AN UPSTREAM ZONE? PROBABILITY OF INTERVENTION FAILS WILL IT BE DETECTED (IS IT VISIBLE/DETECTABLE. ASSESS IN TERMS OF FILTER EROSION CRITERIA IF UNFILTERED .

4 .PROBABILITY = 1. AVAILABILITY OF EQUIPMENT/MATERIALS) PROBABILITY OF BREACH GROSS ENLARGEMENT OF THE PIPE SLOPE INSTABILITY OF THE DOWNSTREAM SLOPE UNRAVELLING OR SLOUGHING OF THE DOWNSTREAM SLOPE SINKHOLE DEVELOPMENT OR CREST SETTLEMENT PROBABILITY OF FAILURE Figure 3.Flowchart for Internal Erosion of the Embankment into or at the Foundation A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-8 .) TIME TO FAILURE INTERVENTION ACTIONS (RELEASE. ASSESS IN TERMS OF FILTER EROSION CRITERIA IF UNFILTERED .0 PROBABILITY OF PROGRESSION WILL THE SOIL HOLD A ROOF? WILL CRACK FILLING ACTION STOP EROSION? WILL FLOW BE RESTRICTED BY AN UPSTREAM ZONE? PROBABILITY OF INTERVENTION FAILS WILL IT BE DETECTED (IS IT VISIBLE/DETECTABLE.Failure Modes and Load Partitioning SECTION 3 LOADING CONDITION PROBABILITY OF INITIATION OF EROSION SCREENING CHECK FAILURE PATH: SOIL OR ROCK FOUNDATION? IF FOUNDATION IS ROCK. HOW OFTEN INSPECTED. WILL EROSION INTO OPEN ROCK JOINTS INITIATE? IF FOUNDATION IS SOIL WILL EROSION INTO COARSE GRAINED SOIL INITIATE? IS THERE A CONTINUOUS PATHWAY OF OPEN JOINTS IN THE ROCK IN THE BASE OR SIDES OF THE CORE TRENCH OR AT THE CORE-FOUNDATION CONTACT ? WILL BACKWARD EROSION PIPING INITIATE ? OR WILL SCOUR INITIATE AT THE CORE-FONDATION CONTACT ? OR WILL EROSION INITIATE IN A HYDRAULIC FRACTURE ACROSS THE CORE TRENCH ? IS THERE A CONTINUOUS PATHWAY OF COARSE GRAINED SOILS IN THE SOILS IN THE BASE OR SIDES OF THE CORE TRENCH OR AT THE CORE-FOUNDATION CONTACT ? WILL BACKWARD EROSION PIPING INITIATE ? OR WILL SCOUR INITIATE AT THE CORE-FONDATION CONTACT ? OR WILL EROSION INITIATE IN A HYDRAULIC FRACTURE ACROSS THE CORE TRENCH ? PROBABILITY OF CONTINUATION PROBABILITY THAT THE EXIT WILL BE A FILTERED OR UNFILTERED EXIT IF FILTERED EXIT.

there remains the potential that even with the “… efforts of even the most experienced engineers the most significant potential failure mode may occasionally be overlooked.” A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-9 . by assembling construction photographs and reports and inspecting the dam as part of the failure modes assessment. The development team has strived to address all failure modes that. However. Zoning of the embankment. • • • • • The general event tree structure as described in Section 3. An example is given in Section 3. The sequence of events and seepage pathways should be documented by annotating cross sections and longitudinal sections of the embankment and its foundation to help visualise the failure path.Failure Modes and Load Partitioning SECTION 3 3.3.4. as summarized in Figures 3.1 Identification of Failure Paths Overview A potential failure path is a sequence of events starting from an initiating mechanism. and which may lead to an uncontrolled release of the reservoir.1.3. The development of the failure paths should consider the following. It is also reduced by having the failure modes assessment done by a team which includes the engineer and geologist most familiar with the dam. flaw or seepage path in the dam or its foundation. occur in embankment dams. and facilitated by a person experienced in failure modes analysis. Potential initiating mechanisms for each of the failure locations. The risk analysis team should go through a discussion of all potential failure paths and develop a thorough understanding of the sequence of events and the potential location of seepage and erosion paths through the embankment and foundation.2. Foundation geology and stratigraphy. as pointed out by the late Ralph Peck in his paper “The Risk of the Oddball”. The possibility of overlooking potentially important failure modes is reduced by considering the particular details if the dam and it’s appurtenant structures. in their experience. such as a defect. and Filtered and unfiltered exit points of seepage. Procedures are also suggested to handle failure modes not well covered in this methodology. in the end.3 3.1 to 3. conduits through the embankment. including the configuration of internal filter and drainage measures. dam operating and surveillance staff. such as details of walls retaining the embankment.

There is no upstream zone of material to limit flows. or sinkhole development).e.. Seepage flows through this gap achieve sufficient velocity (i.” A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-10 .3. intervention efforts are unsuccessful because the efforts simply do not stop the erosion process or the erosion progresses too rapidly allow meaningful intervention efforts to be implemented. 500 A Downstream Slope Protection (Cobbles) B Zone 2 (SM) Zone 1 (SC) Zone 2 (SM) C Drainage Blanket “Initiating event: Poorly compacted layer Failure Path Description: The reservoir rises to elevation 466 feet. Upon saturation from the reservoir. The downstream cobble layer (slope protection) does not filter the Zone 2 material or the material does not have sufficient overburden and the slope protection layer blows off from reservoir pressure reaching the layer. which is 1 foot above the historic high reservoir elevation. Wet spots and flowing water appear on the downstream face of the embankment. Eventually this process of increasing seepage velocities and erosion progresses to full breach. slope instability. If the seepage expressions are seen and reported. A low density zone exists in the Zone 2 at this elevation due to a thick lift being placed during original construction of the embankment that was not compacted well by the equipment being used. Riprap on Gravel Bedding El.e. The breach occurs by mechanisms typical for this type of embankment (i. as a result of collapse settlement of the poorly compacted layer). gross enlargement of the developing pipe. the bottom of this layer settles and separates from the upper portion of the layer leaving a gap (i.Failure Modes and Load Partitioning SECTION 3 3. but these seepage expressions are not seen by either the public or project personnel. The gap widens which leads to increased seepage velocities and more erosion of the Zone 2 material. sufficient gradient and gap width) to initiate erosion and begin a concentrated leak erosion process..2 Examples The following sketch depicts an example of the description of a failure path for an initiating event involving seepage through a poorly compacted layer in the embankment. The riprap bedding layer does not function as a crack stopping material because it does not have sufficient volume due to its limited thickness or it is not of the proper gradation to be filtered by the downstream slope protection material. A roof forms through the Zone 2 material. It is followed by a description of the failure mode for seepage path A. 460 El.e.

E F G Adjacent spillway walls.5 . Related to irregularities in the foundation profile. In the foundation. B C D. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-11 . Figure 3.Typical embankment dam showing some key features associated with potential internal erosion failure paths.Failure Modes and Load Partitioning SECTION 3 Crest Spillway crest Crest Outlet conduit Spillway walls Intake tower PLAN A E B C D ELEVATION No filters above FSL FSL Rockfill Earthfill core F G Filters Rockfill Alluvium Rock SECTION LEGEND A. From embankment to foundation. Adjacent outlet conduit.

Examples of embankment crest details which may result in relatively high likelihood of internal erosion.6 . Failure paths that could be considered for this dam include: a) b) c) Internal erosion adjacent the spillway walls at A and B. and E for piping in the lower part.Failure Modes and Load Partitioning SECTION 3 Figure 3. where there are likely to be low stresses and a potential for cracking and hydraulic fracture due to differential settlement. Internal erosion over irregularities in the foundation e.5 shows a fairly typical embankment dam. Figure 3. Internal erosion adjacent to and into the outlet conduit C. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-12 . Continuation is also more likely because often the detailing of the dam design. at D for piping in the upper part of the embankment. it is more likely that features likely to lead to initiation of internal erosion are in the upper part of the dam. d) e) f) For many dams. Internal erosion for the remainder of the embankment. Internal erosion from the embankment to foundation at G. will give no or little filter protection. with a concrete spillway structure. This is because cracking due to differential settlement over large scale irregularities in the foundation profile is more likely to be present near the crest.6 gives examples of this with increased likelihood of flood loading near the crest of the dam.g. Figure 3. or as built. for example in high permeability layers. Internal erosion in the alluvium foundation (F).

7 is an example where there is a significantly different probability of internal erosion above the berm than below because the upper part is essentially a homogeneous dam while the sandy gravel in zone 2 may act as a filter. The failure path is evaluated by listing the adverse factors that make the failure mode “more likely” and the favorable factors that make the failure mode “less likely”.Failure Modes and Load Partitioning SECTION 3 Figure 3. These are based on the team’s understanding of the dam and background material.4.4 3.7 . This is best managed in the analysis by considering internal erosion above the berm (the upper part of the embankment) separately to below the berm (the middle and lower parts of the embankment). 3. Flow path A Flow path B Figure 3.1 Failure Path Screening Overview The purpose of the failure path screening process is to systematically review the potential failure paths/modes that have been identified and eliminate those from further consideration that are assessed to have negligible contribution to risk. experience shows that it is necessary to do the analysis in this detail to allow proper consideration of the probability of failure.Example of an embankment with significantly different probabilities of internal erosion above and below the top of the downstream berm While considering failure paths in this amount of detail may seem to be a lot of extra work. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-13 . and in reality it makes assessment of conditional probabilities easier because factors are not being lumped together.

5. the “upper” part of the embankment is the upper 1/3rd measured at the maximum height section and continuing across the valley at this level. (1) There is no vertical cliff in contact with the embankment OR (2) A wide bench is present at the base of the cliff (Wb/Hw > 2.1 to 3.1. and using the initiating mechanism screening criteria. The rationale for the inclusion or exclusion of each potential failure path should be fully documented. Table 5. The width of valley to dam height ratio W v /H > 2. These screened failure paths are not carried forward into the risk analysis.Screening of initiating mechanisms – Internal erosion due to concentrated leaks in transverse cracks in the upper part of embankment dams Exclude the Failure Path if the Following Conditions are Satisfied No exclusions apply – always include this failure path Exclude if. refer to Figure 5.1.5.2.4. Each failure path is then screened by the analysis team based on the consideration of the list of the more likely and less likely factors.2) IM3 – Transverse cracking due to cross valley arching Exclude if.2.2 Internal Erosion Through the Embankment In Table 3.1.5 Reference Section and Table Section 5.2 the “lower” part of the embankment is below the top 1/3rd.2).1 . Table 5.2. Table 5. Section 5. OR (3) The abutment slope below the cliff is gentle (β1 < 25°.1.3 Initiating Mechanism IM1 – Transverse cracking due to cross valley differential settlement IM2 – Transverse cracking due to differential settlement adjacent to a vertical cliff at the top of the embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-14 . (refer to Figure 5.Failure Modes and Load Partitioning SECTION 3 Screening of the failure paths is also evaluated using the screening criteria for the various initiating mechanisms which are summarized in Tables 3. 3.1 Section 5.3). refer to Figure 5. Table 3. In Table 3. The primary intent is to identify those failure paths that are clearly so remote as to be negligible or non-credible.

7 IM5 – Transverse cracking due to differential settlements in the foundation beneath the core IM6 – Transverse cracking due to differential settlements due to embankment staging IM7 – Cracking in the crest due to desiccation by drying Exclude if. (1) The dam is zoning type homogeneous earthfill. OR (3) Finite Element Analyses have demonstrated that stresses are such that hydraulic fracture is very unlikely. Reference Section and Table Section 5. Table 5.11 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-15 . Section 5. earthfill with filter drains or zoned earthfill.Failure Modes and Load Partitioning SECTION 3 Initiating Mechanism IM4 – Transverse cracking resultant on cross section settlement Exclude the Failure Path if the Following Conditions are Satisfied Exclude if.9 Section 5. The reservoir stage being considered is below the likely depth of desiccation cracking. Table 5. There is no compressible soil in the foundation below the core Exclude if. OR (2) Evidence from relative settlements of core and shoulders that the materials have a similar modulus.2.2.3. Table 5.4 Section 5.5. The embankment construction was not staged Exclude if.2.2.2.

OR This mechanism only applies where there has been a seasonal shutdown during construction. Table 5. The reservoir stage being considered is below the level of the seasonal shutdown surface.Failure Modes and Load Partitioning SECTION 3 Initiating Mechanism IM8 – Cracking on seasonal shutdown layers during construction and staged construction due to desiccation by drying Exclude the Failure Path if the Following Conditions are Satisfied Exclude if. Below that any desiccation cracks should have swelled and closed.5 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-16 .2. OR Very good control and clean up practices used – desiccated layers removed from the embankment and replaced with new soil or adequately reworked to specified moisture content. or the embankment has been staged.5.14 IM13 – Cracking due to earthquake No exclusions apply – always include this failure path Section 5. OR This mechanism only applies above the level of saturation of the core. Reference Section and Table Section 5.

OR (2) Dam has a wide core (W/H>1.Screening of failure paths – Internal erosion due to concentrated leaks in transverse cracks in the middle and lower parts of embankment dams Exclude the Failure Path if the Following Conditions are Satisfied Exclude if. Shoulders poorly compacted or dumped. Table 5. IM11 – Transverse cracking or hydraulic fracture in the lower part of the embankment due to differential settlement in the foundation under the core IM12 – Transverse cracking at the foundation contact due to small scale irregularities in the foundation profile under the core Already considered in upper part of dam.3. Uniform abutment profile without benches Exclude.2.3 and 5.2. (1) For all dam zoning types other than central core earth and rockfill (or gravel shells).3. Table 5.3.4.16 Section 5.1. Core compacted >98% SMDD.3. The persistence of the irregularity across the width of the core is less than 50% of the core base width Section 5. Sections 5.Failure Modes and Load Partitioning SECTION 3 Table 3.9 Reference Section and Table Section 5.18 Failure Path/Location IM9 – Transverse cracking due to cross valley differential settlement IM10 – Transverse cracking due to differential settlement causing arching of the core onto the shoulders of the embankment Exclude if. Table 5.3. and puddle core earthfill dams.0) OR (3) Core has higher modulus than shells.20 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-17 .2 . Table 5.

with good mixing. rubber tyres.1. moisture content 2% dry to 1% wet of OWC.4 Reference Section and Table Section 6.2 for cohesionless soils Failure Path/Location IM14 – Poorly compacted or high permeability layer in the embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-18 . (1) For cohesive soils (Plasticity Index > 7).1.6 Section 6. OR If the reservoir stage being considered is below the likely depth of freezing Section 6. OR (4) Compacted soil foundation IM16 – Poorly compacted or high permeability layers in the crest due to freezing Exclude if. IM15 – Poorly compacted or high permeability layer on the corefoundation contact Exclude if. with good documentation and records. bonding and compaction of contact fill.2. >75% relative density.Failure Modes and Load Partitioning SECTION 3 Table 3.2. OR (3) Uniform well compacted soil foundation. and soils well compacted (contact soil compacted using special compaction methods (e.Screening of failure paths – Internal erosion due to concentrated leaks in poorly compacted or high permeability zones in the embankment Exclude the Failure Path if the Following Conditions are Satisfied Exclude if.2.3. Table 6. (1) Contact soils are well compacted on a regular foundation surface with good documentation and records OR (2) Uniform or regular rock surface or surface treated with shotcrete or concrete to correct slope irregularities. Table 6. compaction wet of OWC). use more plastic material. ≥98% standard dry density ratio. The climate is such that temperatures do not fall below freezing point except possibly overnight or for a day or two.2. OR (2) For cohesionless soils and soils with PI ≤ 7. Table 6.1 for cohesive soils (PI>7) Table 6. All soils are very well compacted with lift thicknesses less than 8 inches.3 .g.

11 IM19A.2. Table 6. IM19B – Erosion into a (non-pressurized) conduit Exclude if. OR If the reservoir stage being considered is below the likely depth of desiccation cracking Reference Section and Table Section 6.3. backfilled to the surface with concrete Section 6.2.13 IM20 – Poorly compacted zone associated with a spillway or abutment wall IM21 – Crack/gap adjacent to a spillway or abutment wall Exclude if.4. Table 6.3.19 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-19 . OR Very good control and clean-up practices were used – Frozen layers removed from the embankment and replaced with new soil or adequately reworked to specified moisture content.2.3.Failure Modes and Load Partitioning SECTION 3 Failure Path/Location IM17 – Seasonal shutdown layers during construction and staged construction surfaces due to freezing Exclude the Failure Path if the Following Conditions are Satisfied Exclude if. The climate is such that temperatures do not fall below freezing point except possibly overnight or for a day or two. (1) There is no conduit passing through the embankment.9 IM18 – Poorly compacted or high permeability zone around a conduit through the embankment Exclude if. Table 6.4.3.1.2. Table 6. (1) There is no spillway or abutment wall in contact with the embankment Exclude if. OR (2) The conduit is totally embedded in a trench excavated in non-erodible rock.17 Section 6. Section 6. (1) There is no spillway or abutment wall in contact with the embankment Section 6. Table 6. (1) There is no conduit passing through the embankment. OR (2) Careful internal inspection of conduit showing no evidence of open joints or cracks.

(1) There is no spillway or abutment wall in contact with the embankment OR (2) A wide bench is present at the base of the wall (Wb/Hw > 2. Table 6.4.23 Note. Table 6. Section 6.5.4. refer to Figure 5.Failure Modes and Load Partitioning SECTION 3 Failure Path/Location IM22 – Differential settlement adjacent to a spillway or abutment wall Exclude the Failure Path if the Following Conditions are Satisfied Exclude if.2). refer to Figure 5.5.21 IM23 – Wrap around details for connection of embankment dam to concrete gravity dam Exclude if.2) Reference Section and Table Section 6. (1) There is no wrap around connection of an embankment dam to a concrete gravity dam. OR (3) The abutment slope below the wall is gentle (β1 < 25°. Cohesionless soils are soils with zero Plasticity Index A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-20 .4.

Section 7. it terminates beneath the dam. (1) The foundation soil has a Plasticity Index ≥ 7.e. (1) The foundation soil is cohesionless. No exclusions apply Section 7. suffusion.5 Section 8 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-21 . OR (2) The proportion of the finer fraction is less than 40% of the total mass of the soil.4 . (Note that erosion across the cut-off trench is considered separately in internal erosion of the embankment into or at the foundation). (1) The foundation soil has a Plasticity Index ≥ 7. (1) The soil layer beneath the dam is isolated by a cut-off trench founded in non-erodible rock.Failure Modes and Load Partitioning SECTION 3 3.Screening of failure paths – Internal erosion through the foundation Exclude the Failure Path if the Following Conditions are Satisfied Exclude if. refer to Figure 7.1) Exclude if. OR (2) If the cohesionless soil or soil with PI ≤ 7 layer is not continuous below the embankment (i.4.3 Internal Erosion Through the Foundation Table 3. Exclude if. erosion in a crack) IM24 – Backward erosion in a cohesionless soil foundation Suffusion in a cohesionless soil in the foundation IM25 – Suffusion in a cohesionless soil in the foundation IM26 – Erosion in a crack in cohesive soil in the foundation IM27 – Erosion in defects in a rock foundation Exclude if.2 Reference Section and Table Section 7 Initiating Mechanism All modes of internal erosion of the foundation (backward erosion.4 Section 7.

shotcrete. refer to Section 10. relic joints or solution features.g.1. Section 9.4 Internal Erosion of the Embankment into or at the Foundation Table 3.4).e.Screening of failure paths – Internal erosion in the embankment into or at the foundation Exclude the Failure Path if the Following Conditions are Satisfied Exclude if. slush grouting mortar treatment) Reference Section and Table Section 9. satisfy the No Erosion criteria.4.4 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-22 . (1) Rock foundation below the core is comprised of rock containing closed rock defects (<1 mm wide) or defects open less than 3D 95 .of the fine limit of the core OR (2) Rock foundation below the core has been adequately treated (e.5 . OR (2) Soil foundation below the core is comprised of sands (SP or SW) which are filter compatible with the embankment materials (i.Failure Modes and Load Partitioning SECTION 3 3. and the soil does not contain macrostructure such as root holes.3 Initiating Mechanism IM28 – Internal erosion of the embankment into or at a rock foundation IM29 – Internal erosion of the embankment into or at a soil foundation Exclude if.075mm)). (1) Soil foundation below the core is comprised of fine grained soils with greater than 12% fines (fraction finer than No 200 sieve (0.

For many dams this level is achieved in most years of operation and has an annual probability of being reached approaching 1. The probabilities of failure estimated for each of the flood loading partitions can be used as point estimates for developing fragility curves (refer to Section 4. • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-23 . For storages which are operated to store water for irrigation or hydropower.5. It is also known as “historic high reservoir level” and “water surface of record”.6(a).2).5. For a ‘dry dam’. and at 1 in 50 AEP flood in Figure 3. It is an important level because the embankment and its foundations have been tested up to this level.9.6(b). For example at 1 in 1000 Annual Exceedance Probability (AEP) flood in Figure 3.2 Pool of record level The pool of record (POR) level is the maximum level the reservoir has reached during its operation. For a flood control storage the normal loading is the reservoir level that is sustained year in and year out. 3. Changes in design.3 Partitioning of reservoir levels The reservoir levels should be partitioned to coincide with: • • • The top of active conservation (full supply) level Flood pool of record level (also called historic high reservoir level).5 3. there is no normal water level and all water levels are associated with storing a flood and are regarded as flood loadings. Embankment crest level.Failure Modes and Load Partitioning SECTION 3 3. For most of these dams there is usually a minimum pool level set.1 Partitioning of the Reservoir Levels “Normal” and “flood” loading levels It has been USACE and Reclamation practice to determine the risks for “normal” and “flood” loading separately. the top of “normal” loading is the top of active conservation level (“full supply level”) or for example the level of the crest in an un-gated spillway. 3.0.5.

g.5 Annual Exceedance Probability of the reservoir level 1.0199 0.6 . For flood control dams which have not yet experienced a flood reaching the reservoir spillway crest level this may be used as a partition boundary. Topographic features such as major changes in foundation profile if these are above the pool of record.48 0. If it is the policy of the agency to separate “normal” and “flood” risks then one reservoir partition level will need to coincide with the reservoir level at the boundary between these cases.5 0. Table 3.0001 0.6 shows an example of reservoir level partitioning.0 Reservoir Level Partition Range Probability/annum the reservoir level is in this range A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-24 .Failure Modes and Load Partitioning SECTION 3 • Geological features which occur above a particular level in the foundation (e. a highly permeable gravel layer). • For flood control storages it will probably be necessary to partition at for example 1 in 100 AEP and 1 in 1000 AEP flood levels even if these do not coincide with design features. Table 3.Example of reservoir level partitions Reservoir Level RL100 Significance Minimum pool level of record (or “Minimum pool of record”) RL100 – RL200 RL200 Top of active conservation level (or “Full pool”) RL200 – RL205 RL210 Pool of record level (or “Historic high reservoir level”) RL205 – RL215 RL215 Embankment crest level (or “Dam crest flood”) >RL215 0.02 0.0001 0.

not the proportion of time the reservoir is above the level. It is recommended that the minimum peak ground acceleration level be taken as 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 3-25 .Failure Modes and Load Partitioning SECTION 3 3.01 0. Partition the loads to form a table as shown in Table 3. transverse cracks are known to exist above a change in foundation slope).9 for earthfill and rockfill dams. This is done by developing the plot of reservoir level versus the proportion of time the reservoir level is exceeded. This is modelled as the first node in the event tree before the earthquake loading.1g.0 Annual exceedance Probability of the earthquake loading 0.0 0.6 Earthquake Load Partitioning Evaluate the peak ground bedrock acceleration and earthquake magnitude versus Annual Exceedance Probability (AEP) for the site. then the minimum PGA to be analysed could be reduced to 0. It will also be necessary to consider the reservoir level at the time of the earthquake.001 0.30g 7. This is because internal erosion and piping often develops quite quickly and may go from initiation to breach in hours or days.g. Table 3. If a particularly vulnerable element exists within the dam (e. Probabilities of failure should be developed for each of the earthquake loading ranges. 3.0001 0.Example of earthquake load partitions Earthquake peak ground acceleration Representative earthquake magnitude 6.5.7 .8 for earth dams and Figure 5. or cracking may be undetected.20g 6.99 < 0.0 0.10g The earthquake peak ground acceleration values should be selected to coincide with the damage class contours given in Figure 5. In some situations the reservoir may rise to higher levels before repairs to cracks caused by seismic loading can be affected.0009 0.4 Assessing Frequencies of Reservoir Loading For internal erosion and piping it is the annual probability that the level in the partition is exceeded regardless of how long for.7. The possibility of these scenarios should be assessed.05g.0001 Probability/annum the loading is in this stage 0.009 0.

In poorly compacted zones initiation of erosion may be a result of the voids between aggregated soil particles giving higher permeability and continuous open paths.1. They are used in the event trees to calculate probabilities of failure for each load partition. allowing for under reporting of the incidents in the database. desiccation. or erosion was stopped by filters. and do not include the assessment of whether erosion initiates in the crack. the cracks sealed by swelling. Estimated frequencies of the occurrence of cracking.Application of Tables for Estimating Conditional Probabilities SECTION 4 4 Application of Tables for Estimating Conditional Probabilities 4. They have been determined from analysis of historic dam accidents and failures. The probabilities have been assessed using the expert judgement of the workshop attendees. Where practical these have been anchored to historic data. hydraulic fracture or poorly compacted or high permeability zones in embankments are presented in Table 4. These probabilities are conditional on the reservoir or earthquake loading and are known as the “system response”. 4. or in poorly compacted zones. or collapse settlement of the poorly compacted layer leading to a flaw or continuous open path in which water can flow and erode the sides of the flaw as happens in a crack. This has mainly been possible in the estimation of the probability of initiation of erosion in concentrated leaks and is discussed further in Section 4. with Sections 5 to 9 covering Initiation of Erosion. These tables have been developed to model the physical processes so far as practical. and Section 13 Breach. The estimation of conditional probabilities is covered in Sections 5 to 13. hydraulic fracture or poorly compacted or high permeability zone. Intervention and Repair. Section 12 Detection. Section 11 Progression. Tables are presented within the sections to give guidance on the estimation of conditional probabilities.1 are for a crack.2 and in the Supporting Document.2 Historical Frequencies of Cracks and Poorly Compacted or High Permeability Zones in Embankments Concentrated leak erosion may occur in cracks caused by differential settlement. cracking which may have been present in the dams in the database but were sufficiently resistant to erosion for erosion not to initiate. hydraulic fracture or poorly compacted or high permeability zone being present. These historic frequencies have been used as a basis for anchoring the estimated probabilities of a flaw being present in which erosion may initiate.1 General Approach For each failure mode the conditional probabilities at each node in the event tree are estimated for each reservoir and earthquake load partition. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-1 . Details of the database and the analysis are given in the Supporting Document. were above the Pool of Record (POR) of the dams. The frequencies in Table 4. Section 10 Continuation.

Estimated historical frequencies of cracking. hydraulic fracture or poorly compacted or high permeability zones First filling 0.014 0.004 Reservoir Level below Pool of Record 0. Location of cracking/Poorly compacted or high permeability Zone In embankment (dam body) Associated with conduit Associated with concrete wall or structure through embankment Estimated historical frequencies of cracking.3 meters). A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-2 . Table 4. hydraulic fracture or poorly compacted or high permeability zones in embankment dams.0003 per annum The historical frequencies for cracking or poorly compacted or high permeability zone in the dam body were further subdivided into the various mechanisms of crack formation. This is so the above pool of record frequencies apply when the reservoir is testing significant new areas of the dam.0007 per annum 0.01 0.1 .2 presents the estimated historical frequencies for each of the mechanisms.004 Reservoir Level above Pool of Record 0. This was necessary so as to avoid the double counting of the historical frequencies when the probabilities for each initiating mechanism are added together. Table 4.014 0.001 per annum 0.Application of Tables for Estimating Conditional Probabilities SECTION 4 When applying the historic frequencies to predict future behaviour it is assumed that the “Reservoir Level above Pool of Record” values apply when the reservoir level exceed the pool of record level by at least one foot (0.01 0.

2004).0004 As described in the Supporting Document. The historical frequencies for the “Below Pool of Record” case are quoted as annualized frequencies.001 100% 63% (47%) (16%) 37% 0. For ease of application these are applied directly as conditional probabilities as the majority of dams have a seasonal cycle of reservoir fluctuation and reach a particular reservoir level each year.Application of Tables for Estimating Conditional Probabilities SECTION 4 Table 4.1 and Table 4.2 represent the average frequencies for cracks/ hydraulic fracture/poorly compacted or high permeability zones across the population of dams in the database. USACE and Australian dams would have average probabilities estimated based on these historical data less than the average historical frequencies presented in the tables. the historical frequencies presented in Table 4. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-3 .0002 0. These represent a large number of dams of varying age and varying levels of engineering design and construction practice. hydraulic fracture or poorly compacted or high permeability zones Mechanism for Cracking or Poorly Compacted or high permeability Zone Proportion of Cases All incidents (cracking.007 0.002 0.Historical frequencies for cracking or poorly compacted zone in the embankment dam body.005 0.g. They do not represent the historical frequencies for the “average dam” a term which has been applied previously (e.0005 0.014 0. Estimated historical frequencies of cracking.2 . Fell et al 2003. It would be expected that the portfolio of well engineered Reclamation. The database of incidents is relatively small and because of this there should be no further subdivision of the failure and accident statistics. hydraulic fracture and poorly compacted or high permeability zone) Cracking and hydraulic fracture In upper part In middle/lower part Poorly compacted or high permeability zone in upper and lower parts (total) Reservoir Level above Pool of Record Reservoir Level below Pool of Record (per annum) 0.

the use of historic performance data for anchoring the conditional probabilities for internal erosion through the foundation and internal erosion from the embankment into the foundation is not recommended. Where there is historic data to “anchor” the probabilities these are shown. (0. These tables are structured to show: • • The Factors The Relative Importance of this Factor (RF) with numeric weightings (usually three Factors with Relative Importance weightings of 3. [0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-4 .4. Where the “anchor” probabilities are estimated by expert judgement of the Toolbox development team they are shown with rounded brackets e.g.3m) above this historic high reservoir pool level. and interpolation between the bracketed probability values should be based on log interpolation. and historic performance data provides very little information to aid in the assessment.4 4. The “Below POR” figures are for reservoir level stages with a representative level up to and 1 foot (0. For this reason. The conditional probabilities on these tables are on a log scale. 2 and 1). Generally there are four for each Factor. The “Above POR” figures are for representative reservoir level stages at least 1 foot (0.1 Estimating Conditional Probabilities Estimating Conditional Probabilities Using Relative Importance Factors and Likelihood Factors Most of the Sections for estimation of conditional probabilities are structured so there is a Table of Factors affecting the likelihood. with Likelihood weightings of 4.0005] for Table 5. Likelihood Factors (LF) for which there are descriptions.g. “Below POR”.3m) above the historic high reservoir pool level (POR). 3. and hence historic frequencies for internal erosion in and into the foundation are more difficult to interpret and apply.003). The tables have two or more sets of probabilities.2. • There is then a second table which links ∑ (RF) x (LF) to the conditional probability. 4.3 Historical Frequencies for Internal Erosion in and into the Foundation The conditions in the foundations of dams are inherently more complex and varied than in the dam body. One of the key issues for these modes of internal erosion is whether continuous seepage paths or open defects are present.Application of Tables for Estimating Conditional Probabilities SECTION 4 4. e. 2 and 1.

6. and the analysis methods themselves have uncertain outputs. 4. insufficient to affect the outcome. The effect is dependent on the failure mode. (Section 5. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-5 . 4.Application of Tables for Estimating Conditional Probabilities SECTION 4 These anchor probabilities are a form of a “base rate frequency” and the approach used in the tables is a base rate frequency approach.3 Estimating Conditional Probabilities Using Probability Estimate Tables In some Sections. Those carrying out the risk analysis are required to choose which of the descriptors for each Factor best reflects conditions at the dam.4. For some conditional probability estimates it has been necessary to modify the format of the table to better model Relative Importance or Likelihood distributions. 4. In the tables the term “negligible” means that the contribution to the probability of failure would be very small indeed. “Scenario Tables” are used.g.4. e. those describing the assessment of the probability of initiation of erosion in a crack. Where there is little data this assessment should be made on the best available information and using judgement based on geological conditions and experience elsewhere on dams of similar age and design. These tables present ranges of conditional probabilities within which the risk analysis team are to select their best estimate based on the details of the dam they are analyzing. the Toolbox development team have carried out analyses to simplify the estimation of conditional probabilities from input data which itself has significant uncertainty.2). and the experience of the team members. The assumptions made to develop these tables are described in the Supporting Document. These have been developed by the Toolbox development team based on published information. how the embankment is partitioned for the analysis.5 Length Effects The effect of the length of the embankment being considered may have an influence on the assessed probability of internal erosion and piping. and how the conditional probabilities are assessed.2) and initiation and progression of backward erosion in cohesionless soils (Section 6.4.2 Estimating Conditional Probabilities Using Scenario Tables Where the number of factors affecting the estimation of conditional probabilities is few or to be based on limited data.

road pavement) over the core. e. zoning and material properties?” In statistical terms if all sections of the embankment are exactly the same. high permeability zones in the embankment (e. such as a conduit. high permeability layer on the core-foundation contact. The length can be accounted for by considering it in determination of the probability.e. The failure modes which are potentially affected by length are cracking due to desiccation (either by drying and/or freezing). internal erosion of the embankment into or at a rock or soil foundation. and differential settlement of a major change in foundation profile. P = 1 – (1-P1) x (1-P2) x (1-P3)…etc.6. different specifications and construction methods.g. different construction materials. The notes for all of the tables are located after Table 4.g. contact with a wall. then each section should be considered separately and the probabilities added using De Morgan’s rule.g. different cover (e. In many embankments. The following tables describe whether length effects are applicable to each of the internal erosion failure paths/location. If each section is completely independent of the others. these failures modes contribute most to the likelihood of internal erosion and piping. “what is the probability of a through going crack in the core due to desiccation in this 1000 feet length of embankment which so far as can be ascertained all has the same geometry.Application of Tables for Estimating Conditional Probabilities SECTION 4 Many failure modes are independent of length because they are related to specific features in the embankment. and backward erosion in cohesionless soils in the foundations of dams. For these failure modes it is the likelihood of initiation of erosion for which length may be a factor. due to poorly compacted layers in the core). i. then they are perfectly correlated and the probability of at least one crack in the whole of the embankment is the same as for one section regardless of length. the method of accounting for length effects. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-6 . and if so.

so it will be sufficient to calculate for this case only In most cases one abutment cliff will be much larger than the other. so it will be sufficient to calculate for this case only IM2 Transverse cracking due to differential settlement adjacent a vertical cliff at the top of the embankment IM3 Transverse cracking due to cross valley arching Crack/Gap No length effect but consider each abutment and add the calculated probabilities Long Section No length effect Crack Long Section IM4 Transverse cracking resultant on cross section settlement No length effect for case (a) Long Section (a) Long Section (b) No length effect for case (b) but each part contributes and add the calculated probabilities No length effect for cases (a) and (b). No length effect for case (c) but each part contributes and add the calculated probabilities In most cases one part of the embankment will be larger than the other. so it will be sufficient to calculate for this part only IM5 Transverse cracking due to differential settlements in the foundation beneath the core Long Section Compressible soil Long Section (a) (b) Long Section (c) IM6 Transverse cracking resulting from differential settlements due to embankment staging Long Section Crack Stage 2 Stage 1 No length effect.Application of Tables for Estimating Conditional Probabilities SECTION 4 Table 4. If more than one staging surface add the probabilities A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-7 .3 – Length Effects – Transverse cracking in the upper part of the embankment Initiating Mechanism IM1 Transverse cracking due to cross valley differential settlement Long Section Sketch of Failure Mode Crack Effect of length on failure probability No length effect but consider each abutment and add the calculated probabilities Comments In most cases one abutment or change in foundation profile will be much larger than the other.

Same as IM6 IM8 Cracking on seasonal shutdown layers during construction and staged construction surfaces due to desiccation by drying Notes. Then the correlated condition exists.6 for notes. so length effects will be negligible.Application of Tables for Estimating Conditional Probabilities SECTION 4 Initiating Mechanism IM7 Cracking in the core near the crest due to desiccation by drying Sketch of Failure Mode Long Section Effect of length on failure probability Length effects may apply. Long Section Crack Stage 2 Stage 1 Same as IM6 Refer to the end of Table 4. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-8 . See Notes (1) and (2) Comments In many cases the nature of the embankment design and construction materials will be such that all sections are the same.

4 – Length Effects – Transverse cracking in the middle and lower parts of the embankment Effect of length on failure probability No length effect but consider each abutment and add the calculated probabilities No length effect for case (a) Long Section (a) Initiating Mechanism IM9 Transverse cracking due to cross valley differential settlement Sketch of Failure Mode Comments In most cases one abutment or change in foundation profile will be much larger than the other.Application of Tables for Estimating Conditional Probabilities SECTION 4 Table 4. Long Section No length effect.No length effect for case (c) but each part contributes and add the calculated probabilities IM11 Transverse cracking or hydraulic fracture in the lower part of the embankment due to differential settlement in the foundation under the core Long Section Compressible soil Long Section (a) (b) Long Section (c) IM12 Transverse cracking at the foundation contact due to small scale irregularities in the foundation profile under the core Notes. so it will be sufficient to calculate for this case only In most cases one part of the embankment will be larger than the other. See Note (3) Refer to the end of Table 4.6 for notes. so it will be sufficient to calculate for this part only Crack Long Section IM10 Transverse cracking due to differential settlement causing arching of the core onto the shoulders of the embankment Long Section (b) No length effect for case (b) but each part contributes and add the calculated probabilities No length effect for cases (a) and (b). but allow for the number of irregularities and add the calculated probabilities. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-9 .

Then the correlated condition exists. Refer to the end of Table 4. IM22 Poorly compacted or high permeability zone. IM17 Cracking in the crest or seasonal shutdown layers during construction due to desiccation by freezing Long Section Poorly compacted layer Length effects may apply. If there is more than one conduit treat each separately and add the probabilities IM19A Erosion into a (non-pressurized) conduit IM19B Erosion into a (non-pressurized) conduit leading to erosion along the conduit IM20.Application of Tables for Estimating Conditional Probabilities SECTION 4 Table 4. See Notes (1) and (6) In many cases the nature of the embankment design and construction materials will be such that all sections are the same. IM21. If there is more than one conduit treat each separately and add the probabilities No length effect. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-10 .5 – Length Effects – Poorly compacted or high permeability zones in the embankment Initiating Mechanism IM14 Poorly compacted or high permeability layer in the embankment Sketch of Failure Mode Long Section Poorly compacted layers Effect of length on failure probability Length effects may apply.6 for notes. so length effects will be negligible IM15 Poorly compacted or high permeability layer on the core-foundation contact IM16. See Notes (1) and (5) Long Section Length effects may apply. crack/gap associated with a spillway or abutment wall Long Section Erosion into Conduit Crack/Gap Long Section No length effect. See Notes (1) and (4) Comments In many cases the nature of the embankment design and construction materials will be such that all sections are the same. Then the correlated condition exists. so length effects will be negligible IM18 High permeability zone around a conduit through the embankment Long Section High Permeability Zone No length effect. If there is more than one wall treat each separately and add the probabilities Notes.

e. road pavement) over the core. See Note 7 Notes for Tables 4. See Notes (1) and (7). (1) If all sections of the embankment are exactly the same. For desiccation induced cracking see IM6. See Notes (1) and (7).3 to 4. different specifications and construction methods. If each section is completely independent of the others. each section should be considered separately and the probabilities added using De Morgan’s rule. i.12 are determined by expert judgement and are for an embankment about 1600 feet (500 meters) long. so there will be many cracks of maximum depth within the 500 meters.6. See Notes (1) and (7). See Notes (1) and (7). different construction materials.g. Comments See Note 7 Depends on the cause of the cracking. IM26 Erosion in open or in filled defects in a rock foundation Long Section IM27 Internal erosion of the embankment into or at a rock foundation Long Section Length effects may apply. for settlement induced cracking see IM9 or IM11 Length effects may apply. different cover (e. then they are perfectly correlated and the probability of cracking in the whole of the embankment is the same as for one section regardless of length. (2) The probabilities in Table 5.Application of Tables for Estimating Conditional Probabilities SECTION 4 Table 4. These probabilities will apply for embankments shorter than 1600 feet (500 meters) without adjustment for length. See Note 7 IM28 Internal erosion of the embankment into or at a soil foundation Long Section Length effects may apply. The spacing of the cracks is likely to be about 5x crack depth. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-11 .6 – Length Effects – Internal erosion in the foundation and into the foundation Initiating Mechanism IM24 Backward erosion in a cohesionless soil foundation Suffusion in a cohesionless soil in the foundation IM25 Erosion in a crack in cohesive soil in the foundation Long Section Sketch of Failure Mode Long Section Effect of length on failure probability Length effects may apply.

2 and 6. (6) The probabilities in Tables 6. Estimate the probability of backward erosion or suffusion for each section and add the probabilities using De Morgan’s rule. (4) The probabilities in Tables 6. 6.5 are anchored by relation to historic data and are for an embankment about 1600 feet (500 meters) long. There are likely to be a number of sections within 1600 feet (500 meters) which have the conditions potentially leading to a flaw.21 are anchored against historic data. the number of boils is not a factor in assessing the probability as one boil or many boils both mean initiation has occurred at that level. If these occurred at the same reservoir level for dams. These probabilities will apply for embankments shorter than 1600 feet (500 meters) without adjustment for length. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-12 .6.Application of Tables for Estimating Conditional Probabilities SECTION 4 (3) The probabilities in Tables 5. regardless of its length. (5) The probabilities in Tables 6. De Morgan’s rule is P = 1 – (1-P1) x (1-P2) x (1-P3)…etc. Within a section the correlated condition exists.9 and 6. 6. There are likely to be a number of sections within 1600 feet (500 meters) which have the conditions potentially leading to a flaw. 6.4 and 6. for which there were between 2 and 5 small scale irregularities.7. These probabilities will apply for embankments shorter than 1600 feet (500 meters) without adjustment for length. There are likely to be a number of sections within 500 meters which have the conditions potentially leading to a flaw. Any section may have experience one or more sand boils. (7) The foundation should be partitioned so that geotechnical conditions are essentially the same within a section. Use these figures unless there are greatly more than 5 small scale irregularities.1.10 are determined by expert judgement and are for an embankment about 1600 feet (500 meters) long.20 and 5.3 are anchored on historic data and are for an embankment about 1600 feet (500 meters) long. These probabilities will apply for embankments shorter than 1600 feet (500 meters) without adjustment for length.

1 Nature of the estimates of probabilities given by the Toolbox The Toolbox gives “Best Estimate” Probabilities The methods in the Toolbox provide “best estimates” of the conditional probabilities and hence “best estimate” probabilities of failure.e.000 dams in the ICOLD (1986) survey of failures and accidents.2.2 Adjusting the Toolbox Best Estimates 4.1 Allowance for factors not included in the Toolbox methods It is recommended that the Toolbox estimate be adopted except where there are factors not covered in the Toolbox tables which the risk analysis team believe affect the estimate of the conditional probability for the node in question. 4. taking into account the feedback from trials and reviews of the Toolbox in Reclamation and USACE. The additional factors should be described and the reasoning for the revised estimate provided in the risk analysis report.2 What to do if the Toolbox estimates seem incorrect. Probabilities for some of the most important initiating modes within the embankment are calibrated against historic performance of dams from a large database of around 10.6. It would not be expected that this would generally result in conditional probabilities greatly different to those estimated by the Toolbox.Application of Tables for Estimating Conditional Probabilities SECTION 4 4.2. construction of the event tree) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-13 .2.6. This is discussed in detail in Section 4. They are designed to avoid systematic bias towards conservative or non-conservative probabilities. This may be due to 1) the logic (i. 4. The methods in the Toolbox are likely to be more reliable in assessing relative probabilities between failure modes and between dams than assessing absolute values. In these cases the best estimate should be determined by adjusting the toolbox estimate to allow for the additional factors. There may be cases where the toolbox estimates of failure probabilities are significantly different to what the risk analysis team would have expected. These factors may include observations or monitoring data. The estimates are determined by expert judgement based on analyses and laboratory tests modelling the physical processes. or physical factors not allowed for in the Toolbox.6. Where this has not been possible expert judgement of the team developing the Toolbox based on extensive experience in dams and risk assessment has been used.6 4.6.

then the failure mode could be developed into its own event tree or adjustments made to the portion of the toolbox event tree in question followed by assigning probabilities using one of the mapping schemes given in Appendix E or even use the original toolbox RF and LF factors if they are still appropriate. If there seems to be an issue with the probabilities assigned.3 Limitations of the methods used in the Toolbox The Toolbox is based on the (2007) state of the art on modelling the mechanics of initiation of internal erosion in cracks and other flaws. The Agency Person who manages the Toolbox should oversee the corrections/adjustments. or expert judgement anchored on historic data such as those detailed in Fell et al (2003. If in the end it appears that the factors should be adjusted or changed. the analyst is then equipped to assess the appropriateness of the prescribed probabilities and attempt an adjustment. Experience has shown that the logic presented in the toolbox has stood the test of time and is normally considered appropriate. If in the end there still is an issue. 2004). Be sure that the failure mode is well understood and that the risk analyst has a good understanding of the available case histories upon which the toolbox has been built around. In this case the issue should also be referred to one of the current toolbox developers for consideration in future updates. The methods available are sufficient to form the basis of the Toolbox. check with Agency personnel familiar with the toolbox to ascertain that a problem still exists.Application of Tables for Estimating Conditional Probabilities SECTION 4 does not well match the perceived failure mode for the dam being analysed. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-14 . The suggested steps to follow for each of these situations are: a) If the issue appears to be in the logic. fully discuss the factors presented to ascertain that the proposed new factors are indeed more important. possibly assisted by one of the mapping schemes given in Appendix E. look carefully at the failure mode being modelled to identify where the perceived difference has occurred. If there is still perceived to be an error in the toolbox logic. Once the analyst is well armed with this information. The adjustment should be made. In this case the issue should also be referred to one of the current toolbox developers for consideration in future updates. If there seems to be an issue in the RF and/or LF factors. by backward erosion and suffusion. Issues with the logic are typically due to a less-than-full understanding of what is already presented in the toolbox. 2) the RF and LF factors do not well represent what the team originally thinks are the key factors or actual observances.6. and then fully documented. and are to be preferred to methods based only on historic data. or 3) the probability value suggested in the toolbox does not well represent what the risk analyst thinks is appropriate. make the changes/adjustments and document well what was done and why. the risk analyst should be sure they are well aware of the database of historic precedent and experienced judgment that has been extensively used in the development of the method’s probabilities. b) c) 4.

For these failure modes the risk analysis team should develop an event tree to model the failure mode. expert elicitation techniques used in this document continues to be actively researched and/or studied. The methods used here are a significant improvement on the methods described in Fell et al (2003. It is likely that these developments in understanding will result in improved methods for assessing probabilities of initiation. analysis. It is recommended that those using Fell et al (2003. 4. and by expert judgement for the other nodes. modelling. more extensively researched and the methods are less subjective than for initiation. and breach are more subjective and largely based on case studies and expert judgement. Conditional probabilities within the event tree should be estimated using the Toolbox where the nodes are common to the Toolbox event tree. provided the data to do the analyses is available.Application of Tables for Estimating Conditional Probabilities SECTION 4 The mechanics of continuation (filter action) are fundamentally simpler. and breach mechanics. Most of the logic. laboratory testing. Appendix E details how this should be done. When this occurs it will be necessary to revise the Toolbox. 2004) now use the methods described in this Guidance Report.6. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-15 . Case histories continue to occur that give more insight to the process involved. 2004) which have been used in Australia and as input to the Reclamation methods. The modelling of progression.4 Assessment of probabilities of failure for failure modes which are not covered by the Toolbox In some dams there may be failure modes which are not well modelled by the Toolbox. continuation. and progression of internal erosion. detection and intervention.

1 Modelling uncertainty in the estimates of conditional probabilities Purpose of this section The Toolbox has been developed to provide “best estimate” values of conditional probabilities for the nodes in the event trees for all failure modes. This uncertainty may arise from limitations in the Toolbox methods used to estimate probabilities (a type of model uncertainty). the data available for the dam being analysed (epistemic uncertainty) and uncertainty associated with unpredictable variations of a random nature in data relied on for the risk analysis (aleatory uncertainty). uncertainty about the true accuracy and/or applicability of an analytical model used to assess the data. and or is somewhat contradictory.2 Sensitivity analysis There will be some cases where the quality and quantity of data available to do the risk analysis is limited.7.7. and measurement and parameter uncertainty of properties of materials in the dam and its foundation. Alternatively only the upper and lower estimates are carried through so decision makers can gauge the importance of this data on the risk analysis. The effects of uncertainty in the estimates of conditional probability can be examined either with sensitivity analysis or modelling uncertainty in the event tree. Both approaches are described in this section. Reference should be made to Reclamation (2001b). For example data on filters or transition zones may be limited.Application of Tables for Estimating Conditional Probabilities SECTION 4 4. Handling Uncertainty.7. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-16 . This approach should be adopted when uncertainty is not being modelled as detailed in Section 4. Dam Safety Risk Analysis Methodology. In some more detailed risk analysis it may be required that the uncertainty which is inherent in such estimates is modelled.7 4. In these cases it is recommended that the best estimate and the range of the best estimate for the node probability is calculated from this data and carried forward in the risk analysis as a sensitivity analysis. 4. Appendix T. This means that there will be a best estimate and two other estimates representing the range of estimates of the frequency of failure to be reported for this failure mode. Sometimes where there is considerable uncertainty or contradictory information in the data being used for the risk analysis a sensitivity analysis to gauge the effect of the likely range of estimates of conditional probability on one or two of the most critical nodes on the event tree may be modelled.3.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 4-17 .4.2 to 8. 4. Sections 8.3 to 13. Section 9.2 to 6. In these tables there is: i) Uncertainty in the historic data “anchor” probabilities. Reclamation practice is to not require uncertainty analysis for Comprehensive Facility Reviews (CFR) .1 Situations where uncertainty analysis may be required Whether uncertainty analysis is required is a matter for the Agency for whom the risk analysis is being carried out. Australian practice as discussed in ANCOLD (2003) is to carry uncertainties through the analysis and/or do sensitivity testing on important event tree nodes where the data available makes the assessment of that probability difficult.3. The USACE equivalent to a Reclamation CFR is the Periodic Assessment (PA).7.3.2 in Table 5. negligible. USACE does not require an uncertainty analysis.7.g. Where there are no historic data to “anchor” the probabilities there is uncertainty in the estimates based on the expert judgement of the Toolbox development team.3. These in turn affect the values in the tables in between. This covers the probability versus (RF)x(LF) tables in Sections 5 2 and 5.005]. except that the Senior Engineer doing the analysis may assign a range to the estimate based on his/her experience in estimating uncertainty in more detailed risk analyses. The following sections provide some information on how uncertainty may be modelled. and in the minimum and maximum probabilities (e.4). and 0.4. Instead a sensitivity analysis may be required as for Australian practice as detailed below.Application of Tables for Estimating Conditional Probabilities SECTION 4 4. The details of how to do this should be determined by the Agency to meet their needs.2 Modelling uncertainty for event tree nodes where relative importance factors and likelihood factors tables are used to estimate conditional probabilities. This uncertainty would then be used in Monte Carlo analyses to determine the distribution of estimates of the probability of failure. and Sections 13. Sections 6. Section 12.7.7. 4.5.3 Uncertainty analysis In some situations the Agency for whom the risk analysis is being carried out may require that the uncertainty of the estimate of conditional probability in each node of the event tree be modelled. [0.9.3. For Issue Evaluation Risk Analysis (IERA) and Risk Reduction Risk Analysis (RRRA) uncertainty analysis will typically be carried out by Reclamation.

Application of Tables for Estimating Conditional Probabilities

SECTION 4

ii) Uncertainty in the ability of these tables and the factors they are based on to model the relative probabilities. iii) Uncertainty in the analysis and laboratory testing data upon which the methods may be based. An example would be the numerical analysis upon which Table 5.3 is largely based. iv) Uncertainty resulting from limitations of the available investigations, design, construction and monitoring data relating to the probability being assessed. An example would be limitations of knowledge of the degree of compaction and borrow area variability in Table 6.1. (a) Model uncertainty Uncertainties (i) (ii) and (iii) are model uncertainties. Table 4.7 shows best estimate, and equivalent likely low and likely high probabilities. This can be used to develop likely minimum and likely maximum probability versus (RF)x(LF) tables. Table 4.7 is based on a dissection of the basis upon which the historic probabilities anchor points and the minimum and maximum values were determined. Details of how this was done are given in Section S4.7 of the Supporting Document. If the probability for (RF)x(LF) =6 is less than or equal to 0.0001, then the likely low probability should equal the best estimate. Table 4.7 allows definition of the anchor point, minimum and maximum probability values. The risk analyst should then interpolate to determine the intermediate values on the (LF)x(RF) table. An example is given in the Supporting Document Section S4.6. Reclamation Risk Analysis Methodology –Appendix T, Handling Uncertainty, use the terms “Reasonable Low” to represent the 10th percentile bound, and “reasonable high” to represent the 90th percentile bound. The range of 0.2x (likely low) to 5x (likely high) represents more stringent percentiles, possibly < 1% to >99%. It is for the agency doing the uncertainty analysis to judge what range the model uncertainty represents and to decide what range to adopt.

4-18

Application of Tables for Estimating Conditional Probabilities

SECTION 4

**Table 4.7 – Best estimate, likely high and likely low equivalence table
**

Best Estimate Probability 0.0001 0.001 0.01 0.02 0.05 0.1 0.2 0.3 0.5 0.9 Likely High Probability 0.0005 0.005 0.05 0.1 0.2 0.4 0.6 0.8 0.95 0.999 Likely Low Probability 0.0001 0.0002 0.002 0.004 0.01 0.02 0.04 0.07 0.15 0.4

Note. Likely high probabilities are assumed to be 5 times best estimate, and likely low probabilities 0.2 times best estimate. See Section S4.7 for details of the calculations to develop the values in the table.

(b) Data uncertainty The uncertainty in the probability estimate resulting from limitations in the data (factor (iv) above) should be assessed by assessing the likely low and likely high (RF)x(LF) values from the table, and using these in the Probability versus (RF)x(LF) tables to estimate the range of probabilities resulting from data uncertainty. (c) Combining the model and data uncertainty. To combine the model and data uncertainty, use the likely low, best estimate and likely high probability from the relevant probability versus (RF)x(LF) table. An example is given in the Supporting Document, Section S4.6. This is a severe test of overall uncertainty and Agencies will need to develop their policy on how to combine these components of uncertainty. 4.7.3.3 Modelling uncertainty for event tree nodes where scenarios and examples are described, and a range of probabilities provided. This covers the Tables where scenarios are described, examples are given, and a range of probabilities provided from which the risk analyst makes a selection based on the available information, and the analysts degree of belief. Examples are Tables 10.13, 11.1, 11.2, 11.3, 12.8 and 13.6.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008

4-19

Application of Tables for Estimating Conditional Probabilities

SECTION 4

For these cases it is recommended that: • • The risk analysis team estimate is taken as the best estimate. The likely low and likely high probability is estimated by the analysis team, within the range shown in the table.

Where the Toolbox table indicates a probability of 1.0, this should generally be adopted for best estimate and likely low and likely high estimates. These are only used in the Toolbox where there is very high degree of confidence based on physical factors that a probability of 1.0 is applicable. In cases where the information available to the risk analysis team strongly supports adopting best estimate and/or likely low or likely high probabilities outside the range in the tables, the risk analysis team may adopt this value, but it would not be expected that this would result in probabilities greatly different to those estimated by the Toolbox. The factors leading to this probability estimate should be described and the reasoning for the revised estimate provided in the risk analysis report. 4.7.3.4 Modelling uncertainty for event tree nodes where single value estimates of probability are provided. This covers tables where analysis has been carried out to combine a number of input variables, and the Toolbox development team have allowed for uncertainty in the input variables to allow for uncertainty. This includes Tables 5.29 to 5.35, 6.26, 6.27, and 7.4, 7.5. i) Probability of initiation of erosion in concentrated leaks, Tables 5.29 to 5.35

As described in Sections S5.4.2.4 these tables have been developed allowing for: • Probability distributions in the maximum crack width and crack width at depth versus crack width at the surface. The initial shear stress of the soil. Using these to run Monte Carlo analyses.

• •

The values in Tables 5.29 to 5.35 are median values. Tables in Supporting Document Section S5.4.2.4 have the equivalent tables for the 10% and 90% values, representing the likely minimum and likely maximum probabilities.

4-20

Application of Tables for Estimating Conditional Probabilities

SECTION 4

ii) Probability of initiation of erosion and progression in cohesionless soils, Tables 6.26, 6.27, and 7.4, 7.5 As described in Section S6.6.2.7 these probabilities are obtained by comparing the actual average seepage gradient to that required to initiate and progress backward erosion. The probabilities were assessed by expert judgement allowing for the uncertainty in the method used to assess whether backward erosion would initiate and progress. Tables 6.26, 6.27, and 7.4, 7.5 are best estimate values. Tables in Supporting Document Section S6.6.2.7, S7.3.2 and S7.3.3 have the equivalent tables for the likely minimum and likely maximum probabilities. These have been developed by expert judgement. 4.7.3.5 Selection of the Probability Distribution The risk analysis team should select the probability distribution they believe best fits their best estimate, likely low and likely high probability estimates. The alternatives which may be considered are explained in Reclamation (2001b). For many cases a triangular distribution is likely to be suitable.

4.8

Summarizing (Making the case)

Once a risk estimate is prepared with the use of this toolbox, the analyst needs to make summary of the key factors that generated the estimates of probability of failure. An exercise of ‘making the case’ is important so that reviewers and decision makers can quickly focus on the story being told. This can be done relatively easily by reviewing each of the components of the estimate and select those that drive most of the final estimate. Once these are determined, they should be reviewed and the main factors that contribute most to the actual estimate of this component should be brought forward into an engineering summary of the failure mode. This engineering summary should normally only focus on those factors key to the overall estimate. By doing this the redundancy, or lack thereof, in the design can be demonstrated. In the case with much redundancy, i.e. dams with many components that contribute to the overall risk being low, it is important to highlight this. For example, dams with a low chance for concentrated leakage; a non-erodible core; some filtering components should be viewed as an overall robust situation with reasonable redundancy. In contrast, a homogenous dam with an erodible core where the risk is low due almost solely to a low probability of a concentrated leak is a case with little redundancy. It is important to summarize this situation for reviewers and decision makers.

4-21

Application of Tables for Estimating Conditional Probabilities

SECTION 4

4.9

Combining Probabilities

Risk is computed by finding the product of probabilities and consequences for each path of the event tree. The principles for combining the probabilities on the event trees and for the different failure paths is as follows; • For mutually exclusive failure paths, as occurs on a specific event tree, the conditional probabilities of failure should be added. For failure paths that are not mutually exclusive, the total probabilities of failure from the event trees should be calculated using De Morgan’s rule; P = 1 – (1-PIM1) x (1-PIM2) x (1-PIM3)…etc. The method for presenting the probabilities of failure will depend on which method is intended to be used for computing the risk. • If the risk is to be calculated for static, hydrological and seismic loading events separately, then refer to Section 4.9.1. If the risk is to be calculated using fragility curves, then refer to Section 4.9.2.

•

•

4.9.1

Adding Probabilities for Static, Hydrological and Seismic Loads

The following guidelines are provided for computing the annual probabilities of failure for the static, hydrological and seismic load cases. Annual Probability of Failure for the Static Load Condition; The annual probability of failure for the static load case is as follows; • For each static reservoir level partition, combine the response probabilities of failure for each failure mode using De Morgan’s rule. P = 1 – (1-PIM1) x (1-PIM2) x (1-PIM3)…etc. Section 3.5 gives guidance on selecting the reservoir level partitions that correspond to the static load case.

4-22

Application of Tables for Estimating Conditional Probabilities

SECTION 4

•

The annual probability of failure for the static load case is then calculated by summing the product of the total response probability of failure for each load partition by the annual probability of the loading condition.

Annual Probability of Failure for the Hydrological Load Condition; The annual probability of failure for the hydrological load case is as follows; • For each hydrological reservoir level partition, combine the response probabilities of failure for each failure mode using De Morgan’s rule. P = 1 – (1-PIM1) x (1-PIM2) x (1-PIM3)…etc. Section 3.5 gives guidance on selecting the reservoir level partitions that correspond to the hydrological load case. • The annual probability of failure for the hydrological load case is then calculated by summing the product of the total response probability of failure for each load partition by the annual probability of the loading condition.

Annual Probability of Failure for the Seismic Load Condition; The annual probability of failure for the seismic load case is as follows; • For each seismic load partition, combine the response probabilities of failure for each failure mode using De Morgan’s rule. P = 1 – (1-PIM1) x (1-PIM2) x (1-PIM3)…etc. Section 3.6 gives guidance on selecting the seismic level partitions. • The annual probability of failure for the seismic load case is then calculated by summing the product of the total response probability of failure for each load partition by the annual probability of the loading condition.

4.9.2

Development of Fragility Curves

The following guidelines have been developed for users who intend to use fragility curves to quantify the risk. The probabilities of failure estimated for each of the reservoir level and earthquake loading partitions can be used as point estimates for developing fragility curves. Separate fragility curves should be developed for each of the potential failure modes identified in the screening process.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008

4-23

Application of Tables for Estimating Conditional Probabilities

SECTION 4

The process to develop fragility curves for reservoir level loading is summarized as follows; • • Select the reservoir level partition points as described in Section 3.5. Estimate the conditional probabilities of failure for each of the three modes of internal erosion and for each reservoir level. The conditional probabilities for each failure path should be added as follows; P = 1 – (1-PIM1) x (1-PIM2) x (1-PIM3)…etc. • Plot the conditional probabilities vs reservoir level on a log-linear scale. Join the individual point estimates using straight lines. An example is shown in Figure 4.1.

1.0E-03 Conditional probability of failure Top of Active Conservation Flood of Record Level

Internal erosion in the foundation Internal erosion from the embankment into the foundation

1.0E-04

1.0E-05

1.0E-06

1.0E-07 80 100 120 140 160 180 200 220 240 260 Reservoir Level (ft)

Figure 4.1 – Example fragility curve

4-24

Embankment crest leve

Minimum Pool Leve

Internal erosion in the embankment

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

SECTION 5

5

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

5.1

a)

Overall Approach

Estimate the probability of a transverse crack for normal and hydrologic loading for each of the initiating mechanisms that can lead to cracking and low stress zones in which hydraulic fracture can occur. These are: Upper parts of the embankment (Section 5.2) • • • • • • • Differential settlement cross valley (IM1) Differential settlement adjacent to a cliff (IM2) Differential settlement cross valley arching (IM3) Differential settlement as a result of cross section settlement (IM4) Differential settlement in the foundation beneath the core (IM5) Differential settlements due to embankment staging (IM6) Desiccation cracking (IM7, IM8)

Lower and middle parts of the embankment (Section 5.3) • • • • Differential settlement cross valley (IM9) Differential settlement as a result of cross section settlement due to arching (IM10) Differential settlement due to soil in the foundations (IM11) Differential settlement due to small scale irregularities in the foundation profile (IM12)

This is done using Tables 5.1 to 5.15 in Section 5.2 for cracking in the upper part of the dam, and Tables 5.16 to 5.21 in Section 5.3 for cracking in the middle and lower parts. The embankment is split into the upper and middle and lower parts to take into account the likely locations for each of the crack initiating mechanisms. The inputs to these tables are modified by observed settlements or cracking using Tables 5.22 and 5.23 in Section 5.3.5. For most dams not all mechanisms will be present and those mechanisms are assigned a zero probability. The details for screening the mechanisms are described in Section 3.4. b) Estimate the maximum likely crack width at the surface of the core for each of the initiating mechanisms which apply as described in Section 5.4. Then estimate the likely crack width at the reservoir level being considered.

5-1

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

SECTION 5

c)

Estimate the probability of initiation of erosion for this mechanism given this estimated crack width, seepage gradient across the core, and the properties of the soil in the core (P IC ) using Section 5.4.

d)

Estimate the probability of initiation of erosion (P I ) = (P C ) x (P IC ) for each initiating mechanism. The probability estimates for initiation of erosion for each initiating mechanism are not added together, but are carried through the event trees for each failure path.

5.2

Estimating the Probability of Transverse Cracking (P C ) in the Upper Part of the Dam

5.2.1

Likelihood of a Transverse Crack Due to Cross Valley Differential Settlement (IM1)

Table 5.1 - Factors influencing the likelihood of cracking or hydraulic fracturing in the upper part of embankment dams - Cross Valley Differential Settlement (IM1)

Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Uniform abutment profile without benches Narrow bench very low in the abutment b/h2 <0.5, h2/h1 >1.5 Refer to Note (a) Slope of abutments under embankment (1) core Height of embankment

Notes:

Neutral (2) Wide bench low in the abutment b/h2 > 1; h2/h1 > 1

More Likely (3) Wide bench in upper half to one third of the abutment b/h2 > 1; 0.5 < h2/h1 < 1 or narrow bench in upper half to one third of the abutment, b/ h2 > 0.5; h2/h1 < 0.25

Much More Likely (4) Wide bench near the crest in the abutment b/h2 > 1. 0 < h2/h1 < 0.5

Cross valley profile under embankment core(1)

(2)

**Gentle abutment slope β 1 < 30
**

o

**Moderate abutment slopes 30 < β 1 < 45
**

o o

Steep abutments 45o < β 1 < 60o High dams 100 ft to 200 ft (30 m to 60 m)

Very steep abutments, β 1 > 60o Very high dams >200 ft (60 m) Refer to Note (b)

(1)

Dams less than 50 ft (15 m) high

Dams 50 ft to 100 ft (15 m to 30 m) high

(a) See Figure 5.1 for definitions of b, h1, h2, β. (b) For dams higher than 400 ft (120 m) assign a likelihood factor of 5. (c) The method of applying this type of table is described in Section 4.4.

5-2

007] 0.002 [0.Definition of terms used to describe cross valley geometry A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-3 .00005 0.005 0.0005 0.0005] [0.2 .00001 0.2 Above POR 6 9 11 13 18 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot Valley Centerline Dam crest h2 H b β1 Abutment h1 β2 Figure 5.1 .0001 0.02 Below POR 0.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.00015 0.Probability of a cracking or hydraulic fracture in the upper part of embankment (Relative importance factor (RF) x dams-Cross Valley Differential Settlement versus ∑ (Likelihood factor(LF)) 0.05 0.

5 Note. probability=zero (1) Dams less than 50 ft (15 m) high Dams 50 ft to 100 ft (15 m to 30 m) high High dams 100ft to 200 ft (30 m to 60 m) Very high dams >200 ft (60 m) Height of embankment Note: (a) See Figure 5.25 Cross valley profile under embankment core(a) (b) Slope of abutments under embankment core(a) Moderate abutment slopes 25 < β 1 < 45 o o Steep abutments 45o < β 1 < 60o Very steep abutments.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.4 . β1 (b) This mechanism only applies for Wb/Hw< 2.0 < Wb/Hw <2.02 0. probability=zero (2) Gentle abutment slope β 1 < 25 o Neutral (2) Bench adjacent to cliff 1.002 0.5 Table 5.Probability of a cracking or hydraulic fracture in the upper part of embankment(Relative Differential settlement adjacent a cliff at the top of the embankment versus ∑ importance factor (RF)) x (Likelihood factor(LF)) negligible negligible negligible negligible negligible negligible negligible [0.25 < Wb/Hw < 1.Factors influencing the likelihood of cracking or hydraulic fracturing in the upper part of embankment dams-Differential settlement adjacent a cliff at the top of the embankment (IM2) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Wide bench Wb/Hw > 2.02 Below POR 0.3 . If this condition is present. Wb/Hw < 0. If this condition is present. Hw.005] 0.0005] negligible [0.2 for definitions of Wb.2 Above POR 6 9 11 13 14 19 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-4 .0 Much More Likely (4) No or very narrow bench adjacent to cliff. β 1 > 60o Note.5 More Likely (3) Narrow bench adjacent to cliff 0.

5. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-5 .Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 β1 Figure 5.2 .Cracking or hydraulic fracture adjacent cliffs due to differential settlement of the embankment. Note that this mechanism only applies for Wb/Hw< 2.

05 0. H.5 .001 0.Factors influencing the likelihood of cracking or hydraulic fracturing in embankment due to cross valley arching (IM3) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Moderate abutment slope β 1 . and the upper part is not.3 for definitions of W v .6 .β 2 < 45o Neutral (2) Moderate steep abutment slopes 45o<β2.Probability of cracking or hydraulic fracturing in embankment due to cross valley arching versus (Relative importance factor (RF)) x (Likelihood factor(LF)) ∑ negligible negligible 0.00005 negligible negligible 0.0005 0. β 1 . Table 5. β 2 60o Cross valley geometry under embankment core(a) (2) W v /H > 0.4< W v /H<0.2 Above POR 6 9 10 13 17 21 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-6 .004 0.0005 0. β 1 and β 2 (b) If the soil in the lower part of the core is poorly compacted or subject to collapse compression on saturation.75 Note. β2<60o More Likely (3) Steep abutments 60o < β 1 . If W v /H > 2.25< W v /H< 0. Exclude this failure mode Height of embankment (1) Dams less than 50 ft (15 m) high Dams 50 ft to 100 ft (15 m to 30 m) high High dams 100 ft to 200 ft (30 m to 60 m) Very high dams >200 ft (60 m) 0.0001 0.007 0.02 Below POR 0.β 2 < 75o Much More Likely (4) Very steep abutments.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.β 2 > 75o β 1 near vertical.4 Very narrow deep valley W v /H < 0.25 Slope of abutments under embankment core(a) Note: (a) See Figure 5.75 Narrow deep valley 0. increase weighted factor (LF x RF) by 1 or 2.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-7 .Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 β2 β1 β2 β1 β2 β1 β2 β1 Figure 5.3 – Longitudinal profiles of the dam showing the definition of terms for cross valley arching.

uncompacted (dumped) rockfill EITHER Embankment zoning and Compaction of outer zone Excluding sloping core earth and rockfill dams OR Embankment zoning and Compaction of outer zone. Rockfill (or “gravel fill”). earthfill with filter drains or homogeneous (all materials similar modulus) Note: if have evidence from relative settlements of core and shoulders that the materials have a similar modulus.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 5.well compacted. and core .7 – Factors influencing the likelihood of cracking or hydraulic fracturing in the upper part of embankment .well compacted.2. with well compacted shoulders and core. rockfill or gravel” fill compacted by dozer tracking or by small rollers in thick layers Much More Likely (4) Central core earth and rockfill.2 Likelihood of Transverse Cracking Resultant on Cross Section Settlement due to Poorly Compacted Shoulders (IM4) Table 5. then probability = 0 (3) Core sloped o steeper than 45 but within limits of sloping core embankment. rockfill or gravel” fill compacted by dozer tracking or by small rollers in thick layers Core sloped o flatter than 45 . Core sloped o flatter than 45 . and with similar moduli. and core . earthfill with filter drains. More Likely (3) Central core earth and rockfill (or “gravel fill”). Construction staged with rockfill in the lower part of the dam compacted to a higher modulus than (a) the upper part. sloping core earth and rockfill dams Core sloped o flatter than 45 Rockfill (or “gravel fill”). modulus of outer zones lower than core.cross section settlement resulting from poorly compacted shoulders (IM4) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Zoned earthfill. Rockfill (or “gravel fill”). Neutral (2) Zoned earthfill. Central core earth and rockfill (or “gravel fill”). OR uncompacted (dumped) rockfill A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-8 .

02 and 0.2 may be applied.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Likelihood Factor (LF) Factor Relative Importance Factor (RF) (2) Less Likely (1) W/H>3 Neutral (2) 1.Probability of a transverse crack or hydraulic fracture resultant on cross section settlement resulting from poorly compacted shoulders versus (Relative importance ∑ factor (RF)) x (Likelihood factor (LF)) 0.00002 0.4 for definition of width W and Height H.5< W/H<1. See Figure 5.0001 0.5 Core geometry Width (W)/ Height (H)(b) Height of embankment Note.00001 0.5<W/H<3 More Likely (3) 0. Table 5.02 and 0. a probability of cracking between 0.007] 0.5 Much More Likely (4) W/H<0.0002 0.0005 0.00005 0.0005] [0.02(1) Above POR 6 Note: 9 11 13 18 24 RF x LF (1) If there is a large difference in modulus between the core and shoulders.8 . Width should be taken at the base of the core.002 Below POR 0.002 [0.2 may be applied. a probability of cracking between 0. Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-9 . (a) (b) (1) Dams less than 50 ft (15 m) high Dams 50 ft to 100 ft (15 m to 30 m) high High dams 100 ft to 200 ft (30 m to 60 m) Very high dams>200 ft (60 m) If there is a large difference in moduli.0002 0.

(b) Limit of what constitutes a sloping core dam.4 .Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 core slope β1 Figure 5. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-10 .Sloping core dam (a) Definitions of terms.

9 – Factors influencing the likelihood of cracking or hydraulic fracturing in the upper part of embankment dams. See Figure 5.settlement resulting from differential settlements in soil in the foundation (IM5) Relative Importance Factor (RF) (3) Likelihood Factor (LF) Less Likely (1) Rock foundations or uniform soil foundations(a) Neutral (2) Shallow soils or soils with gradual variation in depth and compressibility sufficient to cause differential settlement of less than 0.2.5 for typical scenarios which may lead to differential settlement.5% of the embankment height Steep o Factor Much More Likely (4) Deep compressible soil in the (b) foundation sufficient to cause differential settlement of >0.5% of embankment height Foundation geology and geometry (c) Slope of the sides of the compressible zones (d) Height of embankment (2) Gentle Moderate o Very steep o α < 30 (1) 30 < α < 45 o 45 < α < 60 o α > 60o Very high dams > 200 ft (60m) high Dams less than 50 ft (15 m) high Dams 50 ft to 100 ft (15 m to 30 m) high High dams 100 ft to 200 ft (30m to 60 m) high Notes: (a) (b) (c) (d) If there is no compressible soil in the foundation this mode does not apply Including soils which collapse on saturation and which have not been treated or removed during construction.5 for definition of slope α A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-11 .2% to 0.2% of the embankment height More Likely (3) Moderate depth of compressible soil in the foundation sufficient to cause differential settlement of 0. See Figure 5.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 5.3 Likelihood Of Transverse Cracking Due To Differential Settlements In Soil In The Foundation Beneath The Core (IM5) Table 5.

00005 negligible negligible 0. Note: (1) This mode does not apply if there is no compressible soil in the foundation beneath the core.0005 0.0002 0.Probability of a crack or hydraulic fracture due to differential settlement in the (Relative importance factor (RF)) x (Likelihood factor (LF)) foundation versus ∑ negligible negligible 0.5 – Typical scenarios which may lead to differential settlement in the foundation.03 0.02 Below POR 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-12 . Figure 5.003 0.002 [0.2 Above POR 6 8 9 11 13 18 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.0005] [0.10 .007] 0.

If the reservoir stage being considered is below the likely depth of desiccation cracking. In most cases the latter will apply. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-13 .Longitudinal section through staged embankment 5.13.6 .2. SECOND For cases where the reservoir stage is above the base of potential desiccation cracking. There is a potential for differential settlement to occur if the “existing” (first stage) embankment is a significantly higher modulus than the remainder of the embankment. If this is the situation.5 Likelihood of Transverse Cracking due to Desiccation (IM7. follow the procedure below. IM8) Desiccation In The Crest Of The Dam (IM7) FIRST Consider the maximum likely depth of desiccation cracking for the soil in the core of the dam. use the method described in Section 5. the probability of a crack due to desiccation cracking can be assumed to be zero. and the climate using Table 5.2.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 5. this mode may be ignored. Final embankment crest level h2 b Existing Embankment h1 Figure 5.4 Likelihood Of Transverse Cracking Due To Differential Settlements Due To Embankment Staging (IM6) This situation arises where the embankment has been staged during construction.1 to assess the likelihood a crack or hydraulic fracture will result. If however there is no or little difference in the modulus.2.

less than 10 inches (250 mm) rainfall. Table 5. High summer temperatures High plasticity Much More Likely (4) No surface layer.9 6 9 11 16 20 24 RF x LF A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-14 . Seasonal climate with annual rainfall greater than 20 inches (500 mm) and no prolonged hot dry periods Medium to low plasticity More Likely (3) Thin ( less than 3 inches (75 mm)) surface gravel layer with no pavement (a) cover or Low plasticity granular transition layer over core Monsoonal or other distinct wet and dry periods in the year.0001 0. with non-plastic granular material.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.11 – Factors influencing the likelihood of cracking in the upper part of embankment dams. uniform rainfall throughout the year Neutral (2) No road pavement (a) cover . asphalt or bitumen seal.Probability of a transverse crack. 6 inches to 12 inches (150 mm to 300 mm) thick.001 0.12 . cracking in the crest due to desiccation by drying versus (Relative importance factor (RF)) x (Likelihood factor(LF)) ∑ 0. Summer maximum temperatures>85°F (>30°C) Medium to high plasticity Arid climate.1 0.01 0. and/or Rock fill or non plastic granular layer at least 3 ft (1 meter) thick (2) Temperate climate. Dam core extends to crest level Crest zoning and surface layer over core Climate Plasticity of core material (1) Low plasticity to non plastic Note: (a) Road pavement cover may comprise concrete.5 0.cracking in the crest due to desiccation by drying (IM7) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Road pavement (a) cover with base layer 300 mm or more thick.

5 Note: (a) Road pavement cover may comprise concrete. Below that any desiccation cracks should have swelled and closed.15. SECOND Where the mechanism applies. and multiply this probability by the assessed likelihood the layer will be continuous across the core. less than 10 inches (250mm ) rainfall.14 and Table 5.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5. If the seasonal shutdown layer is below the Pool of Record. Desiccation Cracking On Seasonal Shutdown Layers And On The Surface Of Staged Embankments (IM8) FIRST (a) This mechanism only applies above the level of saturation of the core.15. asphalt or bitumen seal.13 – Screening Tool. or the embankment has been staged.14 are to be assessed according to the conditions across the width of the core. high summer temperatures Monsoonal or other distinct wet and dry periods in the year. estimate the probability of desiccation cracking using Table 5.5 Maximum likely depth with no surface layer (dam core extends to crest level) (feet) 23 (meters) 7 Climate 13 4 20 6 10 3 16 5 6 2 15 4. Summer maximum temperatures >85°F (>30°C ) Seasonal climate with annual rainfall greater than 20 inches (500 mm) and no prolonged hot dry periods Temperate climate. Maximum likely depth of desiccation cracking for a gravel surface layer with no road pavement cover based on climate Maximum likely depth with gravel layer with no road pavement cover(a) (feet) Arid climate. (b) This mechanism only applies where there has been a seasonal shutdown during construction. uniform rainfall throughout the year 15 (meters) 4. The descriptions in Table 5. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-15 . use the “Below POR” probabilities in Table 5.

less than 10 inches (250 mm ) rainfall. surfaces scarified. poor moisture control practices Construction practices regarding clean-up of desiccated layers after construction shutdowns or the surface of the earlier stage of the dam Climate Seasonal climate with annual rainfall greater than 20 inches (500 mm) and no prolonged hot dry periods Arid climate. regardless of the other factors (2) Temperate climate. but depth of scarifying insufficient or difficulties with moisture control Monsoonal or other distinct wet and dry periods in the year. No attempt to scarify or remove desiccated layers. Desiccated layers removed from embankment and replaced with new soil or adequately reworked to specified moisture content. Summer maximum temperatures >85°F (>30°C ) Plasticity of core material (1) Low plasticity to non plastic (LL < 20%) Medium to low plasticity (20% < LL < 40%) Medium to high plasticity (40% < LL < 50%) High plasticity (LL > 50%) Much More Likely (4) Poor control. High summer temperatures A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-16 . surface re-compacted. More Likely (3) Moderate control. moisture adjusted to specified range. If this condition is present.14 .Factors influencing the likelihood of cracking on seasonal shutdown layers during construction and staged construction surfaces due to desiccation by drying (IM8) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (4) Less Likely (1) Very good control and clean-up practices.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5. Attempts to scarify desiccated layers. probability=0. uniform rainfall throughout the year Neutral (2) Good control and practices.

01 0.001 0.1 0.Probability of a transverse crack on seasonal shutdown layers during construction and staged construction surfaces due to desiccation by drying versus (Relative importance factor (RF)) x (Likelihood factor(LF)) ∑ negligible negligible negligible negligible negligible 0.01 0.001 0.15 .9 Above POR 7 10 12 13 18 22 28 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-17 .Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.1 Below POR 0.0001 negligible 0.

17 . 1 < h2/h1 < 1. 1. 45o < β2 < 60o High dams 100 ft to 200 ft (30 m to 60 m) Much More Likely (4) Wide bench in lower half of the abutment b/h2 > 1. β2 < 30 (1) o Neutral (2) Minor bench in lower half of the abutment.001 0. not persistent across the core Moderately abutment slopes.5.3.01 0.005 Below POR 0.Factors influencing the likelihood of cracking or hydraulic fracture in the middle and lower parts of embankment dams-cross valley differential settlements (IM9) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Uniform abutment profile without benches Gentle abutment slope.5 < h2/h1 < 2 Steep abutments.5 Very steep abutments.001] 0.16 .0001] [0. β2 > 60o Very high dams >200ft (60 m) Cross valley profile under embankment core Slope of the abutments under embankment core Height of embankment (2) Dams less than 50 ft (15 m) high Note: See Figure 5.Probability of a transverse crack or hydraulic fracture in the middle and lower parts of embankment dams due to cross valley differential settlements versus (Relative ∑ importance factor (RF)) x (Likelihood factor (LF)) negligible negligible negligible negligible negligible negligible [0.05 Above POR 6 9 12 13 18 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-18 .3 Estimating The Probability Of Transverse Cracking Or Hydraulic Fracture (P C ) In The Middle And Lower Parts Of The Dam Likelihood of Transverse Cracking or Hydraulic Fracture Due To Cross Valley Differential Settlement (IM9) 5. h1. h2.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 5. 30o < β2 < 45o Dams 50 ft to 100 ft (15 m to 30 m) high More Likely (3) Bench in lower half of the abutment b/h2 > 0.1 for definitions of b. Table 5. β2.1 Table 5.

at the phreatic surface. or more than 3% dry of OWC and < 95% (b) SMDD Very high dams> 200 ft (60 m) Height of embankment Notes: (1) Dams less than 50 ft (15 m) high (a) The most likely location for arching to occur is in the upper to middle part of the dam.18 – Factors influencing the likelihood of cracking or hydraulic fracturing in the middle and lower parts of embankments .25 Factor Core geometry Width (W)/Height (H) Relative stiffness of core and shells (2) Core has higher modulus than shells. It may also lead to softened zones and even a crack in the vicinity of the contact between the saturated and unsaturated parts of the core.2 Likelihood Of Transverse Cracking Or Hydraulic Fracture Due To Differential Settlement Causing Arching Of The Core Onto The Shoulders Of The Embankment (IM10) This mode is applicable to central core earth and rockfill (or gravel shells) dams and puddle core earthfill dams.e. Core compacted to >98% SMDD at a moisture content between -2% and +1% of standard OWC Dams 50 ft to 100 ft (15 m to 30 m) high Core lower modulus than outer stiffness. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-19 . Shoulders well compacted and high modulus. Shoulders poorly compacted or dumped.25< W/H<0.5 Much More Likely (4) Very narrow core.0 (c) Neutral (2) 0. 0. Table 5. It is not applicable to all other the following dam types (including dams with a sloping core). i.0 More Likely (3) Narrow core.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 5. Shoulders well compacted and high modulus. W/H<0. Core compacted (c) >98% SMDD Modulus of core same or marginally lower than shoulders. (b) In core materials compacted dry of optimum moisture and to a density ratio less than about 95%.5<W/H<1.settlement resulting from arching of the core onto the shoulders (IM10) Relative Importance Factor (RF) (3) Likelihood Factor (LF) Less Likely (1) W/H>1. This may lead to arching and low stresses.3. Either core compacted 0% to 2% wet of OWC and between 95% and 98% SMDD or 2% to 3% dry of OWC and < (b) 95% SMDD High dams 100 ft to 200 ft (30 m to 60 m) Core much lower modulus than outer shoulders or subject to collapse compression. collapse compression of the core may occur. Shoulders well compacted and high modulus Either core compacted > 2% wet of OWC.

19 . Table 5. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-20 .settlement resulting from arching of the core onto the (Relative importance factor (RF)) x (Likelihood shoulders of the embankment versus ∑ factor (LF)) negligible negligible negligible negligible [0.01 0.3. Finite element analyses which properly model the history of the dam.3.4 Likelihood of Transverse Cracking Or Hydraulic Fracture At The Foundation Contact Due To Small Scale Irregularities In The Foundation Profile Under The Core (IM12) Small scale irregularities in the foundation profile under the core may comprise steps.05 Above POR 6 9 11 12 13 18 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot. 5.3 Likelihood Of Transverse Cracking or Hydraulic Fracture Due To Differential Settlement In The Foundation Under The Core (IM11) Refer to Section 5. W = width of the core at depth H below the crest of the embankment. 5.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 (c) (d) (e) (f) The likelihood of arching is negligible for embankments with these characteristics The mechanism is considered to be significant only for reservoir levels above the pool of record. Larger scale features are covered in Section 5. Small scale irregularities include those features which have heights less than 10% of the embankment height.Probability of a transverse crack or hydraulic fracture in the middle and lower parts of embankment dams. The probability for this initiating mechanism is assumed to be negligible for the Below POR cases.001] 0. benches or depressions in the foundation rock.3.3 which covers this mode as well as cracking and hydraulic fracture in the upper part of the embankment.2. and its properties including collapse on saturation may be used to assess the likelihood of cracking or hydraulic fracture.2.

benches. (b) An irregularity with less than 50% persistence across the core is assumed to have a negligible contribution to the probability of a transverse crack.002 0.01 Below POR 0. depressions in rock foundations 3% to 5% of the embankment height Narrow core. (2) The probability of a transverse crack across the core is negligible for persistence less than 50% Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-21 . depressions in rock foundations less than 3% of the embankment height 0.5 Neutral (2) Persistent 50% to 75% across the core width Steps.5<W/H<1.5 Much More Likely (4) Persistent 90% to 100% across the core Steps. benches.1 Above POR 6 Notes: 9 11 12 13 18 24 RF x LF (1) Larger irregularities are covered in Section 5.002 [0. W/H<0. benches.3.Probability of a transverse crack or hydraulic fracture in the middle and lower parts of embankment dams due to small scale irregularities in the foundation profile under the core versus (Relative importance factor (RF)) x (Likelihood factor (LF)) ∑ Negligible Negligible negligible negligible negligible 0.21 . Table 5.25 Persistence of the irregularity across the core Small scale irregularities in abutment profile (2) Core geometry Width (W)/Height (H) Notes: (1) (a) Larger irregularities are covered in Section 5.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.0007] [0.2. W/H>1.2.0002 negligible 0. 0.5 More Likely (3) Persistent 75% to 90% across the core Steps. or irregularities treated by slope modification Wide core.20 – Factors influencing the likelihood of cracking or hydraulic fracturing in the middle and lower parts of embankment dams due to small scale irregularities in the foundation profile under the core (IM12) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Persistent across less than 50% of the (b) core Uniform abutment profile.3.005] 0.25< W/H<0. depressions in rock foundation 5% to 10% of the embankment (a) height Very narrow core.02 0.

The maximum multiplier should be less than 10 times and should not “double up” on the factors listed in Table 5. For dams which have experienced settlements larger than the averfage population of that class of dam the probabilities of cracking or hydraulic fracture will be increased. and Sections 5.3.22.3. Select the likelihood column in Table 5.22. The probability from the relevant table is multiplied by the factor from Table 5.22 which corresponds to the maximum settlement measured anywhere in the embankment expressed as a ratio of the maximum embankment height.22 depending on whether the dam has poorly compacted rockfill shells or not.3.Factors To Account For Observations And Measured Settlements Settlement Factors Where there are settlement observations for the dam these can be used to modify the results of Sections 5.1 to 5. If there are significant differential settlements across the valley greater than expected from the mechanisms present.2. allowing for the quantity and quality of the data and the relative improtance of the observations.2. Then obtain the settlement multiplication factors for the upper and middle and lower parts from the corresponding bottom four rows of Table 5. The multiplication factor should be selected taking account of what data is available. but lower corrections apply for the latter. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-22 .5 Probability Of Transverse Cracking or Hydraulic Fracture .3. and middle and lower part of the embankment. an additional increase in probabilities may be applied.22 applies to the assessment of transverse cracking in the upper part. Table 5.3. and for those which have experienced settlements smaller than the average population of that class of dams the probability of cracking or hydraulic fracture will be reduced.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 5.1 to 5.

Long term settlement rates(% per log time cycle in years) dams with poorly compacted shoulders .5% Neutral More Likely Much More Likely <0.0 2 to 5 5 to 10 Notes: (a) Multiplication factors to be applied to Probabilities from Sections 5.2 0.5% 0.1.5% to 3% 0.5% 2 to 5 0.Settlement multiplication factors versus observed settlements Influence on Likelihood Factor Less Likely Observed maximum settlements as percentage of embankment height . (b) Includes dumped rockfill.22 .Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.2 > 0.2 to 0.1% 0.Post construction crest settlement at 10 years after construction other dams .25% < 0.4% 0.2 and 5.3.5% 1.5% 0.2.2 to 0.5% to 1.5% to 1% 0.3.0 2 to 10 10 to 20 0.15% > 1% > 0.0% to 1.0% 0.5 3% to 4% 1. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-23 .7% 0. 5.0 > 4% > 1.2.2 to 0.1.25% to 0.Long term settlement rates(% per log time cycle in years)-other dams Settlement multiplication factors for cracking or hydraulic fracture in the upper part (a) of the embankment based on observed maximum settlements Settlement multiplication factors for cracking or hydraulic fracture in the middle and lower parts (c)(d) of the embankment Dams with poorly compacted rockfill (b) All other dams Dams with poorly compacted rockfill (b) All other dams < 1.3.25% to 0. and rockfill and other granular zones compacted by tracking with bulldozers and by small rollers in thick layers (c) To be applied to probabilities from Sections 5.5% <0.3.5 1.4% to 0.7% < 0.1% to 0. 5.25% 0.05 to 0.5 1.2.2 and 5.15% to 0.3 (d) Multiplication factors assumed to be half those for cracking in the upper part.5 1.0 2 to 5 0.Post construction crest settlement at 10 years after construction dams with poorly compacted shoulders .Core settlement during construction .5% 1.

The multiplication factor should be selected taking account of what data is available.5 to 0. core covered with road pavement or other granular material 1.5.2. Cracking in the middle and lower parts will generally not be observed. Neutral No test pits More Likely Transverse cracks persistent across the top of the core and/or. no test pits No cracking observed. Table 5. open longitudinal cracking Much More Likely Transverse cracks which pits show persist across the core. This factor only applies to the assessment of transverse cracking in the upper part of the embankment. extensive. For dams which display cracking the probabilities will be increased.Cracking observation factors (applies to upper embankment only) Influence on Likelihood Factor Less Likely Cracking observed in test pits to the top of or into the core Cracking Factor (A) No cracking observed when large areas of the top of the core are exposed. allowing for the relative importance of the observations.0 Transverse cracks which persist across the crest and/or. narrow longitudinal cracking 2 to 5 depending on and whether they are in locations in which cracking might be expected Cracking in the surface of the crest. For those which do not display cracking the probabilities may remain the same. careful inspection for cracking 1. and extend below reservoir water level in the reservoir level partition being considered Probability of transverse crack = 1.23 .2 depending on the quality of exposure and whether they are in locations in which cracking might be expected 2 to 20 depending on the (2) width of cracking and whether they are in locations in which cracking might be expected Notes: (1) (2) Apply either Cracking Factor (A) or Cracking Factor (B). Cracking Factor (B) 0.0 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-24 .Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Observation of cracking factor Where there are observations of cracking for the dam these can be used to modify the results of Sections 5. or are reduced depending on how extensive the investigations to locate cracking have been.1 to 5. Evaluate the Cracking Factors (A) and (B) from Table 5. wide longitudinal cracking. extensive.23 and then multiply the largest value of (A) or (B) to the probabilities of transverse cracking in the upper part of the dam. The probability is multiplied by the factor from Table 5. extensive. core exposed on the surface.2.0 5 to 100 depending on width(2) of cracking and whether they are in locations in which cracking might be expected Narrow (<10mm) transverse cracks persistent across the crest and/or. whichever gives greatest probability of cracking The greater the crack width the more likely it represents cracking in the core.1 depending on the extent of exposure and how relevant the exposure is to the possible mechanism of cracking No cracking observed.23.5 to 0.

29 to Table 5. The width C d 1 at depth d 1 = C max ((D. The width required is the width at the mid level of the flow path for the reservoir stage under consideration. the crack is assumed to be uniformly tapered from the dam crest to the base of the crack.4 Estimation Of The Probability Erosion Will Initiate In A Crack Or Hydraulic Fracture In An Embankment (P IC ). soils should be assumed dispersive regardless of laboratory test results. Estimate likely crack width at the reservoir level stage under consideration using Table 5. Based on the soil classification. These tables only apply to soil compacted to 95% to 98% of Standard Proctor maximum dry density at a moisture content between -1% to +2% of optimum moisture content. While reservoir water salinity will affect dispersion. soils A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 • • 5-25 . sinkholes and tunnelling.24. That is.7 it is the average of C d 1 and C d 2 .4.26 . These tables assume the relationship between erosion rate index and soil classification shown in Table 5. The gradient required is the average gradient at the mid level of the flow path for the reservoir stage under consideration.25 and Figure 5.7. In Figure 5.35. Use approximate interpolation between values where necessary.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 5. Where there are signs of dispersive soils in field performance.g. dispersivity. • Estimate the likely crack width from hydraulic fracture at the reservoir level stage under consideration using Table 5. e. and Table 5. For saturated soils. Dispersive soils are soils with Sherard Pinhole test D1 or D2.26). severe gully erosion.d 1 )/L.7 it is (d 2 .27. In Figure 5. the salt content of the reservoir water will in most cases reduce with flood inflows and unless laboratory testing is carried out to assess the initial shear stress with the same reservoir water salts content it should be assumed soils which test dispersive in the laboratory will be so in the dam. depending on the gradient of flow through the crack. and estimated crack width (the larger of the estimates from Table 5.d 1 )/D). estimate the probability of initiation of erosion in the crack and/or hydraulic fracture (P IC ) using Table 5.25. Estimate the gradient of flow through the crack using Figure 5.7.1 (a) For cracking in the upper part of the dam the method to be followed is: • • Estimate maximum likely width of cracking at the top of the core from Table 5. • Assess the soil classification and whether the soils are dispersive. Overall Approach 5.

• If there are Hole Erosion Tests available use the method detailed in (c) below (b) For cracking in the middle and lower parts of the dam The method to be followed is: • Estimate maximum likely width of cracking in the middle or lower part of the dam from Table 5. the salt content of the reservoir water will in most cases reduce with flood inflows and unless laboratory testing is carried out to assess the initial shear stress with the same reservoir water salts content it should be assumed soils which test dispersive in the laboratory will be so in the dam. Assess the soil classification and whether the soils are dispersive. e.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 significantly dry of optimum moisture content. For saturated soils. and the method detailed in (c) below followed. Based on this comparison estimate the probability of initiation • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-26 . Based on the soil type and estimated crack width. • • (c) Procedure to be followed where Hole Erosion Test data is available In cases where Hole Erosion tests are available for the dam core soil. the following procedure should be followed: • • Estimate the crack width and hydraulic flow gradient as detailed in (a) or (b). soils should be assumed dispersive regardless of laboratory test results.29 to Table 5.36. depending on the gradient across the crack. whichever is applicable. Hole Erosion Tests should be carried out to determine the initial shear stress. Hole Erosion Tests should be carried out to determine the initial shear stress. sinkholes and tunnelling. estimate the probability of initiation of erosion in the crack and/or hydraulic fracture (P IC ) using Table 5.35. While reservoir water salinity will affect dispersion.29 to Table 5.g. Where there are signs of dispersive soils in field performance. and the method detailed in (c) below followed. AND Compare this hydraulic shear stress to the initial shear stress of the soil at the compaction and moisture conditions it exists in the core. This includes hydraulic fracture. severe gully erosion. soils significantly dry of optimum moisture content. EITHER Calculate the hydraulic shear stress in the crack for the reservoir stage under consideration using Table 5. These tables only apply to soil compacted to 95% to 98% of Standard Proctor maximum dry density at a moisture content between -1% to +2% of optimum moisture content.35. Dispersive soils are soils with Sherard Pinhole test D1 or D2.28. If there are hole erosion tests available assess which classification should apply to best reflect this value when using Table 5. and poorly compacted soils. and poorly compacted soils. Use approximate interpolation between values where necessary.

• OR/AND Use Table 5.14 2 5 20 50(2) 75(3) 2 5 20 50(2) 75(3) 1 20 50(2) 100(4) 150(6) 1 20 50(2) 75(3) 100(4) 1 10 25(1) 37(1.4.5) 50(2) 1 9 to 11 20 11 to 13 50(2) 13 to 18 75(3) 18 to 24 100(4) ∑ Crack formation mechanism A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-27 . In doing this calculation take account of the uncertainty in the crack width.11 Desiccation cracking on seasonal shutdown layer Table 5.2 Details Of The Method Table 5.21 Cross section settlement due to poorly compacted shoulders Table 5.1 Differential settlement adjacent to a spillway or other wall or cliff Table 5.4.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 of erosion.7 Differential settlements in the soil foundation Table 5.2. and initial shear stress detailed in Section S5.37 to determine which of Tables 5.9 Desiccation cracking at the crest by drying Table 5.4.35 best fits the initial shear stress of the soil tested in the HET and use that table to estimate the probability of initiation of erosion.3 Table 6. 5.24 .29 to 5.Maximum likely width of cracking at the dam crest versus (Relative importance factor) x (Likelihood factor) for cracking in the upper part of the dam Maximum likely crack width at the dam crest in millimeters (inches)relative to (relative importance factor) x (likelihood factor) 6 to 9 Cross valley differential settlement Table 5.

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

SECTION 5

Table 5.25 - Likely crack width at the depth shown versus maximum crack width at the dam crest determined from Table 5.24 for cracking in the upper part of the dam.( Depths in feet and meters)

Likely crack width at depth shown, crack width in millimeters (inches) 5 feet 1.5 meters 1 2 20 40(1.6) 60(2.4) 210(8) 1 5 20 35(1.4) 180(7) 1 5 15 140(5.4) 2 7 110(4.4) 3 90(3.5) 1 60(2.4) 10 feet 3 meters 15 feet 4.5 meters 20 feet 6 meters 25 feet 7.5 meters

(1)

Maximum crack width at dam crest Inches 0.5 1 2 3 4 10 Millimeters 10 25 50 75 100 250

30 feet 10 meters

Note. (1) Check potential crack width resulting from hydraulic fracture and use the larger of the crack widths from this table and Table 5.26 .

Table 5.26 – Examples of estimated maximum depths below the dam crest and widths of cracks formed by potential hydraulic fracture for cracking in the upper part of the dam (A) Embankment abutment

Ratio of depth of zero stress to embankment height at the abutment <0.01 0.02 0.12 0.12 0.09 0.10 0.35 Ratio of approximate maximum depth at which hydraulic fracture may occur to embankment height at the abutment 0.05 0.05 0.3 0.3 0.25 0.25 0.5

Abutment slope degrees

Ratio of bench width to embankment height 0.67 No bench No bench 0.2 0.4 1.0 No bench

Likely width of crack formed by hydraulic fracture in millimeters

15 25 45 45 45 45 60

2 2 5 5 5 5 10

5-28

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

SECTION 5

(B) Adjacent to walls and cliffs in the upper part of the embankment

∑

(Relative importance factor (RF)) x

(Likelihood factor (LF)) from Table 5.3 14 to 19 >19

Likely width of crack formed by hydraulic fracture in millimeters 5 10

(C) Cross valley arching in the upper part of the embankment

∑

(Relative importance factor (RF)) x

(Likelihood factor (LF)) from Table 5.5 10 to 13 14 to 19 >19

Likely width of crack formed by hydraulic fracture in millimeters 2 5 10

**(D) Over low stress zones over irregularities in the foundation
**

Assume that hydraulic fracture may persist from the crest to the foundation level, and the width is 5 mm.

5-29

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

SECTION 5

Table 5.27 Representative erosion rate index (IHET) versus soil classification for non dispersive soils based on Wan and Fell (2002, 2004)

Soil Classification SM with <30% fines SM with > 30% fines SC with < 30% fines SC with >40% fines ML CL-ML CL CL-CH MH CH with Liquid Limit <65% CH with Liquid Limit > 65%

Note. (1) Use best estimate value for best estimate probabilities. Check sensitivity if the outcome is strongly dependent on the results. (2) For important decisions carry out Hole Erosion Tests, rather than relying on this table which is approximate (3) The Representative Erosion Rate index is for soils compacted to 95% standard (Proctor) maximum dry density at optimum moisture content. (4) See Supporting Information Report for information regarding the Representative Erosion Rate index for soils which are significantly drier than optimum moisture content or saturated.

Representative Erosion Rate Index (IHET) Likely Minimum 1 <2 <2 2 2 2 3 3 3 3 4 Best Estimate <2 2 to 3 2 to 3 3 2 to 3 3 3 to 4 4 3 to 4 4 5 Likely Maximum 2.5 3.5 3.5 4 3 4 4.5 5 4.5 5 6

5-30

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

SECTION 5

Table 5.28 - Maximum likely width of cracking in the dam versus

∑

(Relative importance

**factor) x (Likelihood factor) for cracking in the middle and lower parts of the dam
**

Crack formation mechanism Cross valley differential settlement Table 5.16 Differential settlement causing arching of the core onto the shoulders Table 5.18 Differential settlement over small scale irregularities in the foundation Table 5.20 Differential settlements due to settlements in the foundation Table 5.9 Differential settlement causing arching of the core in the cut off trench Table 9.3 0 0 0 to 1 1 to 2 2 to 10 0 0 0 to 2 2 to 10 10 to 20 0 0 0 to 1 1 to 2 2 to 10 0 0 0 to 1 1 to 2 2 to 10 Maximum likely crack width in millimeters relative to (relative importance factor) x (likelihood factor) 6 to 9 0 9 to 11 0 11 to 13 1 to 2 13 to 18 2 to 10 18 to 24 10 to 20

5-31

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

SECTION 5

Figure 5.7 Example of the estimation of crack width and flow gradient in the crack.

5-32

Probability of Initiation of Erosion in Transverse Cracks in the Embankment

SECTION 5

**Table 5.29 - Estimation of probability of initiation in a crack for ML or SM with <30% fines soil types
**

Estimated likely crack width in core for Reservoir stage being considered (mm) 1 2 5 10 25 50 75 100 Probability of initiation of erosion for different seepage gradients Average Hydraulic Gradient 0.1 0.05 0.1 0.6 0.9 1.0 1.0 1.0 1.0 0.25 0.2 0.6 1.0 1.0 1.0 1.0 1.0 1.0 0.5 0.6 0.9 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.95 1.0 1.0 1.0 1.0 1.0 1.0 1.0 2.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 5.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

Note. (1) The gradient is the average hydraulic gradient from the upstream to the downstream of the core at the level of the assumed crack under the reservoir level under consideration. No allowance is made for seepage head losses in the zones upstream or downstream of the core.

Table 5.30 - Estimation of probability of initiation in a crack for SC with <40% fines, or SM with >30% fines soil types

Estimated likely crack width in core for Reservoir stage being considered (mm) 1 2 5 10 25 50 75 100 Probability of initiation of erosion for different seepage gradients Average Hydraulic Gradient 0.1 0.02 0.1 0.6 0.9 0.95 1.0 1.0 1.0 0.25 0.2 0.6 0.95 1.0 1.0 1.0 1.0 1.0 0.5 0.6 0.9 0.99 1.0 1.0 1.0 1.0 1.0 1.0 0.9 0.95 1.0 1.0 1.0 1.0 1.0 1.0 2.0 0.95 1.0 1.0 1.0 1.0 1.0 1.0 1.0 5.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

Note. (1) The gradient is the average hydraulic gradient from the upstream to the downstream of the core at the level of the assumed crack under the reservoir level under consideration. No allowance is made for seepage head losses in the zones upstream or downstream of the core.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008

5-33

0 1.8 0.0 0.0 1.7 0.0 0.6 0.0 2.5 0.9 1.0 1.0 1.9 0.95 1.7 0.0 1.0 1.1 0.0 1.1 0.0 1.0 5.0 1.0 1.0 1.0 1.0 1.1 0.9 0.9 1.0 1.0 0.9 0.7 0.1 0.3 0.95 1.95 1.1 0.0 0.5 0.02 0.31 .95 1.0 1.0 0.7 0.32 .0 1.5 0.95 1.0 2.0 1.0 1.9 0.0 1.0 Note.4 0.0 1.95 0.0 1.95 1.2 0.9 0.0 1.7 1.0 1.0 1.1 0.0 1. or CL-ML soil types Estimated likely crack width in core for Reservoir stage being considered (mm) 1 2 5 10 25 50 75 100 Probability of initiation of erosion for different seepage gradients Average Hydraulic Gradient 0.0 1.0 5.0 1.5 0.0 1.0 0.Estimation of probability of initiation in a crack for SC with >40% fines.0 1.9 0.8 0.0 1.2 0.95 1.25 0.5 0.0 1. No allowance is made for seepage head losses in the zones upstream or downstream of the core.0 0.0 0.0 1.0 0. No allowance is made for seepage head losses in the zones upstream or downstream of the core.01 0.Estimation of probability of initiation in a crack for CL or MH soil types Estimated likely crack width in core for Reservoir stage being considered (mm) 1 2 5 10 25 50 75 100 Probability of initiation of erosion for different seepage gradients Average Hydraulic Gradient 0.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.0 1.0 1.4 0.4 0. (1) The gradient is the average hydraulic gradient from the upstream to the downstream of the core at the level of the assumed crack under the reservoir level under consideration.0 1.0 Note.3 0.0 1. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-34 .5 0.0 1.7 0.1 0.7 0.0 1.02 0.0 1. (1) The gradient is the average hydraulic gradient from the upstream to the downstream of the core at the level of the assumed crack under the reservoir level under consideration. Table 5.03 0.2 0.25 0.95 1.

01 0.01 0.1 0.005 0.0 0.33 .01 0.0 5.9 0.05 0.05 0.34 .9 1.0 1.3 0.1 0.0 1.95 1.3 0.0 1.0 1.8 0. No allowance is made for seepage head losses in the zones upstream or downstream of the core.1 0.0 1.4 0.9 1.5 0.5 0.8 0.6 0.0 1.0 1.1 0.0 1.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.01 0.0 0.5 0. (1) The gradient is the average hydraulic gradient from the upstream to the downstream of the core at the level of the assumed crack under the reservoir level under consideration.0 1.3 0.1 0.95 1.0 1.002 0.6 0.4 0.0 1.5 0.0 0.01 0.0 1.001 0.1 0.05 0.2 0.005 0.0 1.04 0.0 2.0 Note.05 0. Table 5.0 1.0 5.3 0.0 0.1 0.8 0.25 0.005 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-35 .3 0.8 1.0 1.002 0.6 0.6 0.0 1.1 0.3 0.0 1. No allowance is made for seepage head losses in the zones upstream or downstream of the core.0 Note.95 1.95 1.0 0.05 0.25 0.02 0.95 1.0 2.005 0.1 0.0 1.0 0.0 0.0 1.02 0.1 0.02 0.0 1.1 0.0 1.0 1.4 0.4 0.0 1.2 0.0 1.0 1.9 1.Estimation of probability of initiation in a crack for CL-CH or CH with LL<65% soil types Estimated likely crack width in core for Reservoir stage being considered (mm) 1 2 5 10 25 50 75 100 Probability of initiation of erosion for different seepage gradients Average Hydraulic Gradient 0. (1) The gradient is the average hydraulic gradient from the upstream to the downstream of the core at the level of the assumed crack under the reservoir level under consideration.6 0.Estimation of probability of initiation in a crack for CH with LL>65% soil types Estimated likely crack width in core for Reservoir stage being considered (mm) 1 2 5 10 25 50 75 100 Probability of initiation of erosion for different seepage gradients Average Hydraulic Gradient 0.6 0.95 0.

0 1.3 0.0 1.5 1.0 0.36 – Estimated hydraulic shear stress (N/m2) from water flowing in an open crack.3 0.1 0.0 1.0 1.5 5 10 25 50 0.0 0.0 1.0 1.0 1.0 1.02 0.0 1.6 0.0 0. CL-CH) Estimated likely crack width in core for Reservoir stage being considered (mm) 1 2 5 10 25 50 75 100 Probability of initiation of erosion for different seepage gradients Average Hydraulic Gradient 0.0 1. Table 5.6 1.0 1.0 25 50 125 250 500 1250 2500 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-36 .0 1.0 1.0 1.0 2.0 1.0 1.0 Note.0 1.5 0.0 1.0 1.5 5 12 25 50 125 250 1.0 0.25 2.8 1.0 5 10 25 50 100 250 500 2.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.5 2.0 1.25 0.35 . No allowance is made for seepage head losses in the zones upstream or downstream of the core.1 0.5 6 12 25 60 125 0.9 1.0 1.0 1.0 1. versus crack width and flow gradient Crack Width Millimeters 1 2 5 10 20 50 100 Flow Gradient in Crack 0.7 1.Estimation of probability of initiation in a crack for dispersive soils (CL.0 1.0 1.3 0.0 1. (1) The gradient is the average hydraulic gradient from the upstream to the downstream of the core at the level of the assumed crack under the reservoir level under consideration.05 0.0 1. CH.1 0.0 5.0 10 20 50 100 200 500 1000 5.0 1.0 1.5 1 2.0 1.0 1.25 1.0 1.

Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.4.29 to 5.Estimate the likely crack depth and assess the probability erosion will initiate using the method described in Section 5. The earthquake load partition ranges should be selected such that they coincide with the damage class contours shown in Figure 5.33 5.35 Initial Shear Stress Assumed for Assessing Probabilities of Initiation of Erosion 2 Pa 2 Pa 4 Pa 5 Pa 25 Pa 60 Pa 2 Pa Table Number 5.39.< 40% fines.30 5.5 • • Estimation Of The Probability Of Transverse Cracks In The Embankment Caused By Earthquake (IM13) First . Do this for each earthquake load partition.32 5.9 for earthfill and rockfill dams. Second .8 for earthfill dams or Figure 5. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-37 . assume damage class 3. Third .Estimate the probability of transverse cracking and likely maximum crack width from the damage class and Table 5. • • This procedure applies to situations where liquefaction does not occur in the dam or its foundations.6 for details).8 for earthfill dams or Figure 5.9 for earth and rockfill dams.34 5.29 5.37 – Initial Shear Stress assumed for Tables 5.Estimate the likely damage class which the embankment may experience as a result of the from the earthquake and peak ground acceleration on bedrock at the dam site using Figure 5.35 Soil Types ML and SM with < 30% fines SC with .Determine the earthquake hazard for the site (refer to Section 3. If flow liquefaction occurs assume the damage is class 4. For cases where liquefaction occurs but it is not flow liquefaction. SM with >30% fines SC with > 40% fines. and CL-ML CL and MH CL-CH and CH with LL < 65% CH with LL > 65% Dispersive soils 5. Fourth .31 5.

Figure 5.500 > 500 Maximum Relative Crest Settlement (2) % <0.Incidence of transverse cracking versus seismic intensity and damage class contours for earthfill dams (Pells and Fell 2002.03 0.5 >5 No or Slight Minor Moderate Major Severe Collapse Maximum crack width is taken as the maximum width. in millimeters.1.0.30 30 .Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Table 5.80 80 .5 1. Maximum relative crest settlement is expressed as a percentage of the structural dam height.2 0. of any longitudinal cracking that occurs.5 0.03 . 2003) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-38 .5 .38 – Damage classification system (Pells and Fell. 2002.5 .0.8 .150 150 .2 . 2003) Damage Class Number Description 0 1 2 3 4 5 (1) (2) Maximum Longitudinal Crack Width (1) mm < 10 mm 10 .

6 Maximum likely crack width at crest mm (inches) 20 50(2) 75(3) 125(5) 175(7) Damage class A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 5-39 .e.10 Probability of transverse cracking ” 0 1 2 3 4 0.10 Probability of transverse cracking 0. 5. 2003) Table 5.2.01 0.2 0.e. 5.01 0.001 0.10 0.05 0.Incidence of transverse cracking versus seismic intensity and damage class contours for earthfill and rockfill dams (Pells and Fell 2002.9 .05 0.5 Maximum likely crack width mm (inches) 5 20 50(2) 100(4) 150(6) For cases where cross valley or cross section cracking assessment is in upper two “boxes” i.Probability of Initiation of Erosion in Transverse Cracks in the Embankment SECTION 5 Figure 5. RF x LF > 13 in Tables 5. RFxLF ≤ 12 in Tables 5.4 and 5.2.39 – Estimation of the probability of transverse cracking from the damage class For cases where cross valley or cross section cracking assessment is in lower three “boxes i.25 0.4 and 5.

Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6 Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment 6. Backward erosion and suffusion will apply to cohesionless soils and as discussed in Section 6. erosion will occur in cracks or continuous open flow paths formed by collapse of the soil on saturation. Even low plasticity soils may form a crack so soils with a plasticity index between zero and 7 should be considered for both erosion in a crack and backward erosion and suffusion and the highest probability of initiation carried forward in the analysis. This will be one or more of the following: • • • Backward erosion Suffusion Erosion in a crack or flaw resulting from the poor compaction. c) Assess the probability of erosion in the poorly compacted or high permeability zone (P IP ) for the mechanism using the relevant method described in Section 6. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-1 . d) Estimate the probability of initiation of erosion (P I ) = (P P ) x (P IP ) for each poorly compacted or high permeability zone.5). or between aggregated particles of the soil.6.3 and 6.3).6.2. Poor compaction or high permeability layer due to freezing (Section 6. Poor compaction or high permeability layer around a conduit or other structure penetrating the core (Sections 6. Poor compaction or high permeability layer on the foundation contact (Section 6. Adjust this estimate to account for observations (Section 6.1 a) Overall Approach Estimate the probability of a continuous poorly compacted or high permeability zone (P P ) for each of the mechanisms which can lead to a poorly compacted or high permeability zone. For many dams one or more of the mechanisms will not be present. and can be considered as equivalent to erosion in a crack. For cohesive soils. These are: • • • • Poor compaction or high permeability layer during construction within the core (Section 6. b) Assess the erosion mechanism(s) which will apply.4).2).2.1). to soils with a plasticity index ≤ 7.1.2.

1 The scenarios which may lead to poorly compacted or high permeability zones in the core of an embankment include: • • • Poorly compacted layers in the core. Coarser soil layers in the core due to variability in particle size and soil type in the borrow areas. This can result in a cohesionless layer within a cohesive core. A segregated layer in the core due to the presence of coarser particles and poor construction practices.2 Estimation Of The Probability Of A Continuous Poorly Compacted or High Permeability Zone In The Embankment Or On The CoreFoundation Contact Poorly Compacted or High Permeability zone within the core (IM14) 6. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-2 . or a coarser cohesionless layer within a finer cohesionless soil. For these purposes a cohesionless soil is non plastic.2. The effect of these factors on the likelihood of a poorly compacted or high permeability zone is different for cohesive and cohesionless soils so they are treated separately.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6.

(1) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-3 .g. or by pushing into place by excavator or bulldozer or in very thick layers) No control on layer thickness. often > 18 to 24 inches (450 mm to 600 mm) loose and Layer thickness Layer thickness 6 to 10 inches (150mm to 250mm) after compaction Layer thickness at or beyond the limit of compaction equipment (e. no formal compaction (e. or rolled in thick layers beyond the capability of the roller Soil placed with. no compaction by rollers.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.g.1 – Factors influencing the likelihood of poorly compacted or high permeability zones in the embankment-cohesive soils (IM14) Likelihood Factor (LF) Factor Relative Importance Factor (RF) Less Likely (1) Neutral (2) More Likely (3) Much More Likely (4) METHOD BASED ON COMPACTION EQUIPMENT. by horse and cart in old dams. > 12 to 18 inches (300 mm to 450 mm) after compaction) Dry of optimum moisture content (1) and moisture content Around optimum moisture content (1) Well dry of optimum moisture content. LAYER THICKNESS AND MOISTURE CONTENT (a) Compaction equipment (3) (a) As for “neutral” but with good documentation and records (2) Soil compacted by suitable rollers in suitable layer thicknesses Soil placed and compacted by bulldozer.

g. FACTORS APPLYING TO BOTH ALTERNATIVES Borrow area variability.25 (a) Make an assessment based on a combination of available data (b) If soils are well compacted this mechanism will not apply A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-4 .5 Very narrow core. (2) Uniform soils in the borrow areas.g. ≥ 98% standard dry density ratio.25< W/H<0. moisture content 2% dry of optimum to 1% wet of standard OWC (b) Well compacted to e. 2% to 3% dry of standard OWC Variable soils in the borrow areas Moderate site supervision Layers very poorly compacted. LAYER THICKNESS AND MOISTURE CONTENT(a) AND/OR (a) Measured or estimated compaction density ratio and moisture content (COHESIVE SOILS) (3) (a) All very well compacted to e.5<W/H<1.5 Narrow core. 0. W/H>1. < 93% standard dry density ratio. 3% dry of standard OWC Very variable soils in borrow areas including gravely soils.5 Uniform or minor variability in the borrow areas. Good site supervision documented with laboratory tests.g. W/H<0. Good site supervision Site supervision Core geometry Width (W)/Height (H) Notes: 0. 95-98% standard dry density ratio. dry of standard optimum moisture content e. < 90% standard dry density ratio. dry of standard optimum moisture content e.g. Poor site supervision. moisture content 2% dry of optimum to 1% wet of standard OWC Layers poorly compacted.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Likelihood Factor (LF) Factor Relative Importance Factor (RF) Less Likely (1) Neutral (2) More Likely (3) Much More Likely (4) METHOD BASED ON COMPACTION EQUIPMENT. (1) Wide core.

36% to 65% relative density: SPT (N 1 ) 60 9 to 25 Much More Likely (4) Soil placed with. <35% relative density. medium dense.g. very loose to loose. > 12 to 18 inches (300 mm to 450 mm) after compaction) Dry of optimum moisture content(a) Layers moderately compacted. no compaction by rollers. very dense. no formal compaction ( e.g. e. dense. e. >85% relative density with good documentation and records (b): SPT (N 1 ) 60 > 42 All well compacted e. or by pushing into place by excavator or bulldozer or in very thick layers No control on layer thickness. often > 24 to 36 inches (600 mm to 900 mm) loose EITHER (a) Compaction equipment and Layer thickness Layer thickness 8 to 12 inches (200 mm – 300 mm) after compaction and moisture content Around optimum moisture content (a) Well dry of optimum moisture content(a) Layers very poorly compacted. by horse and cart in old dams. or rolled in thick layers beyond the capability of the roller Layer thickness at or beyond the limit of compaction equipment (e.g.g. SPT (N 1 ) 60 < 8 AND/OR(a) Measured or estimated compaction density ratio and moisture content (NONCOHESIVE SOILS) (3) (a) All very well compacted e.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6. 66% to 85% relative density : SPT (N 1 ) 60 26 to 42 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-5 .2 – Factors influencing the likelihood of poorly compacted or high permeability zones in the embankment-non cohesive soils (IM14) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) (a) Less Likely (1) As for “neutral” but with good documentation and records (b) Neutral (2) Soil compacted by suitable rollers in suitable layer thicknesses More Likely (3) Soil placed and compacted by bulldozer.g.g.

00003 0.5<W/H<1. middle and lower parts of the embankment) negligible negligible 0. (1) Wide core.01 Below POR 0.5 Narrow core.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Likelihood Factor (LF) Factor Relative Importance Factor (RF) (2) Less Likely (1) Uniform soils in the borrow areas.Probability of a poorly compacted or high permeability layer in the embankment (Relative importance factor (RF)) x (Likelihood factor (LF)) versus ∑ (applies to upper.5 Neutral (2) Uniform or minor variability in the borrow areas. Table 6. Borrow area variability.0004] [0.3 .25< W/H<0. Site supervision Moderate site supervision Core geometry Width (W)/Height (H) Notes: 0. W/H<0. W/H>1.5 Above POR 6 9 11 13 18 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-6 . Poor site supervision.0003 0. Good site supervision More Likely (3) Variable soils in the borrow areas Much More Likely (4) Very variable soils in borrow areas including gravely soils.005 0.25 (a) Make an assessment based on a combination of available data (b) If soils are well compacted this mechanism will not apply. 0.001 [0. Good site supervision documented with laboratory tests.5 Very narrow core.0001 0.05 0.005] 0.

Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6. no compaction by rollers Layer thickness at the limit of compaction equipment (e. soil poorly compacted or allowing segregation against foundation surface.g.2 Poorly Compacted or High permeability layer on the core-foundation contact (IM15) Table 6. or concrete to correct slope (c) irregularities Neutral (2) Regular rock surface. shotcrete or concrete treatment Poor stripping of soil foundation leading to poor compaction of first lift Poor compaction methods used. OR Foundation preparation below the core. soil foundations Compaction methods for contact zone (3) Uniform well compacted soil (c) foundation Compacted soil foundation (b) (2) As for “neutral” but with good documentation and records (d) Soil compacted using special compaction methods (rubber tyres. compaction wet of OWC) Soil placed and compacted by bulldozer. often > 18 inches (450 mm) or 24 inches (600 mm) loose EITHER Foundation preparation below the core.4 – Factors influencing the likelihood of poorly compacted or high permeability zones on the core-foundation/abutment contact (IM15) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Uniform rock surface or surface treated with shotcrete. or rock surface treated with shotcrete or concrete to correct slope (b) irregularities More Likely (3) Irregular rock surface. > 12 inches (300 mm) or 18 inches (450 mm) after (a) compaction) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-7 . Thick layer thickness. overhangs with no slope correction. with minimal slope correction or treatment or irregular or benched soil with no compaction Irregular or benched soil with no compaction Much More Likely (4) Very irregular rock surface.2. use more plastic materials. rock foundations.

Core geometry Width (W)/Height (H) Notes: (a) (b) (c) (d) (e) Smaller thicknesses are for cohesive soil.5 Much More Likely (4) Very narrow core.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Likelihood Factor (LF) Factor Relative Importance Factor (RF) (1) Less Likely (1) Wide core. 0. larger thicknesses for cohesionless soil. For this foundation preparation the probability of this mechanism is negligible Even situations where soil is well compacted can soften or loosen if the contact is irregular. For most cases this mechanism may be judged as of negligible likelihood and assigned zero probability.05 0.25< W/H<0.Probability of a poorly compacted or high permeability layer on the core (Relative importance factor (RF)) x (Likelihood factor (LF)) foundation contact versus ∑ Negligible Negligible 0.1 Above POR 6 Notes: 9 11 13 18 24 RF x LF (1) This and some other combinations which involve contradictory compaction scenarios are quite unlikely to occur (2) If soils are well compacted they this mechanism will not apply Note: “POR” refers to the Pool of Record level + 1 foot.0005 0.002 [0. Table 6.5<W/H<1. W/H>1.0001 0.005 0.5 More Likely (3) Narrow core.5 Core wall cutoff or grout cap cutoff present Neutral (2) 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-8 .0002 0.0004] [0.25 Continuity of features. W/H<0. assess foundation preparation and compaction methods for the central portion of the section.01 Below POR 0. For homogeneous earthfill dams.005] 0.5 .

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-9 .3.3 Poorly Compacted or High Permeability Layer in the Embankment due to Freezing (IM16.8.2. This section describes how to assess the probability of the presence of such features and the depth to which they may exist. If the reservoir stage being considered is below the likely depth of freezing. follow the procedure below to estimate the probability of frost action affecting the dam and the depth of the frost action.1 At the crest of the embankment (IM16) Freezing conditions can result in frost heave and formation of ice lenses in the crest of dams. the probability of a crack or poorly compacted zone due to freezing can be assumed = zero. loosened and/or cracked soil may be present in which internal erosion may initiate if the reservoir rises sufficiently high. The method for estimating the maximum width of flaw which may result from the frost action is detailed in Section 6.2. IM17) 6.6. The procedure is: FIRST Consider the maximum likely depth of freezing for the soil in the core of the dam. and the climate using Table 6. SECOND For cases where the reservoir stage is above the base of potential freezing. When the ice thaws.6.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6.

SW).GW) and sandy (SP.SW). MH). And clayey silts (ML-CL) Climate Classification of core material (2) Crest zoningsurface layer over core (1) Greater than 6 ft (2 meters) of rockfill over the core Gravely material or rockfill 3 ft to 6 ft (1m to 2m) thick over the core A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-10 . Clean gravel (GP.02mm. silty sands (SM) with > 15% finer than 0.GWGM.02mm High plasticity clays (CH) More Likely (3) Sub arctic or alpine climates where temperatures remain below freezing point for 1 to 3 months Silty gravely (GM.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.45m to 1m) thick over the core No surface layer with dam core extending to crest level or thin ( less than 3 inches (75 mm)) road pavement or gravely material Note (a) If climatic conditions are as described here this mode does not apply Much More Likely (4) Sub arctic or alpine climates where temperatures remain below freezing point for 3 months or more Silts (ML.02mm Neutral (2) Temperate climate where temperatures may remain below freezing point for up to 1 month Gravely (GP. SW-SM. less than 3% finer than 0. GC) and clays with plasticity index < 12 Gravel material or rockfill 18 inches to 3 ft (0.6 – Factors influencing the likelihood of cracking and poorly compacted zones in the upper part of embankment dams due to freezing (IM16) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Other climates where temperatures do not fall below freezing point except possibly overnight or for a (a) day or two.GW) and sand (SP. SP-SM) with 6% to 15% finer than 0.02mm Clayey sands and gravels (SC. soils with between 3% and 6% finer than 0.GP-GM) and silty sandy soils (SM.

8– Maximum likely depth of freezing induced flaws based on climate Maximum likely depth Climate (feet) Sub arctic or alpine climates where temperatures remain below freezing point for 3 months or more Sub arctic or alpine climates where temperatures remain below freezing point for 1 to 3 months Temperate climate where temperatures may remain below freezing point for up to 1 month Other climates where temperatures do not fall below freezing point except possibly overnight or for a day or two.8 1.0001 0.9 6 10 11 12 16 21 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot Table 6.3.Probability of cracking or poorly compacted zones in the crest due to freezing (Relative importance factor (RF)) x (Likelihood factor (LF)) versus ∑ negligible negligible 0.2 On Seasonal Shutdown Layers And On The Surface Of Staged Embankments (IM17) FIRST This mechanism only applies where there has been a seasonal shutdown during construction. and multiply this probability by the assessed likelihood the layer will be continuous across the core.9. SECOND Where the mechanism applies estimate the probability of a freezing layer using Table 6. This likelihood is to be assessed by the risk analysis team based on the information available and their judgement.3 0. 6 4 2 0 (meters) 1.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.7 .1 0.3 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-11 .2.6 0 6. or the embankment has been staged.001 0.

Frozen layers removed from embankment and replaced with new soil or adequately reworked to specified moisture content. less than 3% finer than 0.02mm Clayey sands and gravels (SC. SWSM. GW) and sandy (SP. SP-SM) with 6% to 15% finer than 0. moisture adjusted to specified range. silty sands (SM) with > 15% finer than 0. poor moisture control on recompacting the soil Construction practices regarding clean-up of frozen layers after construction shutdowns or the surface of the earlier stage of the dam Climate (2) Other climates where temperatures do not fall below freezing point except possibly overnight or for a day or two.02mm. MH). surface recompacted. More Likely (3) Moderate control. GP-GM) and silty sandy soils (SM. probability=zero Neutral (2) Good control and practices. No attempt to scarify or remove frozen layers.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.9 .Factors influencing the likelihood of high permeability layer on seasonal shutdown layers during construction and staged construction surfaces due to freezing (IM17) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (4) Less Likely (1) Very good control and clean-up practices. Attempts to scarify frozen layers. Clean gravel (GP.GW) and sand (SP. but depth of scarifying insufficient or difficulties with moisture control on re-compacting the soil Much More Likely (4) Poor control.02mm High plasticity clays (CH) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-12 .02mm Temperate climate where temperatures may remain below freezing point for up to 1 month Sub arctic or alpine climates where temperatures remain below freezing point for 1 to 3 months Silty gravely (GM. SW). GC) and clays with plasticity index < 12 Sub arctic or alpine climates where temperatures remain below freezing point for 3 months or more Silts (ML. And clayey silts (ML-CL) Classification of core material (1) Gravely (GP.SW) soils with between 3% and 6% finer than 0. GW-GM. If this condition is present. surfaces scarified.

1 0.9 AbovePOR 7 10 12 13 18 22 28 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-13 .01 0.Probability of a high permeability layer on seasonal shutdown layers during (Relative construction and staged construction surfaces due to freezing versus ∑ importance factor (RF)) x (Likelihood factor(LF)) negligible negligible negligible negligible negligible 0.10 .001 0.001 0.01 0.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.1 Below POR 0.0001 negligible 0.

close spacing Compaction of earthfill around the conduit (2) Compaction by rollers to>98% standard maximum dry density at OWC1% to OWC+2% No formal compaction. base width not less than conduit width plus 2 meters either side. deep. or poor compaction practices used adjacent to conduits (e.11 – Factors influencing the likelihood of poorly compacted or high permeability zones along outside of a conduit. concrete precast or cast in situ. (IM18) Relative Importance Factor (RF) (3) Likelihood Factor (LF) Less Likely (1) Concrete encased round pipe. backfilled to the surface with (a) concrete Medium width. near vertical sides in soil or rock. widely spaced Compaction by hand equipment.g. thick layers inappropriate for equipment) Narrow. Table 6. Flowable fill (CLSM) Well detailed cut off collars. not concrete encased Conduit type and surround Cut off collars (2) Poorly detailed cut off collars. USBR design Compaction by hand and mechanical equipment to >95% standard maximum dry density at OWC1% to OWC+2% Wide. No cut off collars Factor Neutral (2) Concrete encased round pipe. No desiccation of sides of trench More Likely (3) Masonry. backfilled to the surface with concrete it should not be considered as a conduit through the embankment (probability = zero) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-14 . and slope.3 Probability Of A Poorly Compacted or High Permeability Zone Around A Conduit or Features Allowing Erosion Into the Conduit Poorly Compacted or High Permeability Zone Around A Conduit Through The Embankment (IM18) 6. concrete precast or cast in situ.3. thick layers. Much More Likely (4) Any round pipe (including corrugated metal pipe). sloping sides.12. and/or sides of trench highly desiccated and cracked Conduit trench details (2) Trench totally in non-erodible rock. backfilled with soil. depth.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6. slopes flatter than 1H:1V. brick.1 Assess the probability of a poorly compacted or high permeability zone around the conduit using Table 6. and/or sides of trench desiccated and cracked Note: (a) If the conduit is totally embedded in a trench totally in non-erodible rock. vertical sides. dry of optimum moisture content Poorly detailed cut off collars.11 and Table 6.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-15 .02 0.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6. Note: “POR” refers to the Pool of Record level + 1 foot Figure 6.005 [0.0015 0.01] 0.1 > 20% Above POR 0.0006 0.0009] [0.001 0. for flood protection dams).g.Probability of a poorly compacted or high permeability zone associated with a (Relative importance factor (RF)) x (Likelihood factor(LF)) conduit versus ∑ 0.0001 0.02 Below POR 0. backfilled to the surface with concrete it should not be considered as a conduit through the embankment (probability = negligible) Above POR probabilities apply where the reservoir rise is greater than 20% increase in previously recorded hydraulic head (e.002 0.003 [0.12 .005 0.0002 0.05 0.0005 0.0003 0.5 First Fill (untested) 9 Note: (1) (2) 12 16 20 26 36 RF x LF If the conduit is totally embedded in a trench totally in non-erodible rock.1 Example of poor detailing of seepage collars around a conduit (from FEMA 2005).

This should be considered when assessing the likelihood of breach. Observed Condition Careful inspection showing no evidence of open joints or cracks Careful inspection showing no evidence of open joints. It should be noted that this mechanism can apply for erosion into the conduit from the foundation and from the embankment.3 to 0. If the internal condition of the conduit is not known.11. The likelihood of development of piping along the conduit given erosion initiates into the conduit should be estimated from Table 6.3.005 0. Table 6. then use Table 6.9 1.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6. the type of joints. a) b) If the internal condition of the conduit is regularly inspected and there is good documentation of the inspections. then estimate the probability of erosion into a conduit using Table 6. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-16 . and breach. and the quantity of seepage through the crack or joint.9 0. If corrosion is in progress consider time effects on the estimated probability and report this. and whether it penetrates through the conduit.2 Features Allowing Erosion into a Conduit (IM19) Assess the probability of a feature being present in the conduit which would allow erosion of the surrounding soil into the conduit. or cast iron or steel with advanced stages of corrosion ( 2) Probability of initiation for Erosion into the Conduit Negligible 0. the width of cracks. The two potential failure paths should be considered in progression.05 to 0.0 (1) Open joints or cracks present. but hairline cracks in concrete are present Open joints or cracks present.13. Select best estimate accounting for the extent of corrosion. no evidence of seepage Corroded corrugated metal.1 to 0. evidence of seepage through joint/crack Open joints or cracks present. Erosion into a conduit may be followed by development of erosion along the conduit in the progression stage. Consider the conduit in its current condition. evidence of erosion of soil into the conduit Notes (1) (2) Select best estimate taking account of the width of joint openings.3 0.14. detection and intervention.001 to 0.13 – Factors influencing the likelihood of initiation of erosion into a non-pressurized conduit when internal condition is known. with the ∑ (Relative importance factor (RF)) x (Likelihood factor (LF)) taken from Table 6.16.

Either Conduit type Concrete.14 – Factors influencing the likelihood of initiation of erosion into a non-pressurized conduit when the internal condition is not known (IM19) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Cast in-situ concrete. 30 years steel) Corrugated metal pipe < 5 years Wet conduit. Concrete encased steel or iron. More Likely (3) Precast concrete culverts Cast in-situ concrete culvert with poor or no water stops “New” cast iron or steel pipe (“New” < 20 years cast iron.5 meters/sec)) flows. masonry.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6. not encased (“old” > 60 years cast iron. high velocity (> 15ft/sec (3 meters/sec)) or surging flows “Old” corrugated metal pipe (>10 years). calculated small settlements Neutral (2) Concrete pipe with bell and spigot joints. Conduit operation (2) Dry conduit. brick OR Steel/Cast Iron Pipe Settlement of the conduit (1) Rock foundation or measured small settlements Deep compressible foundation soils calculated large settlements Measured significant settlement and differential settlement. New steel or cast iron with extensive corrosion protection. or dry Shallow compressible foundation soils. brick. medium to high velocity flows Wet conduit. Precast concrete pipes with high quality joints with water stops. junction of shaft and conduit within the embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-17 . with low velocity (<5 ft/sec ( 1. Much More Likely (4) Masonry. inspected regularly Wet conduit. dewatered regularly for inspection. 10 years steel) “Old” cast iron or steel pipe.

Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.0005 0.Probability of initiation of erosion into a conduit versus ∑ (Relative importance factor (RF)) x (Likelihood factor(LF)) 0.01] 0.5 1. Table 6.Probability of the development of piping along the conduit given erosion (Relative importance factor (RF)) x (Likelihood initiates into the conduit versus ∑ factor(LF)) from Table 6.0 9 12 16 20 26 36 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-18 .0001 0.1 0.5 First wetting 6 9 11 13 18 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot .005 [0.1 Below POR 0.0003 0.003 0.11 0.001 0.01 0.02 0.15 .16 .02 0.1 0.05 0.001] [0.

3).4. 3) Differential settlement adjacent to the wall (using Table 6.4 Poorly Compacted or High Permeability Zone or Gap Associated With A Spillway Or Abutment Wall Approach 6.4.4. For mechanisms (2) and (3) estimate the probability erosion will initiate in the crack or gap using the methods detailed in Section 5.4.17 in Section 6.1 • Assess the probability of a poorly compacted or high permeability zone or gap for each of the three mechanisms for a spillway or abutment wall.2). For mechanism (1) estimate the probability erosion will initiate in the high permeability zone using the methods detailed in Section 6. 1) Poorly compacted or high permeability zone associated with the wall (using Table 6.4.19 in Section 6.6. • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-19 .4) • • Adopt the maximum probability of the three mechanisms.21 in Section 6. 2) Crack/gap adjacent to the wall (using Table 6.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6.

4.001 0.0001 0.Probability of a high permeability zone associated with a spillway or abutment (Relative importance factor (RF)) x (Likelihood factor(LF)) wall versus ∑ 0.005 [0.00005 0.g. or poor compaction practices (e.002 0.02 Below POR 0.2 Poorly Compacted or High Permeability Zone Associated with a Spillway Or Abutment Wall (IM20) Table 6.0003] [0.17 – Factors influencing the likelihood of a poorly compacted or high permeability zone associated with a spillway or abutment wall (IM20) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Compaction by rollers to >98% standard maximum dry density at OWC2% to OWC+2% Neutral (2) Compaction by hand and mechanical equipment to >95% standard maximum dry density at OWC2% to OWC+2% Single but with good compaction around the buttress Smooth. little evidence of good compaction Rough and irregular Finish on wall (1) Smooth planar coupled with flat slope (flatter than 0.18 .002 0. masonry/ brick walls) Compaction of earthfill adjacent to the wall Concrete buttresses (2) None Single with poor details such as vertical sides.0002 0.5 Above POR 6 9 11 13 18 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-20 .05 0. placed in very thick lifts or allowing segregation against wall) Several close together preventing good compaction Vertical and horizontal steps (e. dry of optimum moisture content Much More Likely (4) No formal compaction. thick layers.5H:1V) Table 6.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6.01] 0. planar More Likely (3) Compaction by hand equipment.g.

02 Below POR 0.Probability of a gap or crack associated with a spillway or abutment wall (Relative importance factor (RF)) x (Likelihood factor(LF)) versus ∑ 0.01] 0.002 0.2(a) (b) See Figure 6.0004] [0. Cantilever wall.0002 0.00005 0.5H to 1V More Likely (3) Vertical.2(b) (c) Select the factor which best describes the worse of the two conditions Table 6.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6.19 – Factors influencing the likelihood of a crack or gap adjacent to a spillway or abutment wall (IM21) Likelihood Factor (LF) Relative Importance Factor (RF) (3) Factor Less Likely (1) Sloping.0001 0.3 Crack/Gap Adjacent to a Spillway or Abutment Wall (IM21) Table 6. flatter than 0.002 0.1H to 1V to 0.20 .4. subject to cyclic reservoir level conditions.001 0.3 Above POR 6 9 11 13 18 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-21 . Poor freeze /thaw details Slope of wall Wall type. or Vertical with flatter slope on lower part Cantilever wall not normally subjected to cyclic reservoir level conditions Much More Likely (4) Overhanging over the core width.005 [0.5H to 1V Neutral (2) Sloping.05 0. little evidence of good compaction Several close together preventing good compaction Notes: (a) See Figure 6. wall stiffness (2) Very stiff gravity wall or counterfort wall Thin gravity wall Counterfort wall subjected to cyclic reservoir level conditions None or Single with probably good to reasonable compaction around the buttress Concrete buttresses (1) Single or multiple with good details and proven good compaction around the buttress Single with poor details such as vertical sides. 0.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-22 .2 .Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Figure 6.Situations where a gap may form between the dam fill and spillway wall (a) Steep foundation adjacent spillway wall. (b) Change in slope of the retaining wall (Fell et al 2004).

21 .4 Differential Settlement Adjacent to a Spillway or Abutment Wall (IM22) Table 6.5 Note. If this condition is present.25 < Wb/Hw < 1.0 < Wb/Hw <2. Hw.005] 0.0 Much More Likely (4) No or very narrow bench adjacent to wall.2 for definitions of Wb.02 0.Factors influencing the likelihood of cracking or hydraulic fracturing due to differential settlement adjacent a spillway or abutment wall (IM22) Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Wide bench Wb/Hw > 2. probability=zero (1) Dams less than 50 ft (15 m) high Dams 50 ft to 100 ft (15 m to 30 m) high High dams 100ft to 200 ft (30 m to 60 m) Very high dams >200 ft (60 m) Height of embankment Note: (a) See Figure 5.22 .0005] negligible [0. β1 (b) This mechanism only applies for Wb/Hw< 2.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6. β 1 > 60o Note.02 Below POR 0.4.5 Table 6. probability=zero (2) Gentle abutment slope β 1 < 25 o Neutral (2) Bench adjacent to wall 1.25 Cross valley profile under embankment core(1) Slope of abutments under embankment core(1) Moderate abutment slopes 25 < β 1 < 45 o o Steep abutments 45o < β 1 < 60o Very steep abutments. If this condition is present.2 Above POR 6 9 11 13 14 19 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-23 .Probability of cracking or hydraulic fracture due to differential settlement (Relative importance factor (RF)) x adjacent a spillway or abutment wall versus ∑ (Likelihood factor(LF)) negligible negligible negligible negligible negligible negligible negligible [0.002 0.5 More Likely (3) Narrow bench adjacent to wall 0. Wb/Hw < 0.

The effect is likely to be most important for reservoir level stages nearing dam crest level.23 . The path length and hence the gradient varies with the reservoir level.5 Wrap around details for connection of embankment dam to concrete gravity dam (IM23) Figure 6.3 shows typical details of the connection of an embankment dam to a concrete gravity dam. Table 6. There is a potential seepage path along PQRS.2 allowing a gap to form as the embankment settles High embankment Reservoir level approaching dam crest level (1) Uniform slope with no overhangs Well compacted embankment shoulders or zoning with all materials having a similar and high modulus Uniform concrete slopes with at least 0. When assessing the seepage gradient along QR. the likelihood there will be gaps or poorly compacted soil along PQ and RS should be assessed taking account of the factors in Table 6.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6. and gradients will be lower (1) • • Factors which make it likely there is a poor seepage contact along PQ and RS. the probability erosion will initiate should be assessed as for an embankment abutting a very stiff retaining wall (refer to Section 6.3).4.1H to 1V slope Low embankments Reservoir level at least 20 feet below dam crest level so there is a lesser likelihood a crack will persist to reservoir level Notes. There is a potential for the embankment to move away from the concrete dam at PQ and particularly at RS due to settlement of the embankment dam during construction. and gradients will be higher Overhangs in the concrete Poorly compacted shoulders leading to large settlements during and post construction Change in slope of concrete such as shown in Figure 6. there will be no benefit from the seepage path on PQ and RS if there is a gap there which can be seen on inspection. and usually to a lesser extent after construction. For the worst scenarios. and if so. The likelihood of a crack or gap being present.3 for definition of terms A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-24 . It is recommended that this situation be assessed as follows: • The primary control on seepage and initiation of erosion is considered to be along QR.23. (1) See Figure 6.4.Factors to be considered in assessing seepage gradients on wrap-around Factors which make it likely there is a good seepage contact along PQ and RS.

Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Figure 6.3 – Wrap around details for connection of embankment dam to concrete gravity dam A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-25 .

2.2.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6.4.4 are multiplied by the factor from Table 6. The multiplication factor should be selected taking account of what data is available.3 and 6.24. around conduits and adjacent walls A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-26 . This factor applies to the assessment of poorly compacted or high permeability zones in the embankment. allowing for the relative importance of the observations. The probabilities obtained from Sections 6. 6.5 Probability Of Poorly Compacted or High Permeability Zone Factors To Account For Observations Observation of seepage factor Where there are observations of seepage for the dam these can be used to modify the results of Sections 6. 6.3 and 6.

24 .0 No drill holes or CPT tests in core Probability of high permeability zone = 0.5 to 1. (2) Use multipliers towards the lower of the range for reservoir level below POR and towards the upper of the range for reservoir levels above POR.Seepage observation factors Influence on Likelihood Factor Less Likely (1) No seepage observed for dams where there is no potential for seepage to be hidden. Drill hole/CPT Factor (B) or Induced Defect Factor (C).0 Drill holes or CPT tests indicate softened zones in the core Probability of high permeability zone = 1. careful inspection for seepage Neutral (2) Seepage observed at toe of dams with permeable downstream zone or internal drainage systems.9 to 0.5 Multiple drill holes or CPT tests indicate no evidence of high permeability or softened zones Multiplier = 1. or potential for hidden seepage More Likely (3) Wet areas on the downstream slope Much More Likely (4) Concentrated seepage is present on the downstream slope Observed Seepage Seepage Adjustment Factor (A) Observations in drill holes/CPT in the core Multiplier = 0. whichever gives greatest probability of high permeability zones. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-27 .0 Drill hole/CPT Adjustment Factor (B) Multiplier = 0.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.0 Multiplier = 5 to 10 Drilling/Grouting Practices Inducing Defects in the Core No water drilling or pressure grouting through the core Drilling through the core with water causing excessive water losses in the core Multiplier = 2 to 5 High pressure grouting was carried out through the core Adjustment Factors for Induced Core Defects (C) Multiplier = 1.5 depending on quantity and quality of the investigations Multiplier = 1.0 Multiplier = 5 to 10 Notes: (1) Apply either Seepage Factor (A).1 to 0.5 to 1.0 Multiple drill holes or CPT tests indicate persistent high permeability zones are likely to be present Probability of high permeability zone = 0.

6. If the seepage gradients are >4. then the likelihood of backwards erosion and suffusion is negligible under the seepage gradients which occur in a conventional dam.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6. • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-28 . This is the gradient that is required to initiate backward erosion at the downstream end of the layer and also to progress the pipe by backward erosion to the upstream end of the layer. 6.4. From this and the time the reservoir will be above this stage assess whether there is sufficient time to develop the seepage gradient in the layer. • Estimate the average gradient (i pmt ) required to initiate and progress backward erosion from Figure 6. • Estimate the time it will take to develop a seepage gradient in the layer from the estimated permeability of the soil.6 Estimation Of The Probability Of Initiation Of Erosion In A Poorly Compacted or High Permeability Layer In The Embankment. From the particle size distribution of the core material estimate a representative uniformity coefficient Cu = D 60 /D 10 . follow the procedures in Sections 6.25 to assist in this estimate.4 and 6.6.2 Assessment Of The Probability Of Initiation Of Backward Erosion In A Layer Of Cohesionless Soil or Soil with Plasticity Index ≤ 7 The steps to be followed are: • Estimate the average seepage gradient (i av ) through the core at the level of the high permeability layer for the reservoir stage under consideration. If the soil is cohesive with a Plasticity Index > 7.6. allowing for potential collapse of the layer on saturation if the soil is not well compacted.5. Consider if erosion may occur through a crack in the soil using the procedure detailed in Sections 6. consider suffusion for soils with a Plasticity Index ≤ 12. Use Table 6.2 and 6.3 to assess the probability of backward erosion and suffusion. Adjacent A Wall Or Around A Conduit Screening Check On Soil Classification 6.6. For layers below the normal operating pool level the layer will be saturated and the seepage gradient will develop as the reservoir rises.6.1 Is the soil cohesive or cohesionless? If the soil is cohesionless or has a Plasticity Index ≤ 7.6.

for permeability anisotropy.4) CD CL CS CK Cz Cγ Cα CR D L • = = = = = = = = = = Correction factor for (D/L) Correction factor for total pipe length L Correction factor for grain size Correction factor. Correction factor for high-permeability under layer Correction factor for density Adjustment for pipe inclination Correction factor for dam axis curvature Depth of piping sand layer. This is for the anisotropy of the soil layer subject to backward erosion.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 • Correct this average gradient for the geometry. This gives (i pmt ) Corrected where (i pmt ) Corrected = where (C D C L C S C K C z Cγ Cα )i pmt CR (6. and grain size as detailed in the Supporting Information Section C6.γ w ) / γ w .0. Use i av and (i pmt ) Corrected or (i cr ) as inputs.27 for poorly compacted layers. not the embankment core as a whole. and case study data from Sweden which shows erosion may occur at average gradients less than 1. It also allows for application of corrections recommended by Schmertmann (2000) which for a thin layer of permeable soil indicate A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-29 . in direction perpendicular to α (m) Direct (not meandered) length between ends of a completed pipe path.9) i pmt = Maximum point seepage gradient needed for complete piping in the flume test based on the soil coefficient of uniformity Cu (from Figure 6. Adopt this gradient if it is smaller than (i pmt ) Corrected . horizontal to vertical permeability ratio of the zone subject to backward erosion. at least if the layers are poorly compacted. from downstream to upstream exit.2. measured along the pipe path (m) For Cu> 6.0.4. also estimate the critical gradient (i cr ) from i cr = ( γ sat . The table allows for gradients potentially being higher than the average at the downstream side as has been observed on many dams. • Estimate the probability of initiation and progression of backward erosion from Table 6.6.26 for well compacted layers and Table 6.

4 are required to initiate erosion. These probabilities apply for reservoir levels up to and above the pool of record. Table 6.2 for detailed descriptions No formal compaction Tracking by dozer or rolled in layers too thick for the equipment Compacted by rollers in suitable layer thicknesses to normal compaction standards Compacted by rollers in suitable layer thicknesses to normal compaction standards with well documented compaction records Time for developing seepage gradient with collapse settlement Minutes Minutes to a few hours Not applicable Not applicable Time for developing seepage gradient if there is no collapse settlement Not applicable.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 gradients higher than those from Figure 6.4.6. Collapse settlement is highly likely Not applicable.25 . Collapse settlement is highly likely Hours to days for silty sands and sands. assess the likelihood of initiation of erosion using Section 6.Time to develop seepage gradient in cohesionless soils Method of compaction (see Tables 6. Hours to days for silty sands and sands.1 and 6. • For poorly compacted silt/sand/gravel soils which are subject to collapse settlement on saturation. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-30 .

needed for complete piping (initiation and progression for an unfiltered exit) versus uniformity coefficient of soil (Schmertmann 2000).4 . ipmt.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 This relationship applies for Cu between 1 and 6. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-31 .γ w ) / γ w Figure 6.Maximum point gradient. calculate the critical gradient icr = ( γ sat . For Cu>6.

9 0.9 0.95 0.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.9 0.99 0.0 0.99 0.05 0.99 0.99 0.95 0.Estimation of the probability of initiation and progression of backward erosion P IP in cohesionless soils and soils with PI ≤ 7 for well compacted layers Average seepage gradient required to initiate & progress backward erosion (ipmt)Corrected or (icr) 0.05 0.5 0.02 0.0 0.9 2.9 0.99 0.99 0.5 0.99 0.1 0.99 0.5 0.2 0.25 0.95 0.99 0.01 0.95 0.5 1.95 Table 6.26 .99 0.Estimation of the probability of initiation and progression of backward erosion P IP in cohesionless soils and soils with PI ≤ 7 for uncompacted layers Average seepage gradient required to initiate & progress backward erosion (i pmt ) Corrected or (icr) 0.99 0. This is a check on whether the A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-32 .5 1.1 0.2 0.01 0.5 0.99 0.95 0.0001 0.27 .99 0.2 0.9 2.6.5 0.9 0.99 0.05 0.9 0. The finer fraction is defined by the point of inflection of broadly graded or gap graded soils.99 0.5 1.05 0.95 0.0 0.05 0.0 Average seepage gradient across embankment core (iav) 0.1 0.95 6.0 Average seepage gradient across embankment core (iav) 0.99 0.99 0.2 0.9 0.99 0.3 Probability Of Initiation Of Erosion By Suffusion In A Layer Of Cohesionless Soil or Soil with Plasticity Index ≤ 7 (PI ≤ 12 for seepage gradients >4) Check if the soil is potentially internally unstable Check if the proportion of the finer fraction is less than 40% of the total mass of the soil.25 0.25 0.001 0.001 0.95 0.2 0.95 0.2 1.01 0.9 0.1 0.05 0.99 0.99 0.0 0.25 0.2 0.

d60 and d90 sizes (the particle size for which 15%. (1989) Burenkova (1993) Skempton & Brogan (1994) Chapuis et al. If not (there are more than 40% finer particles) the coarse particles will “float” in the finer particles and suffusion is not possible. and Figure 6. 8 P is the probability.GRF] h" = d90 /d15 Figure 6.5 . P Sun (1989) data and UNSW data points B1. (Wan and Fell 2004) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-33 . etc.5 for soils with more than 10% fines passing 0.g.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 coarser particles will form a matrix into which the finer particles will fit.). Then estimate the probability the soil is internally unstable (PIUS) from Figure 6. Assess the probability the soil is internally unstable (P IUS ) From the particle size distribution for the soil. Plasticity Index ≤ 12. etc.).378 LOG(h") . These soils may experience backward erosion.648 h' + 3.90 0. .3.075 mm. If the answer is yes. (1984) Kenney & Lau (1984. . 60% and 90% are finer). P = exp(Z)/[1 + exp(Z)] Z = 2.50 0. determine the d15. and internally unstable soil samples are represented by solid symbols (e.g.2) plotted out of range Data source: Kenney et al. The grading curve is not adjusted for this procedure.05 0. continue to assess the likelihood of the soil being internally unstable as described in the following steps.95 4 2 1 10 100 1000 4000 [woSUNBD2. 85) Lafleur et al. predicted by logistic regression. .Contours of the probability of internal instability for silt-sand-gravel soils and clay-silt-sand-gravel soils of limited clay content and plasticity.10 0. that a soil is internally unstable if it is plotted along the respective dotted line .6 for soils with less than 10% fines passing 0. 10 Data point (447. .70 0. D1 not included in the logistic regression. 17. (1983) Kenney et al.30 0.075 mm (#200 sieve).701 h' = d90 /d60 6 D1 B1 0. (1996) UNSW Internally stable soil samples are represented by hollow symbols (e.

(1983) Kenney et al.95 0.3. (1989) Burenkova (1993) Skempton & Brogan (1994) Chapuis et al.05 0.20 0.10 0.99 1.2 0.1 0. A3 and C1) Internally stable soil samples are represented by hollow symbols (e.2 0.).075 mm (Wan and Fell 2004).30 0. Table 6. (1996) UNSW (only data points 4R.0 0.5 0.Seepage gradients at which suffusion may occur Porosity (volume of voids/total volume of soil) <0.2 0.99 0.2 0.05 0.5 0.1 0.95 1 10 100 200 h" = d90 /d15 Figure 6. .9 0.28.875 LOG(h") .1 0.30 0.5 0.70 4R 2 C1 A3 0. that a soil is internally unstable if it is plotted along the respective dotted line .0 0.99 0.99 0.99 2.99 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-34 .20 to 0.05 0.1 0. and internally unstable soil samples are represented by solid symbols (e.g.50 0. .25 to 0.02 0. .g. (1984) Kenney & Lau (1984.436 h' = d90 /d60 6 4 0.99 0.9 0.02 0.25 0.35 Average seepage gradient across embankment core (i) 0.01 0. Assess the probability that given the soil is internally unstable erosion by suffusion will begin (PSI) Assess the probability that given the soil is internally unstable suffusion will begin under the seepage gradient in the highly permeable layer using Table 6. P Sun (1989) data are not included in the logistic regression. The probability of suffusion and backward erosion should both be assessed and carried forward in the analysis.90 0.5 0. etc.Contours of the probability of internal instability for sand-gravel soils with less than 10% non-plastic fines passing 0.35 >0.).25 0. 8 P is the probability.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 10 Data source: Kenney et al.99 0. etc.6 .05 0. predicted by logistic regression.9 0.99 0.28 . P = exp(Z)/[1 + exp(Z)] Z = 3.591 h' + 2.95 0.05 0. . 85) Lafleur et al.30 to 0.

and silty sandy gravel soils which may be subject to collapse settlement even if the soils are non plastic. The procedure is: • Assess the thickness of the layer of soil which is poorly compacted (T p ). the average gradient through the core at the level of the high permeability layer. • Estimate the amount by which the layer may collapse (C F ) using Table 6. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-35 . The mechanism applies to such zones within the embankment. Then estimate the height of the gap which could result (G) from G = (T p ) x (C F ).4 to estimate the probability of initiation of erosion given this width crack. This is particularly so for dispersive soils. and use the method outlined in Section 5. • Assume G is the height of the crack which is formed. The minimum layer thickness adopted should be 300 mm (12 inches). To model this it is most practical to assume a crack is formed and to assess the likelihood of erosion initiating in the crack. This represents the scenario with the weight of the soil above being supported on non-collapsed soil adjacent.29. The soil collapses on saturation forming a crack or flaw in which the water flows. and the soil properties.6. This method should be also applied to silty sands. since they will erode rapidly in a crack. This is most likely where there is poorly compacted soil against a pipe but is possible within layers of soil. The mechanism is potentially of two types: • • The soil behaves as a series of clods with openings between the clods in which water passes. There may be a single layer or several layers. on the embankment/foundation contact. and adjacent to walls. around conduits.4 Probability Of Initiation Of Erosion In A Poorly Compacted or High Permeability Cohesive Soil Layer and in silt-sand-gravel soils in which collapse settlement may form a crack or flaw It is well documented that internal erosion and piping occurs in poorly compacted cohesive soils.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6.

93-95% standard dry density ratio. Soil placed and compacted by bulldozer. 95-98% standard dry density ratio. Well dry of optimum moisture content.Amount of collapse settlement which may occur on saturation versus compaction properties Amount of collapse settlement as a proportion of the layer thickness 0. moisture content 2% dry of optimum to 1% wet of standard OWC 0. 2% to 3% dry of standard OWC 0. at moisture contents dry of standard OWC Layers very poorly compacted.g. ≥ 98% standard dry density ratio.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6.g. no compaction by rollers. by horse and cart in old dams.02 to 0. no formal compaction ( e. < 93% standard dry density ratio.01 to 0.02 Compacted to e. < 90% standard dry density ratio.g. or by pushing into place by excavator or bulldozer or in very thick layers) No control on layer thickness. moisture content 2% dry of optimum to 1% wet of standard OWC All very well compacted to e.005 As above but with good documentation and records.g. but for the poorly compacted layer within 0. or rolled in thick layers beyond the capability of the roller Layer thickness at or beyond the limit of compaction equipment Dry of optimum moisture content Soil rolled in layers near the limit of the capability of the rollers. but for the poorly compacted layer within 0. 3% dry of standard OWC Layers poorly compacted. dry of standard optimum moisture content e.005 Soil compacted by suitable rollers in suitable layer thickness Around optimum moisture content Will not collapse.g. moisture content 2% to 3% dry of standard OWC Well compacted to e.005 Will not collapse. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-36 .g.05 Description of the method and degree of compaction of the core Soil placed with. dry of standard optimum moisture content e.29 .

Where there is specific information about frost effects for the dam being assessed. Use the method outlined in Section 5. These widths may be applied to the full depth of penetration of frost from Table 6. the average gradient through the core at the level of the high permeability layer. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-37 .Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6. (T p ). • Estimate the amount by which the layer may collapse (C F ) using Table 6.4 to estimate the probability of initiation of erosion given this width crack. and use the method outlined in Section 5. and the soil properties. • For cases where it appears that the soil around the pipe is well compacted assume and where crack widths <5 mm are calculated. assume a crack width of 5 mm to allow for possible shrinkage of the soil from the pipe during construction or in service.4 to estimate the probability of initiation of erosion given this width crack or flaw. and formation of ice lenses.30 should be used. The procedure is: • Assess the thickness of the layer of soil which is poorly compacted. • 6. There may be a single layer or several layers.6 Probability Of Initiation Of Erosion In A High Permeability Soil Due to Frost Action The procedure is: • • Assess the width of the frost induced crack or flaw using the procedure detailed below. In the absence of such data Table 6. To do this take account of the dimensions of the conduit and the trench in which it is placed.5 Probability Of Initiation Of Erosion In A Poorly Compacted or High Permeability Cohesive Soil Layer Around A Conduit The method assumes that the critical case is where poorly compacted soil surrounding the conduit collapses on saturation and a gap is formed.6. that information should be used by the risk analysis team to assess likely defect widths. Then estimate the height of the gap which could result (G) from G = (T p ) x (C F ).29 as a guide.6. and the soil properties.8 as the evidence is the ice lenses may be as thick at the base of freezing as at the surface. Assume G is the height of the crack which is formed. The effects of frost action are complex and include formation of cracks due to heave. The latter may melt in summer months and leave pathways in which erosion may initiate. the average gradient through the core at the level of the high permeability layer.

7 Allowance for the Presence of Unknown and Unpredictable Flaws in the Core of the Embankment Background 6. The flaw could also be due to unpredictable cracking or to a pathway created by installation of instrument cables or tubes. the probability of failure approaches that of the probability of the presence of a flaw. erosion initiates. it is possible that there are unknown and unpredictable flaws in the core which may. In these cases if the dam has a high consequence of failure. under some circumstances.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 Table 6. lead to initiation of erosion. a homogeneous dam). and the probability given there is a flaw.1 Even with well designed and constructed dams.10 28 22 18 13 12 Width of flaw -mm 20 10 5 2 Zero 6.7 24 21 16 12 10 (RFxLF) from Table 6. In most cases they should not contribute significantly to the probability of failure because other more predictable flaws will have a greater probability. The potential for unknown or unpredictable flaws in dams is always a possibility.g. even a very small probability of a flaw can be significant contributor to risk as there are little other defensive measures in place. the unknown and unpredictable flaws are most likely to be a poorly compacted layer or a layer that was allowed to crack or soften during construction. Below the POR there is a greater chance that the flaw has been detected and addressed as a known issue. It is therefore considered important that the responsible dam safety decision makers be aware of this when assessing the safety of the dam.30 .7. In the opinion of the development team. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-38 . The probability of these unknown flaws should be low as the risk analysis has already considered the more predictable modes of initiation. If a dam does not have filters or a transition zone (e. The likelihood of an unknown flaw is greater above the reservoir pool levels that the dam has been tested. for which there is good record of construction. For reasons of continuity the flaw is probably more likely to be in the upper part of the dam where the core width is smaller.Width of frost induced flaw versus (RFxLF) (RFxLF) from Table 6.

a review of the surveillance program at the end of the risk analysis is always a good idea. Surveillance should always be part of the monitoring program for the dam and all dam owners should be aware of this. Assign a probability of the presence of the flaw of 10-4 for reservoir levels above POR and 10-5 for reservoir levels below POR. it is recommended that the procedure detailed below be followed to check the effect on the risk analysis of the presence of unknown and unpredictable flaws. as pointed out by the late Ralph Peck in his paper “The Risk of the Oddball”. in the end.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 This section provides advice on what to do in these situations. This can be where the gradient of flow through the flaw will be greatest.7. or the filter or transition zone is so coarse that there is a significant probability of being in the continuing erosion range. and the assessed probability of a flaw for the initiation modes described in Sections 5 and 6 give very low probability of a flaw being present (≤ 10-4 for reservoir levels above POR and ≤ 10-5 for reservoir levels below POR) and the consequences of failure are high.7. occur in embankment dams. Develop a failure path for this scenario. However. 6. where there is no filter or transition zone. 6. • • • See the Supporting Document for the rationale behind these suggestions. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-39 .” As Peck summarizes.2 Recommended Procedure For dams which do not have filters or transition zones protecting all or part of the core. there remains the potential that even with the “… efforts of even the most experienced engineers the most significant potential failure mode may occasionally be overlooked. in their experience. Procedures are also suggested to handle failure modes not well covered in this methodology. Armed with possibly a fresh look at potential failure modes.3 Suggested size and location of the flaw and the probability of occurrence It is suggested that: • A 2mm wide flaw is assumed to be present in a location that is critical for each reservoir level under consideration. the only recourse is continuing surveillance. The risk analyst should remind the owner of this in the preparation of the report. Complete the evaluation of the event tree for this failure path. The development team has strived to address all failure modes that.

Report this separately to the other failure modes. and other factors considered relevant.4 Assessing the Probability of Failure if the Flaw Were to Exist and Reporting the Outcome of the Analysis Assess the probability of failure for each of the reservoir stages affected by the flaw by assuming the flaw is as described in Section 6. Make the case in the report of how likely (or unlikely) the dam has such a flaw. This discussion could focus on the amount and quality of construction control information. Whether or not this probability of failure is combined with the other failure modes to assess the overall probability of failure is for the Agency for whom the risk analysis is being carried out to decide.Probability of Initiation of Erosion in Poorly Compacted and High Permeability Zones in the Embankment SECTION 6 6.3. the climatic conditions at the dam site that may have been present during construction.7.7. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 6-40 .

including response of the piezometers to changes in reservoir level.1. then follow the procedures in Sections 7. then backward erosion is not possible and the probability of backward erosion may be taken as zero. and/or finite element seepage flow modelling as A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-1 . The layer does not have to be exposed to the ground surface downstream of the embankment.g. 7.1 Screening Check on Soil Classification Is the soil cohesive or cohesionless? If the soil is cohesionless or has a Plasticity Index ≤ 7. If the soil is cohesive with a Plasticity Index > 7.4 for suffusion. a toe ditch).2.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 7 Probability of Initiation of Erosion in a Soil Foundation 7.1 and 1.1 The steps to be followed are: a) Estimate the probability (P cl ) there is a continuous layer of cohesionless soil or soil with PI ≤ 7 from upstream of the embankment. If the seepage gradients are >4. This assessment should consider the potential for localised areas of a thinner confining layer within a distance of 2 x dam height downstream of the dam toe (e. consider suffusion for soils with a Plasticity Index ≤ 12. P cl is likely to be between 0. it may be overlain by a layer of cohesive soil. then the likelihood of backwards erosion and suffusion is zero under the seepage gradients which occur in foundations of a conventional dam. Consider if erosion may occur through a crack in the soil using the procedure detailed in Section 7.2 and 7. b) If there is no continuous layer such as shown in Figure 7. That is. This should be done using local gradients at the toe of the embankment from piezometer readings.0 for most situations c) Assess the probability of “heave” (P H ) or reaching the critical gradient (i cr ) which is a zero effective stress condition. understanding of the depositional conditions in which the soils were and deposited.2 Assessment Of The Probability Of Initiation Of Backward Erosion In A Layer Of Cohesionless Soil or Soil with Plasticity Index ≤ 7 In The Foundation (IM24) Description of Method 7. In some situations there will be some uncertainty and the probability of a continuous layer should be assessed from the geotechnical borehole data. piezometer data.5. at the toe of the embankment for the reservoir stage under consideration. to downstream of the embankment.3 to assess the probability of backward erosion and Section 7.

3.2.3. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-2 . the probability of initiation and progression for the reservoir level at which sand boils have been observed and above is assumed to be 1. (P I ) = (P cl ) x [(P H ) x (P IH ) + (1-(P H ) x (P INH )].3.3 and Table 7.2.0 e) Given that backwards erosion may initiate and progress without a heave condition being present. d) Estimate the probability of initiation and progression of backward erosion given heave has occurred (P IH ) using Section 7. or more approximately using the method shown in Section 7.3.1.2 and Table 7.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 described in Section 7.1 . estimate the probability of initiation and progression of backward erosion given heave has not occurred (P INH ) using Section 7. For situations where sand boils have been observed. f) Estimate the probability of initiation and progression of backward erosion (P I ) from the results of the assessments above. 3 1 3 CLAY SAND CLAY Figure 7.An example of a situation where there is no continuous layer of cohesionless soil in the foundation and backward erosion cannot occur.2 and Table 7.

Probability of Initiation of Erosion in a Soil Foundation SECTION 7 7.2 Estimation of the probability of heave or reaching the critical gradient (P H ) from piezometer data and/or seepage flow net models.2. (a) Calculate the factor of safety against heave or reaching the critical gradient from the following equations: σv u FUT = Where (1) total vertical stress at any point in the foundation – kN/m2 σv = u = pore pressure at the same point – kN/m2 or FUT = where γw hp h γ sat hp γw (2) unit weight of saturated foundation soil – kN/m3 γsat = = = unit weight of water – kN/m3 piezometric head – meters Figure 7.2 .Cross section of an embankment and dam foundation showing seepage flow net and definition of terms The alternative method for estimating the factor of safety is to consider the gradient of the flow net. If the gradient approaches the critical gradient. liquefaction and heave can be expected to occur. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-3 .

γ w ) / γ w (5) Where there is a layer of cohesive or other low permeability soil overlying the soil which may be subject to backward erosion. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-4 .9 0.50 Probability of heave (PH) 0. the most critical gradient usually occurs from the base of the lower permeability layer to the ground surface.80 FUC 2.0 0.20 1.23 1. Table 7.1.77 0.02 to 0.3 1.999 The selected probability should take account of the quality of the data on which the factors of safety are based.00 1.50 0.0 1. Estimated factor of safety against heave FUT 1.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 FUC = critical gradient actual gradient (3) For vertically up seepage.15 1.85 0.2 0.99 0.66 Seepage gradient at toe of embankment as a ratio of icr 0.1 – Estimation of the probability of heave or reaching the critical gradient (P H ) from the calculated factor of safety against heave.9 to 0.0 0. (b) Estimate the probability of heave from FUT or/and FUC from Table 7.12 1.05 1.59 0.3 1.87 1.5 0.1 to 0.005 to 0.99 0.05 to 0. the critical gradient i cr can be calculated from (Terzaghi 1960) i cr = (G – 1)/(1 + e) (4) where G = soil particle density (specific gravity) in tonnes/m3 e = void ratio or from • i cr = ( γ sat .7 1.92 0.05 0.

3 0.3 – Estimation of the probability of heave from embankment geometry and the foundation permeability ratio kh/ kv for situations where there is a confining layer in the foundation Foundation permeability ratio (kh/ kv) C 1 10 100 D 1 10 100 Factor of safety against heave from seepage analyses (FUT ) 0.99 0.0 1.9 0. The table is based on finite element seepage modelling.0 1.3 0.2.05 0.3.5 m below the crest level.009 0.2 – Estimation of the probability of heave from embankment geometry and the foundation permeability ratio kh/ kv for situations where there is no confining layer in the foundation Foundation permeability ratio (kh/ kv) 1 10 100 B 1 10 100 Factor of safety against heave from seepage analyses (FUT ) 1.3 An approximate method for assessing the probability of heave An approximate estimate of the probability of heave (P H ) can be obtained from Table 7.0 1.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 7.0 Seepage Model A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-5 .3 Probability of heave (PH) 1.05 0.2 and Table 7.02 to 0.45 1.999 Seepage Model A Table 7.4 0.2 0.1 to 0. Table 7.001 to 0.6 1.8 Probability of heave (PH) 0.2 0.005 to 0. Note that the seepage pressures are determined by the shape of the geometry. The approach is to assess which of the embankment and foundation arrangements shown in Figure 7. The relative thickness of the foundation to embankment height does not significantly affect the results.4 0.9 1. and the results apply to similar geometries regardless of scale.3 to Figure 7.3 0.0 1. The results for the approximate method apply for the condition where the reservoir level is 0.6 best models the dam under consideration.0 1.

5 .Seepage model D used to represent the cracked downstream clay (Maniam 2004).4 . A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-6 .Probability of Initiation of Erosion in a Soil Foundation SECTION 7 Figure 7. Figure 7.3 .Seepage model A of embankment dam and foundation (Maniam 2004).Seepage model C where foundation is overlain by low permeability clay (Maniam 2004) Figure 7.Seepage model B of embankment dam and foundation (Maniam 2004).6 . Figure 7.

This applies to sand boils with moving sand but not pin boils.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 7.0 for reservoir levels at or above the level at which sand boils have been observed. The average seepage gradient (i avf ) is defined as the hydraulic head difference divided by the seepage path length. If in doubt neglect the blanketing effect. In these sections the probability of progression only covers whether the backwards erosion process will progress within the developing pipes. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-7 .3. This allows for the head losses through the upstream clay layer.3. However whether backwards erosion will then progress to develop a pipe all the way from the downstream toe to upstream of the dam is related to the gradients under the main part of the dam.2 Estimating the probability of initiation and progression of backward erosion where heave or reaching the critical gradient is predicted or sand boils have been observed at higher reservoir levels (P IH ) If heave or reaching the critical gradient is predicted for the reservoir level under consideration it may be assumed that the probability of initiation of backwards erosion = 1.1 If sand boils have been observed it should be assumed that the probability of initiation and progression of backward erosion = 1.3. For practical purposes in most cases this can be taken as the average seepage gradient (i avf ) in the foundation layer beneath the dam at the reservoir stage under consideration. It does not include assessment of whether a roof will form and whether flow limitation may occur. These are covered in Section 11.2. 7. 7. For lower reservoir levels use the procedure in Section 7.0.3 Estimating the probability of backward erosion given heave has occurred Probability of initiation and progression of backward erosion if sand boils have been observed.7. In practical terms this gradient may be estimated as (H2 – H3)/L in Figure 7. However if there is upstream blanketing with lower permeability soil much of the head may be lost in seepage through this “blanket”.

This is discussed further in the Supporting document Section S 7. • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-8 . horizontal to vertical permeability ratio of the zone subject to backward erosion.2.4.2. • Estimate the point seepage gradient required to progress backward erosion using Figure 6. soil bentonite or concrete cut-off wall the gradient of interest is the average gradient in the soil subject to backward erosion from downstream of the cut-off wall to the toe of the embankment allowing for the effectiveness of the cut-off wall.6.3. and the geometry of the embankment and the foundation. This gives (i pmt ) Corrected .7 .Section through embankment and foundation showing definition of terms to estimate the average gradient in the foundation sand.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 Piezometric surface at base of clay layer Reservoir level L Piezometric surface at base of clay layer H1 H2 Clay H3 T Sand Figure 7. Correct this average gradient for the geometry. In the absence of more definite methods it is proposed that the Schmertmann (2000) method be used to make this assessment as follows: • Estimate the average seepage gradient (i avf ) through the cohesionless soil layer in the foundation beneath the central part of the dam (not at the toe where there are likely to be locally higher gradients) for the level for the reservoir stage under consideration. The other factors are the particle size characteristics and permeability of the soil. If there is a sheet pile.4 from Schmertmann (2000). • From the particle size distribution of the foundation material estimate the uniformity coefficient Cu = D 60 /D 10 . and grain size as detailed in the Supporting Information Section C6.

0 1.3.4.0 1.0 1. but their testing did not have a confining layer which is virtually a pre-requisite for heave to occur as required for situations satisfying the requirements of Section 7.0 1.05 1.0 1.5 0.0 7.0 1.5 0.01 0.0 0.0 0.0 1.1 1. To allow for these situations the following is suggested: • Estimate the average seepage gradient (i avf ) through the cohesionless soil layer in the foundation beneath the center of the dam at the level for the reservoir stage under consideration.0 1.4 Average seepage gradient in the foundation (iavf) 0.05 0.0 1. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-9 .05 0.75 1.0 1. so higher probabilities are assigned than for the case where heave has not occurred.γ w ) / γ w . For this latter case see Section 7.25 1.0 1.0 0. also estimate the critical gradient (i cr ) from i cr = ( γ sat .0 1. Table 7.5 1.0 1.001 0.4 – Estimation of the probability of initiation and progression of backward erosion in the foundation given heave is predicted (P IH ) Average seepage gradient required to initiate backward erosion (ipmt)Corrected from Figure 6.5 0.3.5 0.0 0.2 0.3 Estimation of the probability of initiation and progression of backward erosion (P INH ) for cases where heave is not predicted.0.0 1. The testing by Schmertmann (2000) and at Delft and field experience is that soils with uniformity coefficients (Cu) <6 experience initiation and progression of backward erosion at average seepage gradients lower than the critical gradient. The table allows for the assumption that for cases where heave is predicted to have occurred initiation has also occurred.2.75 1.0 0.0 1.05 0.0 0.0 1.5 0.01 0.0 1.0 1. Estimate the probability of initiation and progression from Table 7.1 0.0 1.0 1.2 0. Use i av and (i pmt ) Corrected as inputs.0 1.2 0.5 1.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 • For Cu> 6.0 2.3.2 0.05 0. Adopt this gradient if it is smaller than (i pmt ) Corrected . They recognise that the critical gradient is experienced at the toe.3.25 0.0 1.

Adopt this gradient if it is smaller than (i pmt ) Corrected .γ w ) / γ w .4. for permeability anisotropy Correction factor for high-permeability under layer Correction factor for density Adjustment for pipe inclination Correction factor for dam axis curvature Depth of piping sand layer. • Estimate the point seepage gradient required to initiate and progress backward erosion from Figure 6. Use i avf and (i pmt ) Corrected as inputs. The maximum point gradient (i pmt ). • Estimate the probability of initiation from Table 7.6.4) i pmt = CD CL CS CK Cz Cγ Cα CR D L • = = = = = = = = = = Correction factor for (D/L) Correction factor for total pipe length L Correction factor for grain size Correction factor.5.0. • Correct this average gradient for the geometry.2. in direction perpendicular to α (m) Direct (not meandered) length between ends of a completed pipe path. and grain size as detailed in the Supporting Information Section C6. also estimate the critical gradient (i cr ) from i cr = ( γ sat .Probability of Initiation of Erosion in a Soil Foundation SECTION 7 • From the particle size distribution of the foundation material estimate the representative uniformity coefficient Cu = D 60 /D 10 . measured along the pipe path (m) For Cu> 6. This gives (i pmt ) Corrected . The table allows for application of corrections recommended by Schmertmann (2000) which for many typical A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-10 . horizontal to vertical permeability ratio of the zone subject to backward erosion.where (i pmt ) Corrected = where (C D C L C S C K C z Cγ Cα )i pmt CR Maximum point seepage gradient needed for complete piping in the flume test based on the soil coefficient of uniformity Cu (from Figure 6. from downstream to upstream exit.4.

05 0.5 0.5 0. (P I ) = (P cl ) x [(P H ) x (P IH ) + (1-(P H ) x (P INH )].35 0.95 0.95 0.99 0.001 0.75 1.9 0.01 0.5 0.01 0.9 7. Table 7.2 0.4 are required to initiate erosion.25 0.93 0.003 0.999 0.0 2.5 – Estimation of the probability of initiation of backward erosion for cases where heave is not predicted (P INH ) Point seepage gradient required to initiate backward erosion (ipmt)Corrected from Figure 6.5 0.95 0.001 0.95 0.4 0.2 0.7 0.0 0.5 0.9999 0.9 0.35 0.05 0.93 0.1 0.995 0.2 0.0 Average seepage gradient in the foundation beneath the center of the dam ( iavf) 0.97 0.4 Estimation of the total probability of initiation and progression for this reservoir stage Estimate the probability of initiation and progression of backward erosion (P I ) from the results of the assessments above.5 0.5 0.0001 0.01 0.2 0.3.1 0.7 0.05 0.05 0.5 0.999 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-11 .9 0.75 1.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 situations for a dam on a relatively thin layer of erodible soil compared to the dam height indicate gradients significantly higher than those from Figure 6.25 0.99 0.99 0.9 0.

2. The question is best put in terms of the width of the crack.1 First: Estimate the probability (P cl ) there is a layer of soil beneath the embankment in which a continuous crack or interconnected pattern of cracks may exist.5 Estimation Of The Probability Of Initiation Of Erosion In A Crack In Cohesive Soil In The Foundation (IM26) Overall Approach 7. Multiply this value with the probability there is a continuous layer subject to suffusion in the foundation. It may be useful to consider the question for 1mm. 5 mm. and say 10mm wide cracks.4 Estimation Of The Probability Of Initiation Of Suffusion In A Cohesionless Layer In The Foundation (IM25) First Estimate the probability (P cl ) there is a continuous layer of cohesionless soil from upstream of the embankment.5 (b).4. An example is shown in Figure 5.2 Some Factors To Consider In This Assessment And Suggested Method For Estimating The Probability Of A Continuous Crack (a) For Cracking Due to Differential Settlement The situations which are likely to result in cracking in the foundation soils are essentially the same as those causing cracking low in the embankment due to differential settlement in the foundation. For example “what is the probability of a continuous crack or pattern of cracks 5mm wide”. to downstream of the embankment. it is suggested that if in using Section 5. 7.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 7. Second Use the method outlined in Section 6.5. The probability of initiation = (P cl ) x (P IC ).3 to assess the probability of cracking in the embankment due to differential settlement in the foundation. Cracking may be the result of differential settlement in the foundation. 7.6.5.3 to estimate the probability of suffusion. and much lower likelihood of wider cracks. Second: Estimate the probability erosion will initiate in the cracks (P IC ) using Section 5. the presence of cracking in the A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-12 . Given this. or desiccation cracks in the foundation soil which was not stripped from the foundation. as there may be a significantly high likelihood of narrow cracks.

Whether the cracking is likely to persist below the level of the cut-off for the dam. Whether desiccation cracking is evident in the soil surrounding the dam. then cracking in the foundation due to differential settlement may be assumed to be negligible.11 and 5.3 is for features such as those in Figure 5.5 (a) and (c) which will cause cracking near the crest of the dam. The continuity is likely to be linked to the width of the cut-off.Probability of Initiation of Erosion in a Soil Foundation SECTION 7 embankment is likely to be on the foundation/embankment contact. This can be assessed using observed desiccation crack depths for the site. with wide cut-offs being less likely to have continuous cracking than narrow cut-offs.25 to estimate the depth and width of cracking.2. • • • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 7-13 . (b) For Cracking Due to Desiccation Some factors to consider in assessing the likelihood of continuous or interconnected cracking include: • Whether the soils in the foundation on which the dam was built are susceptible to desiccation in the climatic conditions at the site. and knowing the depth of the cut-off below the ground surface.5 and Tables 5.24 and 5. This can be assessed using Section 5. then this probability also be assigned to the probability of cracking in the cohesive soil in the foundation beneath the embankment.12 considering the foundation soil as the “core material”. If the assessment of cracking in Section 5.2. or using Tables 5.

• Assess the probability of one or more continuous in filled or open defects or solution features in the rock in the foundation beneath the core of the embankment for each of the types of geological features. For each type of geological feature.1.2). • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-1 . how far these occur into the foundation.7).4). While continuity of these features is very important it is unlikely there will be good quality information on this for many dams so the weighting given to it is not high for these cases. The relative importance factors are selected taking account of the data which is likely to be available.12). It is anticipated that for some older dams there may be virtually no such data. Solution features for rock subject to solution such as limestone. The second is based on observations and site investigation data of which there will be greatly varying amounts for different dams.6).3). there are two parts to the assessment. gypsum and salt (Section 8.(Section 8. Assess the probability that the defects are in filled (Section 8. the first is based on the geology and topography of the dam site. Defects associated with landslides and faults and shear zones. The two estimates are combined by a weighted average depending on the detail of the investigation and construction data that is available with greater weighting to the second method where there is good quality data available (Section 8. • Assess the width of the open or in filled defects or solution features. and how their widths vary with depth in the foundation (Section 8. The geological processes considered are: • • • • Defects related to stress relief effects in valley sides (Section 8. An outline of the overall approach is summarized in the flow chart shown in Figure 8.5) Tables have been produced to evaluate the probabilities for assessing the presence of continuous open or in filled defects and solution features caused by these geological processes. This information will be available for all dam sites.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8 Probability of the Presence of Open or In Filled Defects in Rock Foundations 8. dolomite. Defects related to stress relief effects in the valley floor – valley bulge and rebound (Section 8.1 Overall Approach The framework considers the geological processes which can lead to the formation of open or in filled defects or solution features in rock. The steps to assess the probability such defects and solution features exist in the dam foundation below the core are as follows.

and other geological feature) Make estimates based on. Probabilities are estimated for three widths at ground surface level 5mm to 25mm. foundation grouting and cut-off foundation surface treatment. Prepare sketches showing the defects in relation to the cut-off foundation beneath the core. Figure 8. valley bulge. 25mm to 100mm and >100mm) Combine the estimates using a weighted average Is the defect open or in filled? For open defects. grouting is generally assumed to be ineffective. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-2 . width. is grouting ineffective? For in filled defects. • Geologic/topographic factors (stress relief defects). solution features. and • Observations and investigation factors. depth and spatial distribution of continuous open defects and solution features in rock foundations. general foundation under the shoulders.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Probability of one or more continuous open or in filled defects (stress relief valley sides. Does erosion of the infill initiate? Does erosion of the infill continue? Combine the probability estimates for open defects and in filled defects that have eroded Prepare a summary of the probability of occurrence of open defects and solution features categorized by width. depth and spatial occurrence in the embankment foundation.1 – Flow chart for estimating the probability.

unravelling or sinkhole development in the embankment. slope instability.9) Assess the probability that erosion of the infill will initiate given grouting and cut-off walls are not effective (Section 8. This is the end product of what is covered in this Section. (Section 8. For in filled defects. identify the potential failure paths.10). Assess the probability of cut-off walls not being effective in cutting off open or in filled defects or solution features. their width. spatial distribution in the foundation. Combine the probabilities for open defects and in filled defects which will potentially erode. (Section 8. Then for each potential failure path and breach mechanism. Unravelling of the embankment due to seepage exiting into the downstream shell.(Section 9). Assess the probability that erosion of the infill in defects or solution features will continue (Section 8. and potential breach modes due to the open defects and solution features. Possible failure paths are. In particular identify features which will be in contact with the core at the base of the cut-off and in the sides of the cut-off trench. The probability of progression (Section 11).12) Describe the open and in filled defects and solution features. assess: a) b) The probability of initiation of erosion of the core into the open defect or scour of the core at the defectcore interface.11). a) b) c) d) Gross enlargement of the defect – this can release the storage but is usually not capable of breaching the dam unless the feature is large relative to the size of the dam Slope instability of the embankment due to increased pore pressures caused by seepage up into the downstream shell. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-3 . Initiation of internal erosion of the embankment at or into the foundation by backward erosion piping or scour followed by gross enlargement. depth. • • • • • Given the defects or solution features occur below the core of the embankment. and how these relate to the cut-off and general foundation of the embankment beneath the core.8). it is assumed that grouting is not effective.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 • Assess the probability that grouting has not been effective in cutting off the open defects (Section 8.

intervention and repair (Section 12). This is done in two ways: • • Based on the geology and topography of the site. grouting or other treatment. That is considered in subsequent steps (Sections 8.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 c) d) The probability of detection. It does not consider whether these are cut off by the foundation excavation. and water pressure tests and grouting during site investigations and construction. This data will be available for all dams. The weighted average of these estimates is carried forward in the analysis. For solution features it is the maximum width which may be at the surface or at depth.8 and 8.9). 8. drilling. Based on observations. • • • 5mm to 25 mm 25mm to 100mm >100mm Open or in filled defects of each width may all be present and each should be considered. These have been determined because of their likely impact on the probability of breach of the embankment. There will be varying amounts of this data for each dam.2.1 This section assesses the probability of an open or in filled defect such as a joint or joint set or bedding surface being continuous in the dam foundation from upstream of the core of the embankment to downstream of the core.2 Probability of one or more continuous in filled or open defects in the rock in the foundation beneath the embankment (IM27) Overview of method 8. The probability of breach (Section 13). as follows. The weighting relates to the detail of the investigations and construction data available. The probabilities are estimated for features of 3 different widths. The widths for stress relief features are at the ground surface not accounting for the depth of the cut-off of the embankment. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-4 .

Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8. Topography.1. the risk analysis team should assess which of the descriptions in Table 8.2 Estimation of the probability of one or more continuous in filled or open defects in the rock foundation beneath the embankment based on geologic and topographic data The defects are stress relief joints.3 the probabilities for each size defect are independent of each other because all defect sizes may be present. The topography which is likely to lead to stress relief features. sheet joints and bedding surface partings. This is dependent on the depth of valley and steepness of valley sides. The other geological environments where stress relief defects are common are massive rocks such as granite.1 best suits the geology of the site. The probabilities may total > 1. It is known that inter-bedded weak and strong rocks such as inter-bedded shale and sandstone commonly result in open joints due to differential strains due to stress relief on valley formation. • Geological environment.1 to 8. • • Use Tables 8. and these strains being concentrated in bedding surface shears in the weaker rock.0 because they are not conditional probabilities.3 to estimate the probability of a continuous in filled or open defect in the rock in the foundation beneath the embankment. It does not rely on mapping of the dam foundation during investigations and construction. Note that in Table 8. These are commonly called sheet joints. The major variables affecting the likelihood of these being in the dam foundation. Where the geology of the site is not covered in Table 8. Continuity and orientation of mapped defects in the exposure of rock in the abutments.2. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-5 . and their relative importance are.

(a) Some basalts (e.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8. with thick beds of hard rocks MML from Table 8.2 matrix (1) Discontinuous.2 matrix N from Table 8.2 matrix Continuous defects. if inter-bedded with weak tuff. steeply dipping More Likely (3) Granite Basalt columnar Rhyolite. and some ignimbrites) may classify as “Much more likely” A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-6 .2 matrix ML from Table 8. sub-parallel to valley Interconnected defects. subparallel to valley Geological environment likely to give open or in-fill continuous features Topography of the dam site (Steepness and valley depth) Continuity and orientation of defects from surface mapping of the dam abutments (2) LL from Table 8.g.1 .Factors influencing the likelihood of a continuous in filled or open defect in the rock in the foundation beneath the embankment Influence on Likelihood Factor Relative Importance of Factor (3) Less Likely (1) Shale not containing other rocks (sub-horizontal bedded) Uniform sandstone (sub-horizontal bedded) Neutral (2) Schistose. inter-bedded thin beds of tuff or other soft rocks. subparallel to valley Notes. ignimbrite(a) Schistose parallel to abutment Thinly interbedded sedimentary (a) Much More Likely (4) Sedimentary – inter-bedded thin beds of mud rocks and thick beds of sandstone or limestone/dolo mite or conglomerate Volcanics. oblique to valley Discontinuous.

001 0.Factors influencing likelihood for topography Average Abutment Slope (degrees) (a) (b) Valley Depth <20 < 30m (<100 ft) 30 – 100m (100-300 ft) 100 – 300m (300-1000 ft) >300m (> 1000 ft) Notes (a) Use average overall slope except for slopes with large colluvium deposits overlying the rock surface. (b) For slopes with more than 6m (20 ft) high cliffs.02 0.005 0.0005 >100 mm 0.01 0. adopt a category not lower than ML.5 0.0001 RF x LF 6 9 11 13 18 24 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-7 .005 0.3 .001 0.2 .01 0.1 0.0002 0.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8.Probability of a continuous in filled or open feature in the rock foundation (Relative importance factor (RF) x (Likelihood factor beneath the embankment versus ∑ (LF)) Defect width 5-25 mm 0. 20-45 LL N ML ML 45-60 N ML MML MML >60 ML MML MML MML LL LL LL N Table 8. in which case use the slope of the rock surface.05 0.005 0.5 0.2 25-100 mm 0.9 0.001 0.

5 the probabilities for each size defect are independent of each other because all defect sizes may be present. and >100mm defects It is expected that there will be different probabilities for each defect size.5. Estimate the probability of one or more continuous in filled or open defect from Tables 8. The probabilities may total > 1. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-8 . sheet joints and bedding surface partings.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8.4 Effects Of Blasting On The Foundation The potential for blasting of the rock foundation for the cut-off or. 8. Construction photographs are the best guide as to whether such features may exist. This can only be assessed on a case by case approach. Do this for: • • • 5mm to 25 mm defects 25mm to 100mm defects.4 and 8. for example to form a trench into which the diversion conduit is placed. and their likely continuity and opening. to lead to open defects should be assessed.2.2.0 because they are not conditional probabilities. Note that in Table 8.3 Estimation of the probability of one or more continuous in filled or open defects in the rock foundation beneath the embankment based on site investigations and construction data The defects are stress relief joints.

tunnels. very high grout takes. down hole imaging.4 . steeply dipping Granite Basalt(d) Rhyolite.Factors influencing the likelihood of a continuous in filled or open defects in the rock in the foundation beneath the embankment Relative Importance of Factor (3) Influence on Likelihood Less Likely (1) Data shows no open or in filled defects aligning spatially in upstreamdownstream direction Neutral (2) Data shows isolated and/or discontinuous open or in filled defects aligning in upstreamdownstream direction. shears.g. interbedded thin beds of tuff or other soft rocks. Data would include the information noted in Notes (a) and (b). if inter-bedded with weak tuff. high water losses in drilling. or a tortuous path Site investigation and construction data indicates circumstantial evidence of open or in filled defects of this size are unlikely to be present More Likely (3) Data shows some features or single continuous open or in filled defects partially aligning upstreamdownstream direction Site investigation and construction data (b) indicates circumstantial evidence open or in filled defects of this size are present Much More Likely (4) Data shows many or a single open or in filled continuous defects spatially aligning upstream – downstream Factor Spatial continuity of the open or in the foundation (c) Site investigation and construction data indicating open or in filled defects of this size are/are not present Geological environment likely to give open or in-fill continuous features (2) Good quality site investigation and construction data indicates open or in filled defects of this size are very unlikely to be present Good quality site investigation and construction data(a) indicates open or in filled defects of this size are present (1) Shale not containing other rocks (subhorizontal bedded) Uniform sandstone (subhorizontal bedded) Schistose. drill rods dropping For example. drill rods dropping. very high grout takes. with thick beds of hard rocks Notes (a) (b) (c) (d) For example records/photographs from excavations. high leakage rate in dam foundation. ignimbrite(d) Schistose parallel to abutment Thinly interbedded sedimentary Sedimentary – inter-bedded thin beds of mud rocks and thick beds of sandstone or limestone/dolomite or conglomerate Volcanics.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8. and some ignimbrites may classify as “Much more likely”. high water losses in drilling. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-9 . Spatial alignment is likely to be able to be related to geologic features such as stress relief joints. low piezometric levels in the abutments. etc Some basalts (e.

The probabilities are estimated for features of 3 different widths.3. or bedding surface being continuous in the foundation from upstream of the core of the embankment to downstream of the core.5 0. That is considered later.02 0. This data will be available for all dams.001 0.1 0. and water pressure tests and grouting during site investigations and construction.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8.9 0.0005 >100 mm 0.1 0.1 This section assesses the probability of an open or in filled valley bulge or rebound feature such as a joint or joint set.2 25-100 mm 0. Based on observations.001 0.01 0. There will be varying amounts of this data for each dam. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-10 .2 0.0002 0.5 0.0001 RF x LF 6 9 11 13 18 24 8.3 Probability of one or more continuous open or in filled valley bulge or rebound features in the rock in the foundation beneath the embankment Overview of method 8. The widths are: • • • 5mm to 25 mm 25mm to 100mm >100mm This is done in two ways: • • Based on the geology and topography of the site.001 0. The widths are at the ground surface not accounting for the depth of the cut-off of the embankment.Probability of a continuous in filled or open features in rock in the foundation beneath the embankment versus ∑ (Relative importance factor (RF) x (Likelihood factor (LF)) Defect width 5-25 mm 0. It does not consider whether these are cut off by the foundation excavation. grouting or other treatment. drilling.01 0.005 0. thrust fault. These have been determined because of their likely impact on the probability of breach of the embankment.5 .005 0.

and in particular where there are inter-beds of strong rock such as sandstone.0 because they are not conditional probabilities. For other geological environments assume the probability of valley bulge or rebound defects is negligible. with weaker rocks such as shale or claystone on which differential movements due to stress relief are concentrated. The presence of beds of stronger rock underlain by weaker rock in the valley floor and the slenderness of the strut as characterised by the relative thickness of the strong rock bed compared to the valley width.7 the probabilities for each size defect are independent of each other because all defect sizes may be present. Use Tables 8. The topography which is likely to lead to stress relief features. This is dependent on the depth of valley and steepness of valley sides. The probabilities may total > 1.7 to estimate the probability a continuous in filled or open valley bulge or rebound feature in the rock in the foundation beneath the embankment.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 The weighted average of these estimates is carried forward in the analysis. The weighting relates to the detail of the investigations and construction data available. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-11 . • Geological environment.2 Estimation of the probability of one or more continuous in filled or open valley bulge or rebound features in the rock foundation beneath the embankment based on geologic and topographic data The major variables and their relative importance are. • • These assessments do not rely on mapping of the dam foundation during site investigations and construction.6 and 8. Valley bulge or rebound features only occur in near horizontally bedded rocks sedimentary rocks. Note that in Table 8. 8. Topography. Potential for buckling or strut shear.3.

7 .01 0. dolomite or conglomerate (Valley floor width) / (strut thickness) >4 Much More Likely (4) Thin beds of claystone or shale inter-bedded with massive beds of sandstone.05 0.001 0.2 25-100 mm 0.2 matrix N from Table 8.0002 0.6 .2 matrix MML from Table 8. limestone.Probability of a continuous in filled or open valley bulge or rebound feature in (Relative importance factor (RF) rock in the foundation beneath the embankment versus ∑ x (Likelihood factor (LF)) Feature width 5-25 mm 0.1 0.5 0.02 0.2 0. dolomite or conglomerate (Valley floor width) / (strut thickness) >2 More Likely (3) Thin beds of siltstone interbedded with massive beds of sandstone.2 matrix ML from Table 8.5 0. dolomite or conglomerate (Valley floor width) / (strut thickness) >8 Geological environment likely to give valley bulge or rebound features Potential for buckling or strut failure relating to slenderness of the strut Topography of the dam site (Steepness and valley depth) (2) (Valley floor width) / (strut thickness) <1 (1) LL from Table 8.001 0.001 0. limestone.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8.005 0.0005 >100 mm 0.2 0.Factors influencing the likelihood of a continuous in filled or open valley bulge or rebound feature in the rock in the foundation beneath the embankment Influence on Likelihood Factor Relative Importance of Factor (3) Less Likely (1) Shale or siltstone not containing other rocks Uniform (a) sandstone Neutral (2) Thick beds of siltstone or shale interbedded with massive beds of sandstone. Table 8. limestone.2 matrix Notes (a) Likelihood of valley bulge feature is negligible in these geological environments.005 0.9 0.0001 RF x LF 6 9 11 14 18 24 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-12 .

9. and >100mm defects It is expected that there will be different probabilities for each defect size. The probabilities may total > 1.0 because they are not conditional probabilities.3. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-13 .3 Estimation of the probability of one or more continuous in filled valley bulge or rebound feature in the rock foundation beneath the embankment based on site investigations and construction data Estimate the probability of one or more continuous in filled or open defect from Tables 8.8 and 8. Do this for: • • • 5mm to 25 mm defects 25mm to 100mm defects.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8.9 the probabilities for each size defect are independent of each other because all defect sizes may be present. Note that in Table 8.

down hole imaging.Factors influencing the likelihood of a continuous in filled or open valley bulge or rebound feature in the rock in the foundation beneath the embankment Influence on Likelihood Factor Relative Importance of Factor (3) Less Likely (1) Data shows no open or in filled features aligning spatially in upstreamdownstream direction Neutral (2) Data shows isolated and/or discontinuous open or in filled feature aligning in upstreamdownstream direction More Likely (3) Data shows some features or single open or in filled continuous feature partially aligning upstreamdownstream direction Site investigation and construction data (b) indicates circumstantial evidence open or in filled defects of this size are present Thin beds of siltstone interbedded with massive beds of sandstone.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8. limestone. dolomite or conglomerate Much More Likely (4) Data shows many or a single open or in filled continuous feature spatially aligning upstream – downstream Spatial continuity of the valley bulge or rebound features (c) Site investigation and construction data indicating open or in filled features are/are not present (2) Good quality site investigation and construction data indicates open or in filled defects of this size are very unlikely to be present Site investigation and construction data indicates circumstantial evidence of open or in filled defects of this size are unlikely to be present Thick beds of siltstone or shale interbedded with massive beds of sandstone.8 . Spatial alignment is likely to be able to be related to geologic features such as stress relief joints. drill rods dropping. dolomite or conglomerate For example records/photographs from excavations. high water losses in drilling. shears. limestone. dolomite or conglomerate Good quality site investigation and construction data(a) indicates open or in filled defects of this size are present Geological environment in relation to the presence/absen ce of valley bulge or rebound features Notes (a) (b) (c) (1) Shale or siltstone not containing other rocks Uniform sandstone (a) Thin beds of claystone or shale interbedded with massive beds of sandstone. high leakage rate in dam foundation. etc. Data would include the information noted in Notes (a) and (b). low piezometric levels in the abutments. tunnels. drill rods dropping For example. very high grout takes. high water losses in drilling. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-14 . limestone. very high grout takes.

001 0.01 0.4.9 . Approach This section assesses the probability of an open or in filled solution feature in the dam foundation continuous from upstream of the core of the embankment to downstream of the core. gypsum and salt.5 0. The widths are the maximum which may not be at the ground surface.001 0.005 0.0002 0.1 Screening This Section applies only to rock foundations subject to solution such as limestone.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8.2 0. The probabilities are estimated for solution features of 3 different widths as follows: • • • 5mm to 100mm 100mm to 300mm >300mm These widths have been determined because of their likely impact on the probability of breach of the embankment. grouting or other treatment. dolomite.02 0.4 Probability of one or more continuous open or in filled solution features in the rock in the foundation beneath the embankment Overview of method 8.9 0.5 0.0001 RF x LF 6 9 11 14 18 24 8.1 0.Probability of a continuous in filled or open valley bulge feature in rock in the (Relative importance factor (RF) x foundation beneath the embankment versus ∑ (Likelihood factor (LF)) Feature width 5-25 mm 0. That is considered later.001 0.2 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-15 .05 0. They do not consider whether these are cut off by the foundation excavation.005 0.2 25-100 mm 0.0005 >100 mm 0.

• • These assessments do not rely on mapping of the dam foundation during investigations and construction.4. and water pressure tests and grouting during site investigations and construction. Use Tables 8. • Geological environment. and the presence or absence of regional defects such as joints and faults. This data will be available for all dams.11 to estimate the probability a continuous in filled or open solution feature in the rock in the foundation beneath the embankment. The probabilities may total > 1. The presence or absence in the region and dam site of karst and solution features. The weighting relates to the detail of the investigations and construction data available. Based on observations. There will be varying amounts of this data for each dam.2 Estimation of the probability of one or more continuous in filled or solution features in the rock foundation beneath the embankment based on geologic and topographic data The major variables and their relative importance are.0 because they are not conditional probabilities. This is dependent on the depth of valley and steepness of valley sides. drilling. Note that in Table 8.11 the probabilities for each size defect are independent of each other because all defect sizes may be present. The topography which is likely to lead to stress relief defects such as joints and bedding partings along which solution is likely to initiate. 8.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 The probabilities are estimated in two ways: • • Based on the geology and topography of the site. The weighted average of these estimates is carried forward in the analysis.10 to 8. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-16 . Continuity and orientation of mapped valley stress relief or regional defects in the exposure of rock in the dam abutments.

oblique to valley A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-17 .2 matrix Regional defects not evident on good quality air photographs and surface geological mapping N from Table 8.g.g.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8.2 matrix Regional defects evident on good quality air photographs and surface geological mapping Continuous defects. sub-parallel to valley ML from Table 8.g. caves and sinkholes are absent) Low leakage in the foundation Neutral (2) Regional and dam site geology indicates solution features are unlikely (e. subparallel to valley Geological environment relating to development of solution features Presence of stress relief defects based on the topography of dam site And/or Regional Defects Continuity and Orientation of the stress relief and/or regional defects from surface mapping of the dam abutments in which solution features may develop (2) LL from Table 8.g. only isolated caves and sinkholes).2 matrix Regional defects not evident on poor quality air photographs and surface geological mapping Discontinuous. a few caves and sinkholes Moderate to high leakage in the foundation Much More Likely (4) Regional and dam site geology and observations in outcrop indicates solution features are very likely (e.2 matrix Regional defects evident on poor quality air photographs and surface geological mapping Interconnected defects.Factors influencing the likelihood of a continuous in filled or open solution feature in the rock in the foundation beneath the embankment Influence on Likelihood Factor Relative Importance of Factor (3) Less Likely (1) Regional and dam site geology indicates solution features are very unlikely (e. many caves and sinkholes). Low leakage in the foundation More Likely (3) Regional and dam site geology indicates solution features are likely (e.10 . High leakage in the foundation MML from Table 8. subparallel to valley (1) Discontinuous.

9 0.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8.5 0.05 0. not isolated caves.05 0.02 0.01 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-18 .3 Estimation of the probability of one or more continuous in filled or open solution features in the rock foundation beneath the embankment based on site investigations.1 100-300 mm 0.5 0.11 .02 0.005 >300 mm 0.1 0.4.12 and 8.2 0. It is important to consider continuous features.05 0.3 0.13.1 0.001 RF x LF 6 9 11 14 18 24 8. construction and monitoring data Estimate the probability of one or more continuous in filled or open solution features from Tables 8.Probability of a continuous in filled or open solution features in rock in the (Relative importance factor (RF) x foundation beneath the embankment versus ∑ (Likelihood factor (LF)) Solution feature width or diameter. and >300mm solution feature It is expected that there will be different probabilities for each solution feature size.005 0.01 0. 5-100mm 0. Do this for: • • • 5mm to 100 mm solution feature 100mm to 300mm solution feature.

Low piezometric levels in the abutments.g. Low piezometric levels in the abutments. only isolated caves and sinkholes) More Likely (3) Site investigation and construction (b) data indicates circumstantial evidence open or in filled defects of this size are present. etc.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8. Moderate piezometric levels in the abutments.Factors influencing the likelihood of a continuous in filled or open solution feature in the rock in the foundation beneath the embankment Influence on Likelihood Factor Relative Importance of Factor (3) Less Likely (1) Good quality site investigation and construction data indicates open or in filled defects of this size are very unlikely to be present. (c) Data would include the information noted in Notes (a) and (b). very high grout takes.g. tunnels.g. High leakage in the foundation Site investigation. a few caves and sinkholes) Much More Likely (4) Good quality site investigation and construction (a) data indicates open or in filled defects of this size are present. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-19 . high water losses in drilling. high water losses in drilling. many caves and sinkholes) Notes (a) For example records/photographs from excavations. very high grout takes. Low leakage in the foundation Neutral (2) Site investigation and construction data indicates circumstantial evidence of open or in filled defects of this size are unlikely to be present. Moderate to high leakage in the foundation Data shows some features or single open or in filled solution features partially aligning upstreamdownstream direction Regional and dam site geology indicates solution features are likely (e. shears. High piezometric levels in the abutments.g.12 . drill rods dropping. caves and sinkholes are absent) Data shows many or a single open or in filled solution features spatially aligning upstream – downstream Geological environment relating to development of solution features (1) Regional and dam site geology indicates solution features are very likely (e. Low leakage in the foundation Data shows isolated open or in filled solution features aligning in upstreamdownstream direction Regional and dam site geology indicates solution features are unlikely (e. down hole imaging. monitoring and construction data indicating open or in filled solution features are/are not present Spatial continuity of solution features (c) (2) Data shows no open or in filled solution features aligning spatially in upstreamdownstream direction Regional and dam site geology indicates solution features are very unlikely (e. low piezometric levels in the abutments. high leakage rate in dam foundation. Spatial alignment is likely to be able to be related to geologic features such as stress relief joints. drill rods dropping (b) For example.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-20 .02 0.005 0. grouting or other treatment.9 6 9 11 14 18 24 RF x LF 8.13 . That is considered later. The assessment is based on the evidence that such features are present based on observations.001 0. such as joints and bedding surface partings.3 0.1 0. drilling. faults and shears Overview of method 8.1 This section assesses the probability of an open or in filled feature associated with geological processes other than stress relief or solution features being present in the foundation of the dam and continuous from upstream of the core to downstream of the core. It does not consider whether these are cut off by the foundation excavation.Probability of a continuous in filled or open solution features in rock in the (Relative importance factor (RF) x foundation beneath the embankment versus ∑ (Likelihood factor (LF)) 0. The probabilities are estimated for features of 3 different widths as follows: • • • 5mm to 25mm 25mm to 100mm >100mm These have been determined because of their likely impact on the probability of breach of the embankment.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8.5.5 Probability of one or more continuous open or in filled features associated with other geological features such as landslides. The widths are at the ground surface not accounting for the depth of the cut-off of the embankment. These include defects related to landslides in rock. and water pressure tests and grouting during site investigations and construction. and fault or shear zones.

drill rods dropping. drill rods dropping (b) For example. and not isolated features. very high grout takes. high leakage rate in dam foundation. tunnels.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-21 . high water losses in drilling.14 . Spatial alignment is likely to be able to be related to geologic features such as stress relief joints.14 and 8. monitoring and construction data indicating open or in filled geological features are/are not present Spatial continuity of the geological (c) features (2) Geological environment relating to development of such features (2) Notes (a) For example records/photographs from excavations. shears. down hole imaging. Table 8. very high grout takes. (c) Data would include the information noted in Notes (a) and (b).Factors influencing the likelihood of a continuous in filled or open geological feature in the rock in the foundation beneath the embankment Relative Importance of Factor (2) Influence on Likelihood Less Likely (1) Good quality site investigation and construction data indicates open or in filled defects of this size are very unlikely to be present Data shows no open or in filled defects aligning spatially in upstreamdownstream direction Regional and dam site geology indicates the geological features are very unlikely Factor Neutral (2) Site investigation and construction data indicates circumstantial evidence of open or in filled defects of this size are unlikely to be present Data shows isolated open or in filled defects aligning in upstreamdownstream direction Regional and dam site geology indicates the geological features are unlikely to be present More Likely (3) Site investigation and construction (b) data indicates circumstantial evidence open or in filled defects of this size are present Data shows some features or single open or in filled defects partially aligning upstreamdownstream direction Regional and dam site geology indicates the geological features are likely to be present Much More Likely (4) Good quality site investigation. Do this separately for each defect size.15. low piezometric levels in the abutments. etc. 25mm to 100mm features. It is expected that there will be different probabilities for each defect size. monitoring and (a) construction data indicates open or in filled defects of this size are present Data shows many or a single open or in filled defects spatially aligning upstream – downstream Regional and dam site geology indicates the geological features are known or very likely to be present Site investigations. and >100mm features. 5mm to 25mm features. It is important to consider continuous features relative to the width of the embankment.5.2 Estimation of the probability of one or more continuous in filled or open other geological features in the rock foundation beneath the embankment based on site investigations and construction data Estimate the probability of one or more continuous in filled or open defect from Tables 8. high water losses in drilling.

The size of the defects given in the preceding sections are for the widths measured at the original ground surface (i.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8. estimate the width of the open or in filled defects below the original ground surface based the available site investigations data.001 0. For massive rocks subject to sheet jointing the sheet joints may be associated with weathering to soil strength to a width of up to 300mm (12 inches) but are seldom open more than 50mm (2 inches).3 0.005 0.1 Extent of occurrence and width of defects associated with stress relief defects in the valley sides For stress relief defects in valley sides. 8. This may also be used to supplement the site investigations and construction data. The issue of whether the extent of the excavation for the foundation and cut-off trench was sufficient to remove these features from the foundation beneath the core of the dam or whether they exist in the sides of the cut-off trench is covered in Section 8.6 Width and extent of open or in filled defects or solution features in the embankment foundation This section estimates the extent that the open or in-filled defects or solution features extend into the foundation beneath the embankment and the relation between the width of the defects and features to the distance from the original ground surface at the dam site. Assume that the maximum opening is for the first joint from the natural surface.Probability of a continuous open or in filled features in the rock (Relative importance factor (RF) x foundation beneath the embankment versus ∑ (Likelihood factor (LF)) 0. before construction of the dam) or for solution features.6. Assume that the defect openings for the jointed rocks in Figures 8. and construction mapping and grouting records.2.9 6 9 11 14 18 24 RF x LF 8.12. which may be below the ground surface. If there is little or no such data make the assessment based on the assumed geometry of stress relief defects shown in Figure 8.2 (a) and (b) and similar conditions vary linearly with depth and horizontal distance from the ground surface.15 .02 0.e.1 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-22 . the maximum width. The assessment depends on the type of defect.

Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Stress relief joints open for 3 or 4 joints or for a width of up to 30% of valley depth whichever is larger Stress relief joints open for 3 or 4 joints or for a width of up to 30% of valley depth whichever is larger (a) Horizontally bedded sedimentary rocks. Figure 8. (Figures from Fell et al 2004). A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-23 .2 – Assumed distribution of defect depths for defects related to stress relief effects in the valley sides – (a) for interbedded sedimentary rock.

Figure 8.2 – Assumed distribution of defect depths for defects related to stress relief effects in the valley sides – (b) Jointed rocks such as thinly bedded sandstones. some granite). granite or basalt. jointed granites and basalt (see supporting document for explanation of numbers).g. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-24 . (b) Uniform jointed rock such as thinly bedded sandstone. Stress relief joints (sheet joints) to a distance normal to the slope of up to 30% of the valley depth (c) Massive rocks such as some granite.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Stress relief joints to a distance normal to the ground surface of up to 30% of the valley depth. Figures from Fell et al 2004). and (c) for massive rocks (e.

8. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-25 . particularly in relation to the historical groundwater levels which may have been lower in the geological history of the valley.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8.2 can be used as a guide. • How the solution features were formed – stress relief effects or regional effects and the extent these can be expected to occur . It should be noted that these features are known to have occurred to depths of up to 15 meters (50ft) below the valley floor. The depth and spatial distribution of solution features should be assessed based on the following factors. In the absence of other information it should be assumed these features exist to below the level of the strut (massive stronger bed).3 Width and extent of solution features For solution defects. The assessment of the size of the defects is considered in the assessment of probability of open or in-filled defects (Section 8.6. • • It is not practical to develop any general rules for these assessments.6. The geological history of the valley. water losses and water pressure testing and grouting records.2 Width and extent of features associated with stress relief effects in the valley floor – valley bulge or rebound For stress relief effects associated with valley bulge or rebound.6.3). the assessment of the size of the defects is already considered in the assessment of probability of open or in-filled defects. voids. the information given in Section 8. except to say that for solution on stress relief defects. the extent and depth to which the features may occur should be assessed based on the available information from construction and investigations. Observational data – observations from boreholes including in-filled features.

assess the likelihood of grouting not being effective using Tables 8. grouting is assumed to be ineffective and assign a probability for grouting not being effective = 1. Table 8. mapping of foundations during construction shows a mix of in filled and open defects and solution features. Table 8. and the lower scale is for defects >100 mm.5 to 0. mapping of foundations during construction shows open defects or solution features.1 Probability In Filled 0.5 to 0. Localised large leakage in dam foundations. water pressure testing shows scour of in fill. Note that there are three probability scales on Table 8. coring or down-hole imaging shows in fill. Moderate leakage in dam foundations but no identifiable single or few sources. Large leakage in dam foundations. coring or down-hole imaging.99 to 0.18.1 0.95 to 1. the upper scale is for defects 5mm to 25 mm.9 0.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8. complete loss of drill water in boreholes.0.18.05 to 0 8.7 Likelihood of Defects or Solution Features being in filled Given there are defects below the level of the core it is necessary to assess the probability these are open or in filled. coring or down-hole imaging. 0.5 0.05 to 0.5 0. the middle scale for 25 mm to 100 mm. water pressure testing shows scour of infill. • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-26 .17 and 8.8 Likelihood of Grouting Not Being Effective in cutting off open or in filled defects or solution or other features Evaluate the likelihood of grouting not being effective in cutting off the potential open or in filled defect or solution features is as follows.0 0. mapping of foundations during construction shows a mix of in filled and open defects and solution features.95 to 0.9 0. For in filled defects. coring or down-hole imaging.16 should be used to assist in making this assessment.16 – Probability of Defects or Solution Features being open or in filled Scenarios Open Very small leakage in dam foundations. • For open defects.01 to 0. mapping of foundations during construction shows infill.

and the weighting of Orientation taken as (1) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-27 .closure. • Solution features are more likely to develop along the predominant defects sets.2a). Moderate to high leakage in the foundation No or very little reduction in pore pressures across grout curtain (Δhp/hp <10%). Low leakage in the foundation More Likely (3) Grout holes at an acute angle to dip of open defects (10 to 30 degrees) Single line curtain 5m to 6m (15 ft to 20 ft) spacing. tertiary etc holes to close to < 10 lugeons or <25 kg cement /meter (15lb/ft) take. > 10m (30ft) spacing. and there is a large drop in pore pressure across the grouting.2b).3 for definition of Δhp and hp. spacing. no check of closure. I (c) If there is very good instrumentation which can measure the effectiveness of the grouting to a high degree of confidence. single stage. • Defects associated with valley bulge are likely to be horizontal or near horizontal. High leakage in the foundation OR No performance data available at all Notes (a) The dip of the open defect will depend on the type of defect. these are likely to be parallel to the valley slopes (refer to Figure 8. W/C ratio < 3 Significant reduction in foundation pore pressures across the grout curtain (Δhp/hp >60%) Very low leakage (c) in the foundation Factor Neutral (2) Grout holes at an acute angle to dip of open defects (30 to 45 degrees) Single line grout curtain with primary holes 6m (20ft) or less spacing. high (>10:1 ) W/C ratio Orientation of grout holes compared with the (1) open defects Quality of grouting (. tertiary etc holes to close to < 10 lugeons or <25 kg cement /meter (15lb/ft) take. When making this assessment takes account of the amount and quality of the instrumentation. the weighting for Performance may be taken as (3). secondary. • For stress relief defects in the valley wall in inter-bedded sedimentary rock. (b) Refer to Figure 8. these defects are likely to be near vertical and parallel to the valley walls (refer to Figure 8. high (>5:1 ) W/C ratio Much More Likely (4) Grout holes parallel or near parallel to dip of open defects (<10 degrees) Single line curtain. W/C ratio <3 Moderate reduction in pore pressures across grout curtain (Δhp/hp = 30% to 60%). For stress relief defects associated with massive rocks. the duration and range of reservoir levels of the observations. number of lines. no secondary holes to check closure. with primary holes 6m (20ft) or less spacing. w/c ratio) (3) (c) (2) Performance (pore pressures (b) and leakage) (1) (c) Minor reduction in pore pressures across grout curtain (Δhp/hp = 10% to 30%).Factors influencing the likelihood of grouting not being effective for continuous open defects and solution features Relative Importance of Factor Influence on Likelihood Less Likely (1) Grout holes at a wide angle to the dip of the open defect (45 to 90 degrees) Three or more lines of grout curtain. secondary.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8.17 . grout takes.

05 0.05 0.005 0.20 to assess the probability the cut-off has not been successful and a continuous open defect or solution feature remains.9 Likelihood of Cut-off Walls Not Being Effective in cutting off open or in filled defects or solution or other features Where a cut-off has been excavated and backfilled in the rock foundation to intercept the continuous open defect or solution feature.001 0.9 0.2 0.6 0.01 0.99 RF x LF 6 9 11 13 18 24 8.Probability of grouting not being effective for continuous open defects or solution features versus (Relative importance factor (RF) x (Likelihood factor (LF)) ∑ 5-25mm 25-100mm >100mm 0. Table 8.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Δhp hp Rock Foundation Piezometer Grout curtain Figure 8. In assigning the weightings it is assumed that for such cut-offs there will be good quality monitoring of the pore pressure drop across the cut-off.5 0.02 0.1 0.3 – Definition of Δhp and hp.18 .8 0.3 0.5 0. use Tables 8.95 0.002 0.2 0.19 and 8. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-28 .

and/or Concrete. And Concrete. When making this assessment taking into account of the amount and quality of the instrumentation. the duration and range of reservoir levels of the observations. sand gravel backfill Excavated under bentonite. cement. sand gravel backfill Notes (1) Refer to Figure 8. and/or Bentonite cement or soil bentonite backfill Quality of the cutoff (1) Excavated open hole or under water.Factors influencing the likelihood of a cut-off in the foundation not being effective for continuous open defects and solution features Influence on Likelihood Factor Relative Importance of Factor (3) Less Likely (1) Width and depth of the defects and solution features well defined and cut-off extends sufficiently wide and deep Neutral (2) Width and depth of the defects and solution features not well defined cut-off may or may not extend sufficiently wide and deep Moderate reduction in pore pressures across cut-off wall (Δhp/hp = 30% to 60%). or bentonite.3 for definition of Δhp and hp. good clean-up of the base of the excavation. and overlap between panels or piers.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8. Moderate to high leakage in the foundation Much More Likely (4) Width and depth of the defects and solution features well defined and cut-off does not extend sufficiently wide and deep Width and depth of the cut-off relative to the defects and solution features Performance (pore pressures and leakage) (1) (2) Significant reduction in foundation pore pressures across the cut-off wall (Δhp/hp >60%) Very low leakage in the foundation No or very little reduction in pore pressures across cut-off wall (Δhp/hp <10%). A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-29 . sand gravel backfill Excavated under bentonite. Low leakage in the foundation More Likely (3) Width and depth of the defects and solution features not well defined and cutoff probably does not extend sufficiently wide and deep Minor reduction in pore pressures across cut-off wall (Δhp/hp = 10% to 30%). borehole camera inspection to confirm defects are intercepted. poor clean-up of the base of the excavation. and/or Bentonite cement or poorly controlled concrete or cement. High leakage in the foundation OR No performance data Excavated under bentonite. and overlap between panels or piers. or bentonite. moderate cleanup of the base of the excavation. poor overlap between panels or piers.19 . cement.

02 0.001 0.10 Probability that erosion of infill in the defects or solution feature initiates 8.20 .1 0.9 6 9 11 14 18 24 RF x LF 8.005 0. or hydraulic fracture through the infill. incomplete filling of the defect.Probability of a cut-off not being effective for continuous open defects or (Relative importance factor (RF) x (Likelihood factor (LF)) solution features versus ∑ 0. Gaps in the infill material may form by collapse settlement of infill material upon saturation.4 – Potential mechanisms for erosion of infill within a defect or solution feature. Figure 8. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-30 .5 0.1 Overview of Method This section considers the likelihood of erosion of the infill initiating due to seepage flow.10. INFILL FLOW ALONG A CRACK OR GAP CAUSED BY COLLAPSE ON SATURATION SCOUR DUE TO FLOW THROUGH AN ADJOINING OPEN JOINT SOLUTION FEATURE FLOW THROUGH A GAP ON THE SIDE OF THE INFILL Figure 8.4 shows these potential mechanisms.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8. An in-filled feature can be eroded by seepage flows through gaps within or along the sides of the infill material or due to seepage through adjoining open defects causing scour of the infill material.

0 For reservoir levels above the Pool of Record – Long term seepage is irregular or showing long term upward trend.2 Estimation of the probability of erosion of in fill initiating based on performance data Use Table 8. The weighted average of the two estimates is then made using weightings judged by the risk analysis team based on the quality of the input data.1 depending on the frequency and quality of seepage and piezometer data Use first principles approach (Section 5.10. Piezometric levels showing irregular behaviour or long term changes (up or down) No data on seepage or piezometric levels and seepage at toe is hidden (e. 8. For reservoir levels up to the Pool of Record – Probability for Erosion Initiates 0.21 Probability of erosion initiating based on performance data Scenario Long term seepage and piezometric levels show no irregularities such as sudden increases or reductions in pressure or flow. taking into account the opening width of the gap/crack in the infill. deep alluvium) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-31 .005 to 0. drowned.21 to estimate the probability of erosion initiating based on performance data. Table 8.g. Based on first principles.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 The assessment is based on two approaches: • • Based on performance. in terms of any evidence of changes in seepage and piezometric levels with time which would be indicative of erosion of in filled defects.0 1.4) 1. gradient and nature of the infill material.

then estimate the probability of erosion continuing using Section 10. If the exit is unfiltered then the probability of erosion continuing will be 1.35 to estimate the likelihood of erosion initiating. • • • 8.10. Estimate the average hydraulic gradient along the in filled defect. iii. Assess the soil classification of the infill material. saturating of the infill causing settlement and leaving a gap. If no information is available then assume the gap width wgap is 5 mm.11. tunnels or adits). If the exit is filtered.4. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-32 . tree roots.3 Estimation of the probability of erosion of infill initiating using first principles The method is as follows. ii. Use Tables 5.1.11 Probability that erosion of infill continues 8. Consider information from observations in excavations and other exposures (e. using the gap width wgap for the crack width values in the tables.0. This allows for the potential for on-going opening of stress relief defects due to dam construction activities. This is equal to the difference in elevation of the reservoir and the natural ground elevation at the downstream toe (or at the location of the exit point of seepage) divided by the seepage path length along the defect from the upstream toe to the downstream toe.1 Approach The probability of erosion of infill continuing is estimated by: i. • Estimate the width of the gap or crack adjacent to the infill material wgap.29 to 5.g. considering the defect infill as the base soil.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8. Assessing the probability that the exit point of seepage through the open or in-filled defect is filtered or unfiltered.

13 and Figures 10.5. The weighted estimate of the probability (P W ) should be estimated using: P W = w P TG + (1-w) P SC Where w = weighting factor to be assessed based on the quantity and quality of the available data. The second (P SC ) depends on having more detailed site investigations and construction records.Punf. Table 8.2 Probability of Filtered or Unfiltered Exit The likelihood that the exit will be filtered or unfiltered will depend on the embankment zoning and the details of the foundation geology and the geometry of the in filled defect or solution feature.22 provides some guidance in assessing the weighting factor. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-33 . The steps are as follows.0. or deteriorated toe drain systems. Calculate the probability that there will be a filtered exit (Pfe) = 1. The probability of continuing erosion = (Punf x 1..4. 8. there is still the possibility that seepage paths could bypass the filter drain system due to the geological conditions. Assess the filter materials and materials being eroded in terms of the Continuing Erosion criteria as described using the procedure in Section 10.12.12 Combining probabilities for a continuous open defect in rock and describing the defects and solution features 8.11. or there will be limited data.0) + (Pfe x PCE). • • Note that in situations where a foundation filter drain or toe drain system is present. • Estimate the probability that there will an unfiltered exit (Punf) using Table 10.7. 10.1. The first is based on topographic and geologic information (P TG ) which should be available for all dam sites.2 to 8.1 Calculating weighted averages of estimates In Sections 8. which for some dams will be not available.6 and 10. The conditional probability of continuation for this scenario is equal to (Punf) x 1.4 two estimates are made of the probability of the presence of open or in filled defects or solution features. compaction of soil foundations below foundation filters.13 to aid judgment. Examples of these scenarios are included in Table 10.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8.

1 to 0. basic grouting records. no records of foundation treatment some photographs of foundations Extensive good quality site investigations data and water pressure testing.g. small number of boreholes.12.9 0.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Table 8. small number of boreholes or larger number but poor quality boreholes.1 where weighting is applied.8 0. no mapping of the embankment cut-off foundation or grouting records.2 to 0. grouting records. some records of foundation treatment and photographs of foundations Very detailed and good quality site investigations data.2 8.0 0. limited mapping of the embankment cut-off foundation. and grouting records.g.12.22 . no mapping of the embankment cut-off foundation or grouting records. no foundation treatment information or photographs of foundations) Some site investigations data (e.4 to 0.6 to 0. no foundation treatment information or photographs of foundations) Some good quality site investigations data and water pressure testing. This is to be done for each defect or solution feature width using the weighted probabilities from Section 8. foundation treatment and photographs of foundations Weighting Factor (w) 1. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-34 .5. Compute the probability of a continuous open defect in rock foundation below the embankment using the event tree shown in Figure 8.Weighting factors for assessing probabilities of open or in filled defects and solution features Information Available No site investigations or construction mapping or other records Limited site investigations data (e. good quality mapping of the embankment cut-off foundation.2 Summing probabilities A continuous open defect in the rock foundation could already be present or could be formed by erosion of an in filled defect. reasonable quality mapping of the embankment cut-off foundation.4 0.6 0. no or poor quality water pressure testing data.7 to 0. Sparse or poor quality water pressure testing data.

0% Infilled 50.0% 50.0% 5-25mm 10. and sides of the cut-off trench is clear. and breach probabilities.0% Does erosion initiate? No 50.0% A butment stress relief features Continuous open or infilled feature present? 25-100mm >100mm 1. These are required so that the potential failure modes can be clearly visualised for the assessment of progression. detection.13.0% 50.0% repeat tree Figure 8.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 Yes Open 50. intervention and repair. The extension of the defects and solution features upstream and downstream of the core.0% 50. and beyond the embankment should also be shown.1 Describing the defects and solution features in relation to the embankment details It is essential that the risk analysis team document with sketch diagrams (plans and sections) of the spatial distribution of the open and in filled defects and solution features which have been assessed potentially present in the foundation. and cut-off walls (if constructed) should also be shown.0% Is feature open or infilled? Yes Yes 50. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-35 . This should show the features superimposed on the foundation drawings and showing the cut-off and general foundations beneath the core of the embankment so the relationship between the defects and solution features and the base of the cut-off trench. including under the shoulders of the embankment.0% Estimated width below cutoff trench X mm Open feature of X mm width Does erosion continue? No 50.0% Yes 50.0% Open feature of X mm width Is grouting ineffective? No 50. Any foundation grouting.5 – Computation of probability of a continuous open defect or solution feature below the embankment.13 Describing the defects and solution features and failure modes 8.0% Is grouting ineffective? No 50. surface treatment of the cut-off foundation. 8.

It should be expected that there will be more than one failure path.Probability of the Presence of Open or In Filled Defects in Rock Foundations SECTION 8 8. i) Gross enlargement of the defect. iii) Unravelling due to seepage exiting from the defect or solution feature into the downstream shell. For example it is likely the potential stress relief defects may occur in one or both abutments. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 8-36 . Valley bulge features are likely to be in the river section of the foundation.13. The potential failure paths arising from the defects and solution features should be assessed and sketches prepared to show them so the risk analysis team have a clear picture of the failure paths and their relation to the defects and solution features. unravelling or sinkhole development in the embankment. v) Initiation of internal erosion of the embankment at or into the foundation by backward erosion piping or scour followed by gross enlargement. iv) Sinkhole development.2 Describing the failure paths The potential failure paths due to internal erosion in rock foundations are. slope instability. ii) Slope instability due to increased pore pressures caused by seepage from the defect or solution feature up into the downstream shell. but not to the river section of the foundation.

1 General Principles For erosion to initiate from the embankment into the foundation. Refer to Section 9.6 for guidance on estimating the probabilities.7 for guidance on estimating the probabilities. (b) Assess the probability of the initiation of internal erosion by backward erosion or suffusion (Pe) starting at the core-foundation contact given there is a continuous path. Erosion in a crack or hydraulic fracture across the cut off trench. requires open joints in rock or coarse soils in the base or sides of the cut off trench.2 Overall Approach The method to assess the probability of initiation of erosion into these features is: (a) Assess from the available data the probability of a continuous pathway of open joints in rock or coarse grained soils (Ppath) in the base or sides of the core trench or core-foundation contact. Refer to Section 9. Internal erosion may initiate by: • • • Backward erosion or suffusion in a high permeability zone in the core or cut-off trench.3 for guidance on estimating the probability.5 for guidance on estimating the probabilities. (c) Assess the probability of initiation of scour. then the issue is whether erosion can occur along the core-foundation contact. Refer to Section 9. In many cases it will become apparent at this stage that there is little or no likelihood of such features being present and the probability of piping into the foundation may be assessed as negligible. Refer to Section 9. (d) Assess the probability of erosion of the core following hydraulic fracture due to arching in a narrow core trench.Probability of Initiation of Erosion from Embankment into Foundation 9 Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 9. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-1 . For cases where a cut off trench is not present. 9. Scour of the core at the core – foundation contact by water flowing in joints in the rock foundation.

use Section 8 to assess the probability of such defects and features. (PCR) 25−100 etc. • For defects such as joints <5mm wide. This assessment will also determine the spatial distribution of such defects and features. each abutment and the river section.3 Probability of a continuous pathway for erosion of the core of the embankment into a rock foundation (Ppath) IM28 Probability of a continuous pathway of open defects and solution features in the rock foundation (PCR) 9. It may be demonstrated that the maximum width defect left in the cut-off foundation was. the depth of general foundation and cut off excavation. and for the various parts of the foundation. <2mm. mapping and. (PCR) 5−25 .g. slush concrete.3. 5mm to 25mm. Do this for each defect width.g. and written record to demonstrate with high confidence they are not present. e. and into the sides of the cut-off trench.1 to assess the probability the foundation treatment fails to prevent contact of the core with open defects or solution features.2 Likelihood treatment of the embankment cut-off foundation does not prevent contact of the core with open defects or solution or other features (PTI) Use Table 9. while wider features may occur in the sides of the cut-off trench. for the base and the sides of the cut-off trench. e. grouting records. whether they persist to below the base of the cut-off trench beneath the core.0) unless there is very detailed mapping. • 9. 25mm to 100mm. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-2 . it should be assumed that they are present (PCR = 1. dental concrete).g.3.1 This should be assessed from the geology of the foundation. for example. photography. In many cases it will become apparent at this stage that there is little or no likelihood of such features being present and the probability of internal erosion into the foundation may be assessed as negligible. shotcrete. photographs taken during construction.Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 9. The probability of continuous pathways should be assessed by: • For defects and solution features wider than 5mm. This will result in a set of probabilities for the different width features. treatment carried out on the walls and floor of the cut off trench (e.

05 to 0.Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 Table 9. and all open and in-filled defects and solution features were covered with at least 100mm (4 inches) of concrete or good quality shotcrete.9 0. estimate the probability of a continuous pathway for erosion of the core into open defects and solution features from: Ppath = (PCR)x (PTI) Do this for each width of defect and feature as the likelihood of erosion initiating is related to the defect width.3 to 0.2. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-3 .3 Probability of a continuous pathway for erosion of the core into a rock foundation (Ppath) Using the results from Sections 9. the design and construction organisation not known or known but likely to have not paid much attention to inspecting foundations and carrying out surface treatment No construction records available. Probability of the treatment not preventing contact 1. but knowledge that the practice of the design and construction authority was to inspect foundations and carry out surface treatment Evidence that the foundations were mapped.0 0. but not in detail.3.001 9.3.Probability of treatment of the cutoff foundation not preventing contact of the core with open defects or solution features Scenarios Well document evidence that there was no treatment of the cut-off foundation No construction records available. Some evidence that open and in-filled defects and solution features were covered with concrete or shotcrete.1 . Well documented evidence that the foundations were carefully mapped.3.2 0.1 to 0.01 to 0.5 0. or that they were cleaned out to at least 3 times the surface width and treated with slush grout.1 and 9.

Scenarios Adequate treatment of soil foundation contact Site investigations indicate continuous coarse grained foundation soil layers are very unlikely to be present Site investigation data is not available.g. the depth of general foundation and cut off excavation.01 Depending on level of confidence in assessment and degree of continuity of features Site investigation or construction data indicates coarse grained foundation soils are likely to be in contact with the core. photographs taken during construction. In some cases negligible. lateritic profiles). no or inadequate treatment 0. construction photographs 0. but circumstantial evidence indicates coarse grained foundation soils maybe present.001 Depending on the quality of the available data. Geological environments where coarse grained soils are unlikely to be present (e. glacial. mapping and.001 – 0. lacustrine. Table 9. Examples Filter protection provided on downstream side of cut-off trench No evidence of openwork gravel layers Site investigation data is not available. aeolian. Observations of cuts in foundation soil. alluvium.Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 9. but circumstantial evidence indicates coarse grained foundation soils are unlikely to be present. volcanic ash).4 Probability of a continuous pathway of coarse grained layers in soil foundations (IM29) This should be assessed from the geology of the foundation.2. and whether there are filters between the core and trench side. Geological environments where coarse grained soils maybe present (e.5 Depending on extent and degree of continuity of soils A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-4 .0001 to 0. 0. Evidence from drill holes. In many cases it will become apparent at this stage that there is little or no likelihood of such features being present and the probability of internal erosion into the foundation may be assessed as negligible.05 – 0. Guidance on estimating the probability is given for a range of scenarios for soil foundations in Table 9. Circumstantial evidence might include. Observations of cuts in foundation soil. residual soils.g. colluvium. excavation logs.2 – Probability of a continuous pathway for erosion into soil foundation (IM29) Range of Probabilities for Continuous Pathway of Coarse Grained Soils Negligible Assume probability = 0 Negligible Assume probability = 0 Circumstantial evidence might include.

6. and breach. The steps in the assessment are as follows. progression. Otherwise the gradients should be estimated using flow nets modeling the embankment and foundation.2. For cases where there are defects of varying widths the head loss will occur mainly in the narrow defects as discussed in Section 9. The probability of internal erosion by backward erosion or suffusion = Ppath x Pe. • Estimate the probability of a continuous pathway for erosion into the foundation (Ppath) from Sections 9. • • • If the embankment is well instrumented. This considers the likelihood of initiation of internal erosion by backward erosion or suffusion due to seepage through the core exiting into an unprotected continuous pathway in a rock or soil foundation. intervention and repair.5 Probability of the initiation of internal erosion by backward erosion or suffusion starting at the core-foundation contact. Do this for each of the potential defect openings and soil particle sizes and take these estimates forward into the assessment of continuation. or from simplified calculations of the likely gradient across the core. detection. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-5 . the seepage gradients will be the gradients determined from piezometers upstream and downstream of the cutoff trench and/or from the flow net derived from piezometers in the embankment and foundation.4. Estimate the probability of backward erosion or suffusion initiating at the core-foundation contact (Pe) using the methods described in Section 6. This should be accounted for when estimating the seepage gradients through the core.Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 9.3 and 9.6.

6.6 9. intervention and repair. and breach.Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 9.2 How to model scour into defects of varying width and persistence Section 8 is modeled around assessing the probability of defects being present in the foundation which are continuous from upstream to downstream of the core. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-6 . Estimate the probability of erosion of the core material at the core-foundation contact (Pic) using the method for erosion in a crack in the core as described in Section 5.2. progression. Figure 9. Guidance is given in Section 9.6. Assume that the hydraulic shear stresses imposed on the core by the water flowing in the open joints is equivalent to those for an equivalent crack width. The probability of internal erosion by scour = Ppath x Pic. The hydraulic gradient used in the assessment should be based on the estimated seepage gradient on the core-foundation contact.1 shows some examples of this.1 Probability of initiation of scour at the core-foundation contact The Steps to be followed This considers the likelihood that seepage flows within a continuous pathway in a rock or soil foundation may initiate erosion of the core material at the core-foundation contact.6. Do this for each of the potential defect openings and soil particle sizes and take these estimates forward into the assessment of continuation. • Estimate the probability of a continuous pathway for erosion at the core-foundation contact (Ppath) from Sections 9. The steps in the assessment are as follows. • • • 9. and are a constant width.4. There may be foundations where the information available from construction and site investigations indicates that there will be continuous open defects in the foundation but they are of varying width.3 and 9.4. detection.

1 – Examples of foundations with continuous open defects of varying width A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-7 .Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 Figure 9.

Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 The following equations can be used to estimate the hydraulic shear stress on the surface of a cylindrical pipe. Cylindrical pipe: τ = ρw where gH f φ 4L τ = Hydraulic shear stress in N/m2 Density of water in kg/m3 Acceleration due to gravity = 9.8m/s2 Head loss in pipe due to friction in meters Length of pipe in meters ρw = g = Hf = L (b) = φ = Diameter of the pipe in meters Vertical transverse crack τ= ρ w gH f 2W 2( H f + W ) L τ = Hydraulic shear stress in N/m2 Density of water in kg/m3 Acceleration due to gravity = 9.8m/s2 Head loss in crack due to friction in meters Length of crack base in meters Width of crack in meters. or parallel sided transverse crack. where ρw = g = Hf = L W = = A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-8 . The assumptions are: • • • • • (a) Linear head loss from upstream to downstream Steady uniform flow along the crack Zero pressure head at the downstream end Uniform frictional resistance along the surface of the crack or cylindrical pipe Driving force = frictional resistance.

e. is the diameter of the pipe. so the gradient is most affected by the defect width or diameter.1 erosion will initiate in the narrow defects where gradients are high in preference to the wider sections of defect. Darcy-Weisbach equation) q =V πφ 2 4 Re = ρ wV φ μ where f is the friction loss factor for pipe flow.Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 Hence for narrow cracks and cylindrical pipes the hydraulic shear stress τ ∝ ( H f /L. Hence in the situations shown in Figure 9. and W or d) The gradient in a cylindrical pipe is given by Hf 4f V 2 = L φ 2g (i. is the coefficient of dynamic viscosity of water (10-3 kg/ms at 20oC). is the mean velocity of flow along the pipe. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-9 . is the rate of discharge. φ V Q Re μ From these equations it can be seen that the hydraulic gradient ∝ (1/ φ 3 ) whereas the imposed hydraulic shear stress ∝ φ . is the Reynold’s number. Within the accuracy of the calculations. it is sufficient to assume that all the head loss occurs in the narrow defects.

7 Probability of erosion of the core following hydraulic fracture due to arching in a narrow cut-off trench The steps in the assessment are as follows. and breach. • Estimate the probability of a continuous pathway for erosion at the core-foundation contact (Ppath) from Sections 9.3 and 9.Definition of terms for arching across a cut-off trench.Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 9. Do this for each of the potential defect openings and soil particle sizes and take these estimates forward into the assessment of continuation. The hydraulic gradient used in the assessment should be based on the estimated seepage gradient across the cut-off trench. Estimate the probability of a hydraulic fracture occurring across the cut-off trench (Phf) using Table 9. • • • • W β D Figure 9.2 .3 and Table 9. detection. Probability of internal erosion by hydraulic fracture across the cutoff trench = Ppath x Phf x Pic. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-10 . Estimate the probability of erosion through the hydraulic fracture across the cutoff trench (Pic) using the method for erosion in a crack in the core as described in Section 5. Assume that the hydraulic fracture is 5 mm wide. The hydraulic fracture in the cut-off trench needs to coincide with an open joint or coarse grained soil layer to find an unfiltered exit and hence the assessed probability should consider the likelihood of this coincidence.4. intervention and repair.4.4. progression.

00005 negligible negligible 0.4< W/D<0.3 .1 Above POR 6 9 10 13 17 21 24 RF x LF Note: “POR” refers to the Pool of Record level + 1 foot A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 9-11 . D.75 0.0005 0.005 0.Factors influencing the likelihood of hydraulic fracturing within the cutoff trench due to arching Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Moderate slope Neutral (2) Moderate steep slope 45 < β < 60 o o More Likely (3) Steep slope 60 ≤ β < 75 o o Much More Likely (4) Very steep slope Slope of side slopes of cutoff trench (a) β ≤ 45 o β ≥ 75o Note.005 0.2 for definitions of W.Probability of Initiation of Erosion from Embankment into Foundation SECTION 9 Table 9. and β.4 Depth of cutoff trench Note: Very narrow deep trench W/D < 0.25 > 20 ft (6 m) (1) < 6 ft (2m) deep 6 ft to 13 ft (2 m to 4 m) 13 ft to 20 ft (4 m to 6 m) (a) See Figure 9. If this condition is present.001 0.0001 0.75 Narrow deep trench 0.0005 0.Probability of hydraulic fracturing in cut off trench due to arching versus (Relative importance factor (RF)) x (Likelihood factor(LF)) ∑ negligible negligible 0. Table 9.05 0. probability is negligible Geometry of cutoff trench(1) (2) Wide trench W/D > 0.01 Below POR 0.4 .25< W/D< 0.

3).1. estimate the probability for Continuing Erosion based on the guidance given in the applicable section (Sections 10. open joint or crack in a conduit or adjoining concrete structure). For Scenario 3. Scenario 5: Erosion into a toe drain. containing > 5% plastic fines. Soils which are capable of holding a crack or pipe are: – – – – • well compacted shoulder (shell). >30% non plastic Scenario 3: Filter/transition zone is present downstream of the core or a downstream shoulder zone which is not capable of holding a crack/pipe. containing >15% plastic fines well compacted shoulder. Scenario 2: Downstream shoulder of fine grained cohesive material which is capable of holding a crack/pipe.1 Internal Erosion Through the Embankment – Overall Approach Step 1: Assess which of the following five scenarios is most applicable to the dam section and failure path that is under consideration: • • Scenario 1: Homogeneous zoning with no fully intercepting filter.2 or 10. or poorly compacted shoulder (shell). Step 2: For Scenarios 1 and 2.1. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-1 .g. The increase in leakage flows is so small that it is unlikely to be detectable.Probability of Continuation SECTION 10 10 Probability of Continuation 10. The No Erosion branch on the event tree is a “No Breach” branch. Scenario 4: Erosion into a crack or open joint (e. This includes earthfill dams with a chimney filter. • • The example sketches shown in the navigation table (Table A7 in Appendix A) can be used to help evaluate the most applicable scenario. a four way split for filtering behaviour is recommended in the event trees: • Seals with No Erosion – the filtering material stops erosion with no or very little erosion of the material it is protecting.1 Probability of Continuation for Internal Erosion in the Embankment 10.1. containing > 30% non plastic fines poorly compacted shoulder.

Excessive Erosion and Continuing Erosion based on the guidance given in Section 10. but the flows are self healing. Adopt a probability of Continuing Erosion = 1. Leakage flows due to piping can be up to 3 cfs (100 l/s). or if not.0.5. but it eventually seals up and stops erosion.1. estimate the probability for Some Erosion.3 Probability for Continuation – Scenario 2 (Downstream shoulder can hold a crack or pipe) The issue for this scenario is whether the crack/high permeability feature that is present through the core is continuous through the downstream shoulder. Seals with Excessive Erosion – the filter material allows erosion from the material it is protecting.4. estimate the probability for Some Erosion. 10.1.2 Probability for Continuation – Scenario 1 (Homogeneous zoning) There is no potential for filtering action for this scenario. Continuing Erosion – the filtering material is too coarse to stop erosion of the material it is protecting and continuing erosion is permitted. For Scenario 4. but are self healing. Unlimited erosion and leakage flows are likely. Use Table 10. 10. whether it can find an exit. The material characteristics and width of the downstream shoulder zone. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-2 .1.1 to evaluate the conditional probability of continuing erosion. This depends on the following factors: • • The mechanism causing the concentrated leak. • • For Scenario 3. Excessive Erosion and Continuing Erosion based on the guidance given in Section 10. and in the process permits large increases in leakage flow (up to 35 cfs). in particular whether it also causes cracking in the shoulder. The extent of erosion is sufficient to cause sinkholes on the crest and erosion tunnels through the core.Probability of Continuation SECTION 10 • Seals with Some Erosion – the filtering materials initially allow erosion from the soil it is protecting. The maximum leakage flows for the various filter erosion categories described above are used to assess the potential for unravelling and instability in the Breach Mechanism phase.1.

shutdown surface) Leak unlikely to find an exit through the shoulder (i. materials placed in upstream/downstream orientation.5 Along outside of conduits passing through the dam Leak also likely to be common cause through downstream shoulder (e. arching in trench backfill) 0. Material may collapse on wetting.0 0.0 0.1.. very wide downstream shoulder.g. Range of Conditional Probabilities for Continuing Erosion 1.0 High permeability zone in the core or along the foundation contact. poor compaction. high gradient across shoulder.e. features causing cracking in the core are not present below the downstream shell.5 . feature extends part way through the shoulder) 0.1 0.g.1 Conditional Probability Ranges for Continuation (Scenario 2) Predominant Mode of Concentrated Leak Cracking due to differential settlement (cross valley.5 0. foundation. desiccation cracking on sides of excavation. cohesive materials.Probability of Continuation SECTION 10 Table 10. well compacted.1 – 0. 0. less prone to desiccation cracking High permeability feature also likely to be present across the shoulder zone (e.01 – 0. similar compaction methods and lift thicknesses used in core and downstream shoulder. high erodibility. low erodibility.5 . different compaction methods and lift thicknesses used in core and downstream shoulder) Leak likely to find an exit through the shoulder (e. or Cracking due to differential settlement.g.5 – 0.5 . embankment staging). low plasticity materials.0 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-3 . Mechanism causing cracking in the core is also likely to cause cracking of the downstream shell (e. low gradients. common cause cracking). Desiccation cracking near crest. or on construction layer Characteristics of downstream shoulder zone Well compacted. Material likely to hold a crack.g. Poorly compacted.1 – 0.9 Similar plasticity to core Lower plasticity than core. narrow downstream shoulder.1.1.

Figure 10. If the probability of the filter/transition holding a crack from Table 10. • (a) Particle Size Distribution Information is Available The recommended procedure is shown as a flowchart in the navigation tables (Table A7 in Appendix A) and involves the following steps: Step 1: Adjust and select base soil gradings.1 and 0.Probability of Continuation SECTION 10 10. then regrade the base soil grading on the particle size that is missing (i. The two approaches are as follows: • If particle size distribution information is available for the core and filter/transition/shoulder materials (either from construction.1 shows an example.4 Probability for Continuation – Scenario 3 (Filter/transition zone is present downstream of the core or a downstream shoulder zone which is not capable of holding a crack/pipe) The method of assessing the probability for continuation depends on the information that is available on the particle size distributions of the core and filter/transition/shoulder materials.4 (b). If the factor of safety is greater than about 0. If the factor of safety is less than about 0. at the point of inflection of the grading curve).4 (a). the average grading (average base soil grading) and the coarser 5% of the base soils (coarse base soil grading). Plot the particle size distributions for the core material and the filters or transitions which are protecting the core.5. specifications and/or borrow area investigations).0. If the factor of safety for blow out is greater than 0.9 should be applied. If the maximum particle size of the core material is >4. If particle size distribution information is not available for the core and filter/transition/shoulder materials. Select representative gradings of the regraded base soil which are indicative of the finer 5% of the base soil gradings (fine base soil grading).1 it should be assumed the filter/transition will not be effective and probability of continuation PCE = 1. Step 2: Check for the blow out condition. assess the potential for blow out by comparing the seepage head at the downstream face of the core to the weight of soil cover.1.2 is ≥ 0. This is calculated as the ratio of the total stress from the vertical depth of soil (and rockfill) over the crack exit to the potential reservoir head.1.75 mm. assess the potential for them to hold an open crack using Table 10. Between these limits a probability of continuation between 0.1. Step 3: Check if the filter/transition zone will hold an open crack. then follow the proceeding steps. then use the approach described in 10.e. If the filter/transition zone contains an excess of silty or clayey fines.1. In cases where there is limited depth of cover over the filter/transition zone.2. then evaluate the remaining steps by A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-4 . then regrade the core grading such that the maximum size is 4. then use the approach described in 10.75 mm.5 three dimensional effects will be sufficient to make this a non-issue. If the base soil is gap graded.

then adjust the grading curve assuming that 50% of the unstable soil fraction is washed out. the average base soil grading and the coarse base soil grading. Figure 10. Use the DF15 values from the adjusted grading curves for estimating the conditional probabilities of No Erosion.3. then do not adjust the filter grading curves. Estimate the proportion of the filter/transition gradings that fall into each of the particular erosion categories based on the plot of filter/transition grading curves versus Filter Erosion Boundaries (an example is shown in Figure 10. then estimate the grading of the segregated layer assuming that 50% of the finer soil fraction is segregated out leaving the remaining 50% of coarser fraction. PEE fine. This assumes the cracked filter zone will also erode. Excessive and Continuing Erosion boundaries using Table 10. Step 5: Check if the filter/transition zone is internally unstable. Step 6: Evaluate the DF15 values for the No. Table 10. Step 4: Check if the filter/transition zone is segregated.2 shows a graphical method for adjusting the gradation curve to allow for suffusion. Use the adjusted grading curves for the filter/transition zone if required to do so by the preceding steps 4 or 5. average and coarse gradings) as follows.5. Step 7: Estimate the probabilities for No Erosion. and PCE fine. Excessive and Continuing Erosion categories first (PSE.3. PEE and PCE ) and then calculate PNE = 1. • For the fine base soil grading. Do this for each of the representative base soil gradings (fine.4).7 for the fine base soil grading. PSE fine. If there are no filter/transition gradings that fall into the Continuing Erosion category. The suggested approach is to estimate the proportions for the Some. Excessive Erosion and Continuing Erosion in Step 7. Excessive and Continuing Erosion for each representative base soil grading. then the ‘cracked’ filter zone should be ignored and the core evaluated against the zone downstream of the cracked filter. If the probability of internal instability is < 0. Some Erosion. PNE fine. Some Erosion. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-5 . Excessive Erosion and Continuing Erosion in the remaining steps.5 for materials with >10% fines or Figure 6.4 for an example).4 and Table 10. then use Table 10. If the filter is cemented. If the probability of internal instability (PIUS) is ≥ 0.8 to aid judgement in assigning probabilities for Continuing Erosion. This allows for the possibility of the gradations being coarser than indicated by the available information and depends on how much finer the gradings are to the Continuing Erosion boundary.3. Figure 10.(PSE + PEE + PCE).Probability of Continuation SECTION 10 considering the ‘cracked’ filter/transition zone as the base soil and the zone downstream of the cracked filter as the filter material.6 for materials with <10% fines.2 shows a graphical method for adjusting the gradation curve to allow for segregation. Use the DF15 values from the adjusted filter grading curves for assessing the probability of No Erosion. If a continuous segregated layer is likely to be present. Plot the DF15 values for these boundaries on the grading curve limits of the filter/transition material (see Figure 10. Some Erosion. Assess the potential for segregation of the filter/transition/shoulder materials using Table 10.6 and Table 10. Evaluate the probability that the filter or transition zone materials are internally unstable (PIUS) using Figure 6.

PEE ave. PNE ave. PSE coarse. The calculations are as follows. and PCE ave. Excessive and Continuing Erosion boundaries for the estimated gradation of the core materials using Table 10. Some Erosion.9 for the example shown in Figure 10. PSE = (5% x PSE fine) + (90% x PSE ave) + (5% x PSE coarse). The suggested approach is to estimate the probabilities for Some. unprocessed quarry fines or tunnel spoil. • • • • PNE = (5% x PNE fine) + (90% x PNE ave) + (5% x PNE coarse). and PCE coarse.Probability of Continuation SECTION 10 • • For the average base soil grading. SE.4. etc). Excessive and Continuing Erosion (PSE.7. Excessive Erosion and Continuing Erosion branches by the sum-product of the % of base soil gradings and the % of NE. decomposed granitic soils. processed sand and gravels. Use judgement to adjust the calculated percentages to take into account the effects of other factors such as the distribution of the core and filter gradations in the fill. Excessive and Continuing Erosion based on the estimated proportion of the filter/transition gradings that is likely to fall into each of the particular filter erosion categories. PNE coarse.6 and Table 10.g. Estimate the probabilities for No Erosion. The gradation of the soils may be able to be estimated based on the likely geological origin of the materials (e.(PSE + PEE + PCE). PEE coarse. PSE ave. Evaluate the DF15 values for the No. borrow area variability and selective placement of materials. For the coarse base soil grading. (b) Simplified Approach – Particle Size Distribution Information is Not Available • Estimate the particle size distribution of the core materials based on the likely source of materials. Estimate the particle size distribution of the filter/transition/shoulder materials based on the likely source of the materials and whether they were processed or not (e. Some Erosion. PCE = (5% x PCE fine) + (90% x PCE ave) + (5% x PCE coarse). alluvial fine clays and silts. PEE and PCE) and calculate PNE = 1. etc). run-of-pit alluvial sands and gravels. residual soils. PEE = (5% x PEE fine) + (90% x PEE ave) + (5% x PEE coarse). • • • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-6 . Make an initial estimate of the probabilities of the No Erosion. EE and CE for each representative base soil grading.g. An example of summing of the probabilities is shown in Table 10. Estimate the DF15 of the filter/transition/downstream shoulder materials.

Filters falling into the Continuing Erosion category – flows of 35 cfs (1000 l/sec) and increasing.Probability of Continuation SECTION 10 (c) Estimate Potential Leakage Flows Assess the potential leakage flows that could develop if piping were to initiate if piping were to initiate in the core based on the following guidance. • Information from case histories of poor filter performance suggest the potential maximum leakage flows that could develop due to piping are as follows: – – – Filters falling into the Some Erosion category – up to 3 cfs (100 l/sec): Filters falling into the Excessive Erosion category – 3 to 35 cfs (100 to 1000 l/sec). but the factors would need to be carefully considered and justified if they were to be relied on in the assessment. average and coarse) for the assessment of filter compatibility. It is recommended that the maximum leakage flows listed above be used in the assessment of the probability of a breach mechanism developing. Representative Fine Grading Curve Representative Coarse Grading Curve Average Grading % Passing Envelope of Base Soil Gradings Particle Size (mm) Figure 10. Lower leakage flows are likely if upstream flow limitation occurs.1 – Example of the selection of representative grading curves (fine. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-7 .

1 0.Likelihood for Filters with Excessive Fines Holding a Crack Fines Content Fines Plasticity % Passing 0.1 0.001 0.02 0.02 0.05 0.5 0.05 0.001 0.2 .2 – Approximate method for estimating DF15 after washout of the erodible fraction from a suffusive soil or for soils susceptible to segregation. filters derived from crushed limestone).0002 0.g. Table 10.1 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-8 .3 0.005 0.7 Probability of holding a crack Note: Fines susceptible to cementing for filters having a matrix predominately of sand sized particles (e.05 0.5 0.075 mm 5% 7% Non plastic (and no cementing present) 12% 15% >30% Plastic (or fines susceptible to cementing) 5% 7% 12% ≥ 15% Compacted 0.9 Not compacted 0.01 0.Probability of Continuation SECTION 10 % Passing Original gradation (1) Select the point of maximum inflexion of the grading curve (2) Locate the mid point below the point of inflexion Equal distance Estimated gradation curve after washout or segregation 15% (3) Estimate the shape of the gradation curve passing through the mid point Particle size (4) Estimate the D15 after washout or segregation Figure 10.

6 m) Fails segregation criteria in Table 10. > 20 foot wide (>6 m) End dumping from trucks. spread by dozer in thick lifts > 2 feet (> 0.5 m) Construction practices Placed in thin lifts < 2 feet (< 0.3 . >40% passing 4.4.4 25 – 40% passing 4.4. <5 foot (<1.75 mm sieve Narrow zone.75 mm sieve 10 – 20 foot wide zone (3 – 6 m) End dumping from trucks. spread by dozer in thin lifts < 2 feet (< 0.Potential for Segregation of Filtering Materials Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Good construction and stockpiling practices used Neutral (2) Fair construction and stockpiling practices More Likely (3) Poor construction or stockpiling practices Much More Likely (4) Very poor construction and stockpiling practices with no regard for segregation effects Filters/transition s constructed by pushing material over the edge of the core Significant departure from segregation criteria in Table 10.5 – 3 m) Width of zone A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-9 .75 mm sieve (1) Wide zone.6 m) Borderline segregation criteria in Table 10. and % sand (2) Meets segregation criteria in Table 10.75 mm sieve Narrow zone. >50% passing 4.Probability of Continuation SECTION 10 Table 10. <25% passing 4. Gradation – Comparison to USBR/US Corps filter criteria. careful control during construction. 5 – 10 foot wide (1.4.6 m).

0 – 5. Segregation should be assumed to be present.0 – 10 10 – 50 Maximum D90 (mm) 20 25 30 40 50 60 Table 10. US Corps of Engineers 1994) Minimum D10 (mm) <0.5 – 1.5 – Susceptibility of filter/transition zones to segregation versus weighted score (Relative importance factor (RF)) x (Likelihood factor(LF)) Weighted Score from Table 10. 18 – 24 High potential for continuous segregated layers A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-10 .0 1.0 5.0 2. unless investigations show otherwise.0 – 2.3 6 – 10 11 – 17 Segregation Assessment Low potential for segregation Moderate potential for segregation Consideration of Segregation Effects for Filter/Transition Assessment Segregation of filter/transition materials do not need to be considered Segregation of filter/transition materials should be considered.5 0.Probability of Continuation SECTION 10 Table 10. unless investigations show otherwise.4 – Gradation Limits to Prevent Segregation (USDA SCS 1994. USBR 1987.

Base Soil Soils with DB95<0.25g/cm2 in the CEF test (0.5 mm. DF15 ≤ 6.Probability of Continuation SECTION 10 Table 10. 2001).3 mm Soils with 0.4 DB85. Table 10.7)/25 + 0.3<DB95<2 mm Soils with DB95>2 mm and fines content >35% Soils with DB95>2 mm and fines content <15% Soils with DB95>2 mm and fines content 15-35% Proposed Criteria for Excessive Erosion Boundary DF15 > 9 DB95 DF15 > 9 DB90 DF15 > the DF15 value which gives an erosion loss of 0.25g/cm2 contour line in Figure 10.4 . suggest use DF15 ≤ 0.2. For soil group 2 soils.10 DB85 Criteria for No Erosion Boundary DF15 ≤ 9 DB85 (2) DF15 ≤ 0. Base Soil Category 1 2 3 Fines content (1) Design Criteria of Sherard and Dunnigan (1989) DF15 ≤ 9 DB85 DF15 ≤ 0.e.7 mm 1.5 DF15 design.7)/25 + 0.5 DB85 0.1. suggest use the lower limit of the experimental boundary.5 DF15 of Sherard and Dunnigan design criteria 6.75 mm.75 mm.075 mm after the base soil is adjusted to a maximum particle size of 4.075 mm) x (4DB85-0.13. The equation for soil group 4 would be modified accordingly.85% 15 .7 mm DF15 ≤ (40-pp% 0.40% 4 Notes: < 15% (1) The fines content is the % finer than 0.7 DF15 ≤ 4 DB85 ≥ 85% 40 . i. Foster and Fell 1999. Proposed Criteria for Continuing Erosion Boundary For all soils: DF15 > 9DB95 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-11 .7 Notes: Criteria are directly applicable to soils with DB95 up to 4. it is recommended to use a lower DF15 for the no erosion boundary. For soil group 1 soils. For soils with coarser particles determine DB85 and DB95 using grading curves adjusted to give a maximum size of 4.7 .8 .7 mm (2) DF15 ≤ (40-pp% 0.6 .6 – No erosion boundary for the assessment of filters of existing dams (after Foster and Fell 2001).75 mm. (2) For highly dispersive soils (Pinhole classification D1 or D2 or Emerson Class 1 or 2).7)/20+0.7 – Excessive and Continuing erosion criteria (Foster 1999. where DF15 design is given by: DF15 design=(35-pp%0.3) DF15 > 9 DB85 DF15 > 2.075 mm)(4DB850.7 DF15 ≤ 4 DB85 Range of DF15 for No Erosion Boundary From Tests 6.075 mm) x (4DB85-0.

Probability of Continuation SECTION 10 15 Filter DF15 (mm) 10 0. Table 10.1x DF15CE DF15 in dam < 0.medium sand (0.05 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-12 .0001 0.7.8 – Aid to judgement for estimation of probability for Continuing Erosion (PCE) when the actual filter grading is finer than the Continuing Erosion Boundary Comparison of Actual DF15 of the Filter/transition Zone to the Continuing Erosion boundary DF15 in dam < 0.5 x DF15CE Notes: DF15CE = DF15 for Continuing Erosion Boundary from Table 10.075 .001 0.7mm EXCESSIVE EROSION SOME EROSION 0 0 5 10 15 20 25 30 35 40 45 50 Core material % fine .01 – 0.2 x DF15CE DF15 in dam < 0. Probability for Continuing Erosion (PCE) 0.18mm) 55 60 Figure 10.3 – Criteria for Excessive Erosion Boundary.25g/cm Contour of Erosion Loss 2 5 No Erosion Boundary for Soil Group 2 DF15=0.1.

Average Grading Zone 1 . Evaluate the filter erosion boundaries for the representative fine.5 2.Example of plot showing filter/transition gradings compared to Filter Erosion Boundaries. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-13 .01 0.5 Excessive Erosion 30 36 38 Continuing Erosion 0.0 Erosion 0.2 35 30 0.9 3. average and coarse gradings of the core material.18mm) DF15 (mm) Fine Grading 1.6 Continuing Erosion DF15 (mm) 30 36 38 Figure 10.Fine Grading Zone 1 .75 mm Zone 3.Core Material re-graded to maximum size of 4.7 Average 2.Coarse Grading Filter erosion boundaries for the Average core grading Filter erosion boundaries for the Fine core grading Filter erosion boundaries for the Coarse core grading Percent Passing 60 50 40 30 20 10 0 Zone 1 .7 Some 2.4 .Filter Transition 0.4 4 41 29 0.001 0.5 4.1 1 Seive Size (mm) 10 100 1000 Assessment of Zone 1 core against no erosion.7 No Erosion 2.Probability of Continuation SECTION 10 100 90 80 70 Zone 1.7 Coarse Grading 2.075mm (0.1. excessive erosion and continuing erosion criteria Core Base soil sizes (mm) No Erosion Gradation % passing % fine-medium sand DB85 (mm) DB95 (mm) 0.075 .3 50 25 0.7 Excessive Erosion DF15 (mm) 2 2.

and hence a very low probability is assigned based on Table 10. In this example. Excessive and Continuing Erosion (Pxx) Calculation Result Assigned Probabilities PNE fine = 20% Some Erosion (SE) PSE fine = 60% Excessive Erosion (EE) PEE fine = 20% Continuing Erosion (CE) PCE fine = 0% 100% Sum PXX Representative Base Soil Grading PNE ave = 20% PSE ave = 70% PEE ave = 10% PCE ave = 0% 100% PNE coarse = 20% PSE coarse = 70% PEE coarse = 10% PCE coarse = 0% 100% PNE = (5% x PNE fine) + (90% x PNE ave) + (5% x PNE coarse) (5% x PSE = PSE fine) + (90% x PSE ave) + (5% x PSE coarse) (5% x PEE = PEE fine) + (90% x PEE ave) + (5% x PEE coarse) (5% x PCE = PCE fine) + (90% x PCE ave) + (5% x PCE coarse) 20% PNE = 0.8.000 Notes: (a) Even though there are no filter gradings falling into the Continuing Erosion category in this example.Probability of Continuation SECTION 10 Table 10. Excessive and Continuing Erosion for the example shown in Figure 10. Some. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-14 .5% PEE = 0.4) No Erosion (NE) Fine Base Soil Grading (represents 5% of finest grading curves) Average Base Soil Grading (represents 90% of grading curves) Coarse Base Soil Grading (represents 5% of coarsest grading curves) Calculation of Probabilities for No.5% PSE = 0.11 0% PCE = 0.8. This takes into account the possibility of the materials in the dam being coarser than indicated by the gradation curves. the filter gradation envelope is significantly finer than the Continuing Erosion boundary.20 69.9 Example of Estimating Probabilities for No. Some.69 10.0001 was assigned for Continuing Erosion based on the guidance given in Table 10.4 Estimated Proportion of Filter Gradings falling into each Filter Erosion category (from Figure 10. a probability of 0.0001(a) 100% 1.

10 .001 0.75mm particle size. joint or crack in the foundation.Continuing Erosion criteria for erosion into an open defect.2 x JOSCE JOS < 0.11 – Aid to judgement for estimation of probability for continuation for open defects/joints/cracks Comparison of joint opening in the dam (JOS) to the Continuing Erosion criteria Joint Opening in the dam JOS < 0.0001 0. Estimate the conditional probability for Continuing Erosion by estimating the proportion of soil gradations that are coarser than the Continuing Erosion category and using Table 10.5 x JOSCE JOS > JOSCE Probability for Continuing Erosion (PCE) 0. in a wall or conduit) For erosion to continue through an open defect. the defect needs to be sufficiently open to allow the soil surrounding the defect to pass through it.9 Estimate based on the proportion of gradings finer than CE Notes: JOSCE = Joint Opening for Continuing Erosion Boundary = D95 of the surrounding soil A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-15 . Comparison of Soil Gradation to Joint/Defect opening size (JOS) All Soils Continuing erosion (CE) Notes: • • • Erosion condition JOSCE = D95 surrounding soil JOSCE = Joint/defect opening size that would allow continuing erosion of the surrounding soil. Table 10.1.1 – 0. D95 should be based on the average soil grading after regrading on 4.11.01 0.10.1x JOSCE JOS < 0.5 Probability for Continuation – Scenario 4 (Internal erosion into an open defect. Table 10. The recommended procedure is as follows: • Evaluate the opening size that would allow Continuing Erosion (JOSCE) of the surrounding soil using Table 10.Probability of Continuation SECTION 10 10.

Or deterioration likely to be present (e. clear outflow.Probability of Continuation SECTION 10 10.05 Video inspection of drain shows.12.e. no sinkholes) Good design and construction details (e.9 depending on severity of damage 0. thin plastic pipe.1 to 0. Estimate the probability of continuing erosion for erosion into a toe drain using Table 10.5 to 0. nearby source of tree roots) 0.005 to 0. filter surround present.95 if large openings in pipe and surrounding material has moved into pipe.0001 to 0.g. old tile drains. The assessment of erosion into a toe drain considers the observed condition of the toe drain (from video or external inspections) and the design and construction details of the toe drain.1 – 0.1.g. Table 10. proper pipe) Poor design and construction details (e.5 if some damage but no actual surrounding material has moved into pipes 0.005 Drain Inspection Video inspections indicate no deterioration or damage to the pipes Design and Construction Details Good design and construction details (e.005 0. geotextile wrap of drainage gravel). A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-16 .6 Probability for Continuation – Scenario 5 (Erosion into a toe drain) This scenario is applicable if the failure path under consideration involves a seepage path that exits into a toe drain which could lead to continuing erosion of the embankment or foundation materials.0005 to 0.g. and no external evidence of poor performance (i.12 – Probability of continuation for erosion into toe drains Probabilities for Continuing Erosion into Toe Drains 0.g. evidence of internal erosion in outflow) 0. metal pipe > 30 years. good joint details) and deterioration unlikely (metal pipe <20 years) Poor design and construction details (e.0005 to 0.g. filter surround present. geotextile surround) No video inspection of drain. broken pipe. thin plastic pipe. materials found in drain) Or external evidence of sinkholes over drain.0005 0. open joints.

the embankment zoning and the details of the foundation geology. 4) Given the exit is unfiltered the probability of continuation will be 1. Assess the filter materials and materials being eroded in terms of the Continuing Erosion criteria using the procedure in Section 10.1.7).1 Approach The probability of continuation of erosion should be estimated by: a) Assessing the probability that the exit will be a filtered or unfiltered exit.0. The conditional probability of continuation for this scenario is equal to (Punf) x 1. • • Note that in situations where a foundation filter drain or toe drain system is present. there is still the possibility that seepage paths could bypass the filter drain system due to the geological conditions.2 Probability of Filtered or Unfiltered Exit The likelihood that the exit will be filtered or unfiltered will depend on the failure path being considered.0. Examples of these scenarios are included in Table 10. The probability of continuing erosion PCE = (Punf x 1.Probability of Continuation SECTION 10 10.2. The steps are as follows.4.0) + (Pfe x PCE). The probability of continuation will be the product of the probability of an unfiltered exit and the probability assessed considering the filters.2. 10.13 and illustrated in Figures 10.1.5 to 10.2 Probability for Continuation for Internal Erosion Through the Foundation 10. • Estimate the probability that there will an unfiltered exit (Punf) using Table 10. Calculate the probability that there will be a filtered exit (Pfe) = 1. 5) Given the exit is filtered. or deteriorated toe drain systems.13 to aid judgment. compaction of soil foundations below foundation filters.4.Punf.. estimate the probability for Continuing Erosion using the method described in Section 10. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-17 .

01 to 0.01 – 0. filter drain has inadequate discharge capacity and or clean-up of the foundation hinders flow of seepage into the filter drain Filter drain overlying a foundation sand/silt layer.1 – 0.5b) Filter trench partially penetrating through foundation sand/silt layer. or desiccated foundation clay but some of this soil or in filled defects in rock remains beneath the filter trench (Figure 10.001 Foundation filter partially penetrates the foundation material that is erodible Filter drain has adequate discharge capacity 0.001 depending on the width and discharge capacity of the filter drain. clean-up of the foundation allows seepage into the filter drain 0. Low permeability layer is present below the foundation filter drain.5b) Filter trench partially penetrating through foundation sand/silt layer. in filled defects in rock or desiccated foundation clay but erodible soil or in filled defects in rock remains beneath the filter trench (Figure 10. Filter drain has adequate discharge capacity Low permeability soil layer between the foundation filter/toe drain and the foundation material that is being eroded (e. Foundation filter partially penetrates the foundation material that is erodible Filter drain has adequate discharge capacity 0.Probability of Continuation SECTION 10 Table 10.01 Foundation filter partially penetrates the foundation material that is erodible Filter drain has inadequate discharge capacity Foundation filter blanket drain directly overlying the foundation material that is erodible. clean-up of the foundation to allow seepage to flow into the filter drain and discharge capacity of filter drain 0.1 to 0.5a) Filter trench penetrating through foundation sand/silt layer.g. Figure 10. in filled defects in rock or desiccated foundation clay and into a nonerodible material (Figure 10.13 – Probability of by-passing the foundation filter for piping through the foundation or piping from the embankment into the foundation Range of Probabilities of the Seepage Path Bypassing the Foundation Filter Negligible Assume probability unfiltered exit = 0 Scenarios Foundation filter fully penetrates the foundation material that is erodible Filter drain has adequate discharge capacity Examples Filter trench completely penetrates through the foundation sand/silt layer. or desiccated foundation clay (Figure 10.1 to 0.1 0.9 depending on the degree to which filter drain capacity is exceeded 0. desiccated foundation clay (Figure 10.5c) or in filled defects in rock.5 Foundation filter blanket drain directly overlying the foundation material that is erodible. For very narrow widths the probability could be as high as 0.01 depending on the width of the filter drain.5c) or in filled defects in rock. desiccated foundation clay (Figure 10.5 – 0.5b) Filter drain overlying a foundation sand/silt layer.6) Residual soil left over jointed bedrock below the foundation filter A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-18 .

collapsed tile drains.1 – 0. the likelihood of finding an unfiltered exit is already considered in the assessment of initiation of backward erosion. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-19 . root holes.Probability of Continuation SECTION 10 Scenarios Scarifying and rolling the foundation below the foundation filter in soils with macrostructure. An unfiltered exit is implicit if a heave condition or a sand boil is present.5 0. cohesive shell materials HDPE corrugated pipes.7) Range of Probabilities of the Seepage Path Bypassing the Foundation Filter 0.9 Discharge capacity of foundation filter drainage system is not sufficient Thin sand filter layer on high permeability soil or rock foundation Filter materials contain excess of fines (>5% fines passing 0. For the situation of backward erosion piping within a foundation sand layer which has an overlying low permeability layer.0 It needs to be recognized that low permeability strata beneath horizontal drains may prevent them working effectively.1 – 0. or lateritic soils.6 shows an alluvial foundation where the lower permeability strata (A and E) will prevent the seepage in the most permeable sand and gravel strata (B. and D) from flowing into a filtered exit in the horizontal drain. Soil layer that can be piped daylights downstream of the dam Continuous open jointed rock daylights downstream of the dam (Figure 10. desiccation cracks.5 – 0.9 1.075mm) 0. Figure 10.9 Blocked or collapsed toe drains No foundation filter or toe drain system provided.1 – 0. Filter drain has adequate discharge capacity High permeability soil or rock has an unprotected exit downstream Examples Foundation soils or highly weathered rocks containing relict defects. biological growth in pipes 0.

5 – Examples of scenarios of fully penetrating and partially penetrating foundation filter drains.Probability of Continuation SECTION 10 1 2B 2A SAND NON-ERODIBLE ROCK (a) Example of fully penetrating filter drain 1 2B 2A SAND NON-ERODIBLE ROCK (b) Example of partially penetrating filter drain 1 FOUNDATION FILTER DRAIN SAND (c) Example of horizontal filter drain directly overlying erodible foundation material Figure 10. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-20 .

6 . A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-21 .Example of an embankment where there is an unfiltered exit due to day lighting of the foundation sand layer downstream of the dam.Example of an embankment where much of the seepage flow will be to an unfiltered exit. Foundation filter Unfiltered exit Sand layer Figure 10.7 .Probability of Continuation SECTION 10 Figure 10.

Use the method described in Section 10. then reduce the probability for continuing erosion using the judgment probability mapping tables in Appendix E. If the geological conditions of the soil layering are such that the open soil layers at the core-foundation contact are unlikely to daylight at an unfiltered exit. If the geological conditions of the rock mass are such that the open joints at the core-foundation contact are unlikely to daylight at an unfiltered exit.3. 10. The assessment of continuation at the soil layer contact needs to consider the potential volume of embankment material that may be stored within the open jointed rock mass before filtering takes place. then evaluate the probability of continuing erosion at both the contact with the rock defects and also at the contact with the overlying soils.3.1. or need to exit via an interconnected path of another joint set which are less open.3 Probability of Continuation for Internal erosion of the embankment at or into the foundation 10. or need to exit via an interconnected path of another soil layer which is capable of acting as an effective filtering medium.4. then reduce the probability for continuing erosion using the judgment probability mapping tables in Appendix E.. alluvium overlying a jointed rock mass).2 Erosion into coarse grained soil foundation Evaluate the probability of the coarse grained soil foundation layer being sufficiently open and continuous to allow erosion of the core materials.1. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 10-22 . Evaluate the probability of the foundation soils permitting Continuing Erosion of the core materials using the methods described in Section 10.g.5 to evaluate the probability of continuing erosion.1 Erosion into open joints in rock foundation Evaluate the probability of the joint openings being sufficiently open and continuous to allow erosion of the core materials based on the outcomes of the assessment in Sections 8 and 9.Probability of Continuation SECTION 10 10. If the geological conditions are such that the seepage through the rock mass needs to exit through an overlying soil layer (e.

Figure 11. 11. then a pipe through the foundation may not be able to fully develop.1 Internal Erosion Through the Embankment For internal erosion and piping through the dam or piping from the embankment into a rock foundation.2 Probability of Forming a Roof 11.4). Step 2: Estimate the probability that “crack filling” action will not stop the erosion process (Section 11. rockfill or gravel shells).Probability of Progression SECTION 11 11 Probability of Progression 11. In most cases the core of the embankment is capable of providing a roof to a developing pipe in the foundation. Other geological conditions which may form a roof within a soil foundation include where basalts overly the soil layer.1b shows an example of this. the roof of a pipe will be formed by layers of soil in the foundation which are cohesive or have high fines content.1.g. the core must be capable of holding the roof of a pipe. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 11-1 .1. Assess the probability of the embankment and foundation materials supporting the roof of a pipe in the foundation using Table 11.2 Internal Erosion through a Soil Foundation For internal erosion and piping through a soil foundation. However.2. Step 3: Estimate the probability that flow in the developing pipe will not be restricted by an upstream zone (or for example a concrete face slab) so the erosion process continues to develop (Section 11. Assess the probability of the soil forming a roof of a pipe using Table 11.3).2. 11.2). or by the core of the embankment. if there are upstream or downstream zones of non plastic granular material in the embankment that are not capable of supporting a roof of a pipe (e.1 Overall Approach Step 1: Estimate the probability that the soil will “hold a roof” over a pipe (Section 11.

1 Scenarios for holding a roof of a pipe for internal erosion through the foundation A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 11-2 .Probability of Progression SECTION 11 1 SAND (a) Homogeneous earthfill dam ROCKFILL OR GRAVEL SHELLS 2 3 1 3 PIPE COLLAPSE LEADING TO SINKHOLE 2 SAND (b) Dam with gravel or rockfill shells Figure 11.

GM) Granular soils with some cohesive fines (SC-SP.7 to 1.2 to 0.02 to 0.001 to 0. (SP. SC-SW. GM-GP.5 to 1.1 0.0 5% to 15% Plastic Moist Saturated 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 11-3 . sandy clays (CL.01 Lower range of probabilities is for poorly compacted materials (i. use the category that best describes the particular situation.5 5% to 15% Non plastic Moist Saturated 0. GP. and upper bound for well compacted materials. Silty sandy gravel (SM.0001 0. Silty gravels.0 > 15% Non plastic Moist Saturated 0.05 < 5% Non plastic Plastic Moist and saturated Moist and saturated 0.e. CH. GC-GP.0 1.Probability of Progression SECTION 11 Table 11. not rolled). SM-SW. GCGW) Granular soils with some non plastic fines (SM-SP.05 to 0. If soils are cemented. Gravely clays.50% Plasticity of the Fines Plastic Plastic or nonplastic Plastic Moisture Condition Moist or saturated Moist or saturated Moist or Saturated Likelihood of Supporting a Roof 1. GMGW) Granular soils.0 0.5 to 1.1 – Probability of a soil being able to support a roof to an erosion pipe Soil Classification Clays. (SC. CL-CH) ML or MH Sandy clays. SW.0 0. GC) Silty sands. Cemented materials give higher probabilities than indicated in the table. GW) Notes: (1) (2) Percentage Fines > 50% >50% 15% .0 1.

then evaluate the probability for crack filling action as for erosion in the dam using Table 11.Probability of Progression SECTION 11 11. If the eroding foundation soil layer is located immediately below the embankment. There is very little benefit where there is no downstream filter/transition zone.3. For piping through the dam or piping from the dam into a rock foundation. 11.2. This will be of greatest benefit in cases where there is poor filter compatibility between the core and downstream filter due to a lack of sand size particles in the core. the probability of continuation may be high. Estimate the probability of crack filling action not stopping pipe enlargement using Table 11. If the eroding foundation soil layer is located further below the embankment.1 Internal Erosion the Embankment Estimate the probability of crack filling action not stopping pipe enlargement using Table 11. crack filling from an upstream zone can limit the extent of erosion in the core if the materials washed into the crack or pipe are capable of filtering against the downstream filter or transition zone.2 are higher for a well graded core material compared to those for a core which is deficient in sand sizes.2.3. There needs to be a filtering material at the downstream end of the flow path for crack filling action to be effective.3 Probability of Crack Filling Action Not Being Effective 11. The filtering material may be a naturally occurring layer in the foundation. There is less benefit where the materials that are washed in are of similar sizes to those already in the core. hence the probabilities for crack filling in Table 11. then consider the potential for the overlying soil layers to wash into the developing pipe. The washed in materials aid in the filtering action against the downstream zone.3 Internal Erosion of the Embankment into or at the Foundation The issues are the same as for internal erosion through the embankment. but the washed in materials may be capable of filtering against the downstream filter zone and this reduces the potential for the pipe enlarging. and the embankment forms the roof of the pipe. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 11-4 . or the embankment filter.3. 11. In these cases.2 Internal Erosion Through the Foundation The potential for crack filling action for internal erosion in the foundation depends on the zoning of the embankment and the geological features in the foundation.2.

earthfill with horizontal drain.075 – 4.1 If the core is deficient in sand sized particles. puddle core earthfill. zoned earthfill Central and sloping core earth and rockfill (or gravel shoulders) None except for rip rap and filters under these None or none effective 1. then the potential benefits of crack filling are less as the sand size particles are already present.01 to 0.0 0.Probability of Progression SECTION 11 Table 11.75 mm) already present (1) 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 11-5 . If the core is well graded and has sand sizes present. Notes: (1) (2) Crack filling is more likely to stop pipe enlargement when the core zone is deficient in sand size particles and these particles can be provided by washing in from the upstream zone.9 Present Present If the core is well graded and has fine to coarse sand sizes (0.2 – Probability for crack filling action not stopping pipe enlargement – internal erosion through the embankment Downstream Filter or Transition or other granular material Likelihood of Piping Progressing – Crack Filling Action Not Effective Embankment Zoning Upstream Granular Zone Homogeneous. earthfill with corewall. concrete face earthfill.1 to 0. earthfill with toe drain. and washed in sand material aids in sealing the downstream zone (2). Probability dependent on compatibility of particle sizes of granular soils upstream of the core and in the downstream filter transition. hydraulic fill Earthfill with vertical and horizontal drain.0 None Present 1. This aids in sealing of the downstream filter zone.

This is embedded in the system for estimating the likelihood of breach due to the flow through the open joints in the rock foundation (Section 13). This is likely to lead to the pipe forming through the embankment. Leakage out of a pressurised conduit is likely to be limited by the defect in the pipe. estimate the probability that flow in the developing pipe will not be restricted by an upstream zone use Table 11.4. The limitation of flows is therefore not applicable to this scenario and a probability of no flow limitation of 1.Probability of Progression SECTION 11 11. • Erosion initiating at the core-foundation contact where there is no or a shallow cut off trench.1 Flow limitation by upstream zone Estimate the probability that flow in the developing pipe will not be restricted by an upstream zone using Table 11.4.3 Flow into jointed bedrock The possible scenarios are.0 should be used.3. 11. 11. Estimate the potential flows out of the defects in the conduit pipe and this is used in Section 13.0 should be used.3. For this case. This considers the potential for flow limitation due to zoning within the dam or cutoff walls or other structural elements within the dam or foundation.4. The extent of erosion may become limited by the opening width of the rock defects.4 Probability for Limitation of Flows 11. Erosion into open defects in a conduit may lead to the development of a sinkhole on the embankment. and this is considered under the breach node of the event tree (refer to Section 13. Erosion initiating within a deep cutoff trench into open joints in a rock foundation. • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 11-6 .5).2 Flow into/out of open joint in conduits Limitation of flows is not applicable to these failure modes and a probability of no flow limitation of 1.3 to estimate the likelihood for it to cause slope instability.

(1) Mechanism causing cracking or flaw in the core is also likely to affect the upstream zone (e. Homogeneous.05 to 0. clean rockfill) Fill with > 15% cohesive fines. features causing cracking or flaw in the core are not present below the upstream shell.0 1. Mechanism causing cracking or flaw in the core is also likely to affect the upstream zone (e.3 – Probability that flow in the developing pipe will not be restricted by an upstream zone.5 to 0. common cause cracking).g.0 0. features causing cracking or flaw in the core are not present below the upstream zone. earthfill with toe drain. unlikely to support a roof (1). highly likely to support a roof. features causing cracking or flaw in the core are not present below the upstream zone. Fill with 5% to 15% cohesive fines. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 11-7 . Fill with > 30% cohesionless fines. Likelihood for No Flow Restriction 1. common cause cracking). may support a roof.8 0.7 0. Assign probability depending on the confidence that there is not a common cause defect 0.4 if gradient across upstream zone is < 1 0.Probability of Progression SECTION 11 Table 11.g.1 depending on the confidence that there is not a common cause defect 0. highly likely to support a roof.01 to 0.g.3 depending on the confidence that there is not a common cause defect and fines content 0.g.g. common cause cracking).4 to 0.8 to 1.0 0. Mechanism causing cracking or flaw in the core is also likely to affect the upstream zone (e.g. Fill with 5% to 15% cohesive fines.1 to 0. cut-off wall or a concrete element in the erosion path Characteristics of Upstream Zone/Concrete Element/Cutoff Flow Limitation by an Upstream Zone.8 to 1. No zone upstream of core (e.05 to 0.2 if gradient across upstream zone is > 1 Fill with > 15% cohesive fines. may support a roof. Mechanism causing cracking or flaw in the core is also likely to affect the upstream zone (e. Fill with <15% cohesionless fines. common cause cracking). may support a roof. likely to support a roof (1).1 depending on the confidence that there is not a common cause defect 0.1 if gradient across upstream zone is <1.01 to 0. 0. likely to support a roof.0 Fill with 15% to 30% cohesionless fines. Fill with <15% cohesionless fines. unlikely to support a roof (1) Mechanism causing cracking or flaw in the core is also likely to affect the upstream zone (e.2 to 0. features causing cracking or flaw in the core are not present below the upstream shell. earthfill with filter drains) High permeability zone (e.01 to 0. common cause cracking).g. Fill with 15% to 30% cohesionless fines.9 if gradient across upstream zone is > 1 0.

5 0. 0.1 depending on the extent of coverage of the piping soil layer 0.1 depending on the confidence that there is not a common cause defect 0.01 to 0.01 to 0.01 to 0. If so.01 to 0.9. (2) Need to consider potential size of pipe and ability of downstream shoulder to handle flows. cement bentonite Non-cementitious walls – soil bentonite Soil cement bentonite wall Column walls – jet grouting.01 to 0.1 to 0. features causing cracking or flaw in the core are not present below the upstream shell. At gradients > 1. fines can wash out and lead to higher permeability and /or a pipe may develop.1 0. Concrete slab on upstream slope Soil cement wave protection Partially penetrating concrete core wall in dam (for internal erosion and piping along foundation contact) Likelihood for No Flow Restriction 0.05 to 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 11-8 .01 to 0.001(2) 0.0001 well constructed.01 to 0. and the bentonite mixed with the excavated soil using earth moving equipment. soil mixing Open joint. (3) For these walls the soil is excavated by excavator or dragline. water stop. Upstream low permeability blanket (for internal erosion in the foundation) Flow Limitation by a Concrete Element in the embankment. backward erosion is likely.5 Cold rolled – 0. crack or other defect in the conduit 0.2 Notes: (1) Need to check whether the upstream zone materials are susceptible to suffusion and backward erosion.001 to 0.0.01 0.05 to 0. Modern diaphragm walls: Cementitious walls – conventional concrete.001 for piping along the corefoundation contact (depending on height of the wall) Extruded – 0.Probability of Progression SECTION 11 Characteristics of Upstream Zone/Concrete Element/Cutoff Fill with > 30% cohesionless fines.001 serious defects suspected 0. If have good piezometers data to show wall is integral may use low end figure or possibly even lower Flow Limitation by cut-off walls in the foundation (for internal erosion and piping in the foundation). Sheet pile walls Concrete core wall within embankment (1920’s-1930’s).1 0. plastic concrete.2 0.1 to 0. may support a roof.

2. internal erosion in the embankment. or carry out repairs to prevent the dam breaching is usually best considered as two questions: 1. Intervention and Repair SECTION 12 12 Probability of Detection. whether it is possible to intervene (e. and if so. The likelihood of detection and successful intervention and repair is dependent on a number of factors including: a) b) c) d) e) f) g) h) i) j) The category of internal erosion and piping. The type and frequency of monitoring and surveillance at the dam and the training of the staff to recognize a developing internal erosion and piping incident. The overall probability of detection.Probability of Detection. Will this failure path be detected? Will intervention and repair be possible? A probability is assigned to each of these questions. by lowering the reservoir level). suffusion or backward erosion. sinkhole development. and how rapidly it can be drawn down. The reservoir level at the time of the piping incident. unravelling or sloughing of the downstream slope. settlement of the foundation. The breach mode – gross enlargement of a pipe.1 General Principles The likelihood that a particular failure path can be detected. and the materials in the embankment. The mechanism of initiation of internal erosion – erosion in a crack. The ability of those responsible to be able to direct emergency release of the reservoir The availability of materials and equipment to intervene and carry out repair works. instability of the downstream slope. i. The ability to get trained personnel out to the site in the event of a piping incident. It is necessary to use judgement to assess these probabilities. The nature of and the geometry of the materials in the foundation.e.g. The zoning of the embankment. intervention and repair is the product of these two probabilities. Intervention and Repair 12. the foundation or embankment to foundation. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-1 .

Fell et al (2001. and development of a breach.1 working from left top right. Intervention and Repair SECTION 12 12. Based on the case histories and an understanding of the physical processes they provided guidance on the time for progression beyond when a concentrated leak is first observed. and embankment to foundation. Table 12.2 Some Information on the Rate of Internal Erosion and Piping The likelihood of detection and successful intervention or repair depends on the time from when the internal erosion process may be detected to when breach begins.g.3 replaces an original table to assess the likely rate of erosion of the core of the embankment or the soil in the foundation. In these tables the terms for rates are defined as shown in Table 12.3 and 12. 12.g.Probability of Detection. Table 11.1 should be used to estimate the approximate likely time to dam failure after a concentrated leak is first observed.1. Table 12. 11. It is considered to be reasonably applicable to cases where the final breach is by slope instability. Most of the cases studied were for breach by gross enlargement. Table 12. and selecting the likely time for progression and breach which best fits the data. Table 12. very rapid – rapid for 6 hours. Tables 11.4.1 to Table 12. foundation.3 are used in Table 12. 2003) studied case histories of failures and accidents for piping in the embankment. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-2 .2. progression) or the whole process. We would expect breach to occur in a small number of hours but do not have case data to support a more refined estimate. The terms are applied to part (e. Dual descriptors are used to describe intermediate terms e. so the method is applicable to cases where the mechanism is gross enlargement. following development of a pipe. It will probably underestimate the time for breach by sloughing. Sloughing is a slowly developing breach mode which should take days or weeks to lead to breach.1 is used by assigning the values to the first four columns. Breach by sinkhole development is potentially a rapid process in the final stages when the sinkhole emerges into the reservoir.1 and Table 11.3 are based on that study.3 should be used to assess the ability to support a roof and upstream flow limitation respectively.

S = Slow. Intervention and Repair SECTION 12 Table 12. M = Medium.Qualitative Approximate Likely Time Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes R or VR R R-M R R R or R-M M or R-M M or R-M R-M or M M No No No No No No Yes No Yes Yes or No VR or R-VR R VR R-M M or S M or M-S R or R-M S S S Very Rapid Very Rapid to Rapid Rapid < 3 hours 3-12 hours 12-24 hours Rapid to Medium 1-2 days Medium Slow 2-7 days Weeks – even months or years Note: . A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-3 .3 Breach Time From Table 12.1 – A method for the approximation estimation of the time for progression of piping and development of a breach.3 Approximate Likely Time.Probability of Detection. 2003). VR = Very Rapid. and slope instability linked to development of a pipe (Fell et al 2001.2 Upstream Flow Limiter From Table 11. for breach by gross enlargement. Factors Influencing the Time for Progression and Breach Ability to Support a Roof From Table 11. R – Rapid.1 Rate of Erosion From Table 12.

2 . Qualitative Term Slow (S) Medium (M) Rapid (R) Very Rapid (VR) Equivalent Time Weeks or months. Material Description Coarse grained rockfill Soil of high plasticity (plasticity index > 50%) and high clay size content including clayey gravels Soil of low plasticity (plasticity index < 35%) and low clay size content.3 – Influence of the material in the downstream zone of the embankment on the likely time for development of a breach.2 Very Rapid Very Rapid Very Rapid Rapid Very Rapid to Rapid Rapid Rapid Rapid Rapid Rapid to Medium Medium to Slow Gradient along pipe 0.Rate of Erosion of the core or soil in the foundation Best Estimate Erosion Rate Index (IHET) <2 2 to 3 2 to 3 3 2 to 3 3 3 to 4 4 3 to 4 4 5 Time for erosion in the core of the embankment or in the foundation Gradient along pipe 0.5 Very Rapid Very Rapid Very Rapid Very Rapid Very Rapid Very Rapid Very Rapid to Rapid Rapid Very Rapid to Rapid Rapid Medium Soil Classification SM with <30% fines SM with > 30% fines SC with < 30% fines SC with >40% fines ML CL-ML CL CL-CH MH CH with Liquid Limit <65% CH with Liquid Limit > 65% Table 12. silty sandy gravels Sand.4 – Qualitative terms for times of development of internal erosion. all poorly compacted soils. Intervention and Repair SECTION 12 Table 12. silt Likely Breach Time Slow – medium Medium – rapid Rapid – very rapid Very rapid Table 12. 2003).Probability of Detection. even years Days or weeks Hours (> 12 hours) or days < 3 hours A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-4 . piping and breach (Fell et al 2001. silty sand.

3. 2003) caution however. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-5 . The dam zoning and the location of the concentrated leak and whether the leak will be visible to those doing the inspection. intervention and repair. 2. Whether detection is likely depends on: 1. and measurement of monitoring equipment. The frequency of inspections. against over-reliance of these figures for life loss estimates where the estimates are sensitive to the assumed warning times. in the advanced stages of progression and breach formation. and in Table 12. The rate at which the internal erosion and piping. For a homogeneous dam the whole of the embankment is the same soil. and hidden or unknown details within a dam or its foundation may give shorter or longer times. For example if a process may go from initiation or first presence of a concentrated leak to breach in say 6 hours. or by seepage measurement. Settlements. and associated processes. 12. 2.3 Detection 12. Detection is likely to be by: 1.Probability of Detection. 3. This may be by visual observation. Intervention and Repair SECTION 12 Note that the dispersivity of the soil does not significantly affect the rate of erosion so is not listed as a factor in Table 12. Measured higher pore pressures in the foundation and/or embankment. Fell et al (2001. or more sophisticated methods such as thermal monitoring of the foundation or the downstream slope. Observation of increased seepage out of the downstream face of the embankment or in the foundation. there is some chance the leak may be noticed non-the-less. or more likely. it is very unlikely that a piping incident will be detected before breach occurs.2 the soil is considered as the core.3. The times estimated in Table 12. such as instability of the downstream face. and the dam is only inspected or monitored weekly.2003) show that the method gives a reasonable estimate of the time for progression beyond where a concentrated leak is observed and breach and the times are acceptably accurate for the purpose here which is to assess the likelihood of detection.3 as the downstream zone. so in Table 12. However if the dam is visible by the general population. Fell et al (2001.2.1 Some General Principles Detection may be possible in the continuation or early progression phase.1 are only approximate. deformation and cracking in the embankment or area downstream of the dam. occurs.

whether it is successful is dependent on the factors discussed above. and the frequency of inspections and/or reading of monitoring instruments. 12. • • It is known that most internal erosion and piping failures occur at reservoir levels close to or above historic high. it is not practical to walk to inspect. particularly for rapidly developing piping mechanism. and the physical processes are driven by the reservoir water.6.7. so there is 30% to 50% of the time (varying throughout the year) when detection will not be effective. Dense vegetation.Probability of Detection. but also on such practical issues as: • Inspections are seldom practical at night. Hence a good monitoring and surveillance program will have a greatly increased frequency of inspections and reading of critical instruments under such reservoir conditions. so it is less likely small leaks are detected.3. particularly for erosion initiating along a crack. 2003) record that most piping incidents are first identified as a concentrated leak in the progression phase. Visual inspection is a vital tool in detecting internal erosion and piping. Fell et al (2001. This is done using • • Table 12.5 and 9.P nol ). The probability of not detecting the internal erosion P nol + [(1. it is not detected (P nd ) allowing for the time between the first appearance of the concentrated leak. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-6 .2 Assessing the Probability of Not Detecting Internal Erosion The probability of not detecting internal erosion is determined by • Assess the probability of not observing the concentrated leak (P nol ) allowing for the location of the leak for the failure mode under consideration and factors which may mean the leak cannot be observed. or by backwards erosion because the amount of leakage is very small at the start. Suffusion is more likely to be detected by piezometers because the process is slower to develop.P nol ) x (P nd )]. snow cover can all hide the presence of a concentrated leak. However it can be the case that melted snow is a good indicator of areas affected by seepage. Assess the probability that given the leak is observable (1. Many dams are not inspected on weekends. The presence of conditions potentially leading to heave and backward erosion in the foundation may also be detected by piezometers provided they are correctly positioned and read as reservoir levels rise. Intervention and Repair SECTION 12 Detection early in the internal erosion process is usually difficult. further reducing the likelihood of detection. This is done using Tables 9. For very long embankments. runoff from rainfall.

while they may be difficult to observe in the river section because the foundation is high permeability alluvium and the toe overgrown with vegetation. For example there might be a dam where leaks in the abutment area may be readily observed because the foundations are low permeability. backward erosion piping in the foundation because sand boils will form. • • A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-7 . It may be easier to detect some such mechanisms than others e. The toe of the embankment being drowned out by another reservoir or is in the river.Probability of Detection. The internal erosion mechanism.g. which may make it virtually impossible to detect leaks in this part of the dam. and the vegetation is clear. take account of: • The location of the potential leak when assessing the probabilities. Intervention and Repair SECTION 12 When using these tables.

2 0. observation of leakage is easy. etc. Central core earth and rockfill dams. Puddle core.5 0. No seepage collection system. zoned earth and rockfill Much More Likely (4) Tail-water drowns part or all of toe. or/and Vegetation at toe may preclude observation of seepage.Probability of not observing a concentrated leak (P nol ) versus ∑ (Relative importance factor (RF)) x (Likelihood factor(LF)) for internal erosion in an embankment 0. Can a concentrated leak be observed at toe? Dam Zoning which affects whether leaks emerge on the downstream face of the embankment Seepage instrumentation weirs. More Likely (3) Foundation medium permeability soil or rock so leaks may remain in the foundation and not emerge at the toe.Probability of Detection. and No vegetation or only mown grass at toe. or/and Dense vegetation at toe makes observation of seepage difficult. rockfill with core wall. Earthfill with horizontal and chimney drains. Neutral (2) Foundation low permeability soil or rock so leaks will emerge at the toe. earthfill with core wall. and Dense vegetation at toe makes observation of seepage difficult.05 0.1 0.3 0.6 .5 – Factors influencing the likelihood of not observing a concentrated leak (Pnol) Influence on Likelihood of Not Observing Factor Relative Importance (3) Less Likely (1) Foundation low permeability soil or rock. so leaks will emerge at the toe. concrete face earthfill (1) Seepage collected to flow to readily observed measuring weir or real time monitored Seepage partly collected to flow to measuring weir Seepage partly collected to flow to measuring weir but masked by rainfall effects Table 12. or foundation permeable alluvium so leaks may not emerge at the toe. Zoned earthfill dam. Intervention and Repair SECTION 12 Table 12. concrete face rockfill.9 6 9 12 15 18 24 RF x LF A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-8 . Earthfill with toe drain. (2) Homogeneous.

95 0.25 0.1 0.1 0.Probability of Detection. Intervention and Repair SECTION 12 Table 12.5 0.6 0.01 0.5 0.9 0.2 0.7 0. no public nearby Daily.1 0.05 0.15 1 to 2 days 0. no public nearby Weekly.999 0.05 0.05 0.6 0.01 Frequency of Inspection and /or Monitoring < 3 hrs Monthly.1 0.05 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-9 .5 0. public nearby Weekly.75 0.8 0.9 0. no public nearby Monthly.01 0.999 0.9 0.2 0.99 0. and the frequency of inspections and/or reading of monitoring instruments Probability of Not Detecting the Internal Erosion (Pnd) Given the Time for Development of Concentrated Leak to Initial Breach From Table 12.1 0.4 0.99 0.95 0.05 0. public nearby Daily.8 0.7 – Probability that given the leak is observable it is not detected given the time between the first appearance of the concentrated leak.2 3 to 12 hrs 0. public nearby Daily with real time monitoring of leakage 0.5 0.1 12 to 24 hrs 0.1 0.1 2 to 7 days 0.1 Weeks or months 0.99 0.

Table 12.Probability of Detection.8 should be used to assess the probability that given the concentrated leak is detected. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-10 . (ii) Installing pressure relief wells in the foundation of the embankment. or slope instability. (iv) Building a weighting berm to reduce the likelihood of heave. (iii) Building reverse filters over “boils” or areas where eroding material is emerging from the foundation of the embankment. This is done for each pool (reservoir) level partition. In making this assessment consideration should be made for the failure mode and location of the developing pipe. Which is applicable or feasible will depend on the particular circumstances of the dam. or unravelling. intervention and repair is not successful. More than one of these measures may be used together.4 Assessing the likelihood of Intervention and Repair Intervention and repair to prevent the progression of internal erosion and piping and breach can take several forms including: (i) Drawing down the reservoir level using spillway gates or outlet valves. It is not practical to cover all the possible scenarios and those doing the risk analysis are required to make their assessment within the range of probabilities shown. (v) Dumping granular material (sand/gravel/rockfill) into the upstream side of sinkholes to try to block them. It should be recognised that there may be reluctance on the part of the reservoir owner or operator to release water given the lost revenue that may result. or if release of reservoir water is likely to result in property damage and loss of life. for example if levee banks downstream of the dam are likely to be overtopped by the flood resulting from release of the water. Intervention and Repair SECTION 12 12.

8 (a) Note: (a) Use values less than 0.9 to 0.1 to 0.5 only if there is a high degree of confidence in the assessed value. In some cases it will be practical to intervene successfully in this time. In many cases it will be impractical to intervene successfully in this time. and there are personnel. Only in cases where there is a straight forward method of intervention. or it is a small storage or medium storage with large gate discharge capacity which can be drawn down to stop the failure mode. and there are personnel. In cases where there is a straight forward method of intervention. and there are personnel. or it is a small storage which can be drawn down to stop the failure mode.99 1 to 2 days 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-11 . intervention and repair is not successful (P ui ) Time for Development of Concentrated Leak to Initial Breach < 3 hrs 3 to 12 hrs What can be done Probability of Not Intervening 0.8 – Assessment of the probability that given the concentrated leak is detected. equipment and materials available. or it is a small or medium storage with large gate discharge capacity allowing the reservoir to be drawn down to stop the failure mode 12 to 24 hrs 0.Probability of Detection. equipment and materials available will intervention be successful.2 to 0. Only in cases where there is a straight forward method of intervention.99 There is too little time to successfully intervene regardless of the failure mode In most cases it will be impractical to intervene successfully in this amount of time. or it is a small storage or medium storage with large gate discharge capacity allowing the reservoir to be drawn down to stop the failure mode.99 0. equipment and materials available and large resources intervention has a fair chance of being successful. and there are personnel.7 to 0. Intervention and Repair SECTION 12 Table 12. In some cases it will be practical to intervene successfully in this time.95 2 to 7 days 0. and there are personnel.85 to 0. equipment and materials available will intervention be successful. Only in cases where there is a straight forward method of intervention. Where there is a straight forward method of intervention.9 (a) Weeks or months 0. equipment and materials available will intervention be successful. In many cases it will be impractical to intervene successfully in this time.

1 0.1 – Sub-event tree for calculating the probability of not intervening.225 Intervention is Suc cessful 0. Or Probability that intervention fails = P nol + [(1. Intervention and Repair SECTION 12 12.45 No Intervention Not Detected in Time 0.9 0.5 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 12-12 . Probability of Not Intervening 0.5 0.Probability of Detection.1 No Intervention Internal Erosion is Not Able to be Detected Yes No 0.P nol ) x P nd x P ui ] This calculation is represented by the sub-event tree structure shown in Figure 12.5 Interve ntion and Repair is Not Succ essful No 0.5 0.5 Calculation of Probability of Not Detecting and Not Intervening The probability of not detecting and not intervening is calculated as follows.225 No Intervention No 0.1.225 Probability of Not Intervening = 0.45 Yes 0.775 Figure 12. Probability that intervention fails = [Probability of not observing the concentrated leak because it is not observable] + [(Probability it is observable) x (Probability it is not detected in the time between concentrated leak and breach) x (Probability that intervention and repair is not successful)].1 Yes Not Intervene Tree 0.

0 when on the some or excessive erosion branches of the event tree. Section 10.5. This involves estimating the probability of slope instability occurring and the probability of loss of freeboard given instability. This is considered in Section 13.Probability of Breach SECTION 13 13 Probability of Breach 13.Psu) x (1 .1 Overall Approach and Screening For each general failure mode including internal erosion in the embankment.2. in soil or rock foundations. Step 4: Estimate the probability of breach by unravelling or sloughing of the downstream slope of the embankment using Section 13.1 in Section 1 shows the dam zoning types.Pge) x (1 . the seepage flows that develop through the dam or foundation will be limited by the filtering material.4. In most cases the probability of breach will be 1.1 Screening of Breach Mechanisms For most dam types and failure modes.1.3. The probability of breach is often less than 1. the likelihood for breach development will be dominated by one or two of the potential breach mechanisms.Psd)]. The probability of breach will depend on how well the dam can cope with the leakage flows that may occur.0 when it is on the continuing erosion branch of the event tree.1. using the appropriate statistical summation Pbreach = 1 – [(1 . Step 3: Estimate the probability of breach by instability of the downstream slope using Section 13. Step 2: Estimate the probability of breach by gross enlargement of the pipe using Section 13. and from embankment to foundation: Step 1: Screen the breach mechanisms depending on the dam zoning type using Table 13. Step 5: Estimate the probability of breach by sinkhole development using Section 13.1. Table 13.1 also applies to internal erosion in a soil foundation. Step 6: Estimate the overall probability of breach by adding the probabilities for each of the four mechanisms. Table 13.4 provides guidance on the possible magnitude of leakage flows that may develop for the some and excessive erosion conditions. Figure 1. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-1 .Psi) x (1 . For these cases.2. 13. Exceptions to this will be if the pool level drops below the inlet of the developing pipe before a breach mechanism has time to develop. Breach mechanisms will not necessarily be applicable to some dam zoning types or modes of piping and can be ignored.1 lists those breach mechanisms which should be considered in the assessment depending on the dam zoning type and mode of internal erosion for internal erosion in the embankment due to a crack or poorly compacted zone.

except if downstream fill is cohesionless Exclude. and usually is the more critical mechanism A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-2 . except if existing dam has marginal stability * Zoned earthfill and rockfill Central core earth and rockfill (or gravel shells) Concrete face earthfill * Exclude. except if downstream fill is cohesionless Exclude. and from embankment into foundation Breach Mechanisms Dam Zoning Type Gross Enlargement * Slope Instability Sloughing or Unravelling Exclude. except if downstream fill is cohesionless * Exclude Concrete face rockfill (including gravel fill) Puddle core earthfill Exclude Exclude. except if downstream fill can support a roof Exclude. except if downstream fill is cohesionless * Earthfill with corewall Exclude Rockfill with corewall Exclude Exclude. except if downstream fill can support a roof * Key: * Breach mechanism should be included for the probability estimate Breach mechanism should be included for probability estimate. except if downstream fill can support a roof Exclude. except if downstream fill is cohesionless Exclude. except if downstream fill is cohesionless * Exclude. except if existing dam has marginal stability Hydraulic fill Exclude. except if downstream fill is cohesionless * Sinkhole Development Homogeneous earthfill Earthfill with filters * Earthfill with rockfill toe * Zoned earthfill Exclude. except if dam is gravel or low permeability * Exclude.Probability of Breach SECTION 13 Table 13. except if downstream fill can support a roof Exclude. internal erosion in soil foundations.1 – Screening of breach mechanisms for internal erosion through the embankment.

1 does not apply to failure modes involving open or in filled defects and solution features in rock foundations because the leakage flows may exceed the capacity of even free draining rockfill. Breach by slope instability or sloughing/unravelling are usually the more critical mechanisms for suffusion. breach by gross enlargement can be considered negligible in cases where the downstream shell is unable to support a roof of a pipe. Assign probability of breach by gross enlargement Pge = 0. If applicable. Breach by gross enlargement of the pipe can be considered to be negligible when on the “Some Erosion” and “Excessive Erosion” branches of the event tree. 13. although the probabilities for breach are usually relatively low for this mode internal erosion. Breach by slope instability or unravelling/sloughing are likely to be more critical. This considers whether the reservoir will drop below the level of the pipe before the enlarging pipe develops into a breach. Internal erosion by the process of suffusion is very unlikely to lead to the formation of a pipe through the dam or its foundation.2 – Probability of breach by gross enlargement of the pipe – Ability to support a pipe (Screening) Downstream shell/zone Downstream shell comprises free draining rockfill. <5% plastic fines or <15% non plastic fines Ability to Support a Roof Very unlikely for piping through dam Probability of Breach by Gross Enlargement Not a likely mode of breach for piping through the dam. Assess using Table 13. • Breach by gross enlargement of the pipe requires a continuing erosion condition. Use Table 13. For internal erosion through the embankment. soil foundation.2.2 Estimation of the Probability of Breach by Gross Enlargement 13. estimate the probability of breach by gross enlargement using Table 13. and hence the probability of breach by gross enlargement where the mode of erosion is suffusion can be excluded. or coarse sandy gravel Downstream shell comprises sand and gravel.3 • • All other cases Likely A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-3 . Table 13.1 Screening for internal erosion in the embankment.3. and embankment into foundation.2 to assess if this applicable.Probability of Breach SECTION 13 Table 13.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-4 . If there is no intervention.2. in-filled caverns or caves in karst > 2m).g. This will not happen unless the reservoir level drops giving a lower gradient.2 Screening for internal erosion in a rock foundation. the pipe must stay open until it is so large that the settlement of the crest due to the pipe. the process can only stop if one or more of the following occurs: a) The hydraulic shear stresses in the pipe reach an equilibrium condition with the erosion resistance of the soil. 6) The reservoir empties or falls below the entrance of the pipe before a breach mechanism is able to develop. and no intervention to lower the reservoir level. 13.2. as the hydraulic shear stress increases with hole diameter for a constant gradient.Probability of Breach SECTION 13 13.3 Assessment for internal erosion in the embankment. Breach by gross enlargement is only possible for internal erosion through rock foundations if the in-filled defect is sufficiently large such that the embankment may collapse into the void and lowers the crest to below the reservoir level. This is a common consideration where internal erosion may develop in the upper part of the dam under short duration flood loading conditions. Table 13. or collapse of the embankment into the pipe lowers the crest to below the reservoir level. soil foundation and from embankment into foundation For breach to occur by gross enlargement of a pipe. Ignore this failure mode unless the size of the in filled defect is very large (e.3 provides guidance on the probability of breach by gross enlargement assuming there is no restriction on flows.

3 0.0 >2 days 1-2 days <1 day > 2 weeks 1 – 2 weeks <1 week Likely to self limit 0.95 0.1 CH with Liquid Limit > 65% 6 13. The basis for the judgemental probabilities is given in the supporting information document. dispersive soils CL.95 0. Estimate the probability that the amount of crest subsidence is sufficient to cause loss of freeboard.6 to 0. ML. Time for Reservoir Level to Fall Below the Invert of the Pipe Probability of Breach by Gross Enlargement 1. MH or CH with Liquid Limit <65% Hole Erosion Index <=3 2 to 3 4 3 to 5 5 4 to 6 CH with Liquid Limit <65% Or CH with Liquid Limit > 65% HET carried out Notes: (1) (2) (3) Assuming there is no restriction on flows and no intervention to lower the reservoir level IHET from Hole Erosion Tests.8 to 0.8 0. SC.3 – Probability of breach by gross enlargement of the pipe Characteristics of Core Material Soil Classification SM.05 to 0. This needs to be considered on a case by case basis using the methods described in Appendix E.2.3 to 0.4 Assessment for internal erosion in a rock foundation • • Estimate the probability that the roof over the void collapses and causes the crest to subside.8 0.3 to 0.Probability of Breach SECTION 13 Table 13. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-5 . CL-CH.1 to 0.6 0.8 to 0.

10 and Table 13. soil foundation.3 and Tables 13.1 Approach • For internal erosion in embankments. • • • 13. and from embankment into foundation The assessment considers whether internal drainage measures in the dam are able to prevent pore pressures rising in the dam and/or foundation and whether the factor of safety of the dam falls below 1. soil foundations and from embankment to foundation.3.3. Estimate the probability of loss of freeboard due to instability (Psi-lf) using Table 13. It is assumed that seepage in a soil foundation will exit under the embankment.3 Estimation Of The Probability Of Breach By Slope Instability 13.3.4 and Table 13. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-6 .5. estimate the probability of slope instability occurring due to the increased seepage flows (Psi-i) using Table 13.4 and Table 13.11.0 if pore pressures do increase.2 Estimation of the probability of slope instability initiates for internal erosion in the embankment. For internal erosion in rock foundations. use Section 13.5.6 to 13. Estimate the probability of a downstream slide initiating using Table 13.Probability of Breach SECTION 13 13.9. The probability of breach by slope instability is equal to (Psi-i) x (Psi-lf).

or high fines content weathered rockfill in the downstream zones Much More Likely (4) None No or limited zoning of materials and no filter drains Internal drainage measures in dam (2) 3H:1V or flatter 2.Probability of Breach SECTION 13 Table 13. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-7 .3H: 1V Downstream shell materials (1) Sandy gravel <5% fines. or moderate fines rockfill in the downstream zones More Likely (3) Limited Filter drain with excessive fines. silty sandy gravel. Or silty sandy gravel. puddle core earthfill. Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Good Filter drains with good discharge capacity Or free draining rockfill or clean sandy gravel in the downstream zones Downstream Slope For dams with an earthfill downstream zone (a) Neutral (2) Moderate Single stage filter zones Or sandy gravel. free draining rockfill Sandy gravel 520% fines ‘Dirty’ rockfill Cohesive soils. zoned earthfill.75H: 1V 1.5H: 1V Steeper than 1. Coarse grained. earthfill with filter. homogeneous earthfill. earthfill with core wall and hydraulic fill. concrete face earthfill.4H: 1V Steeper than 1.5H:1V 2H:1V Steeper than 1. poor discharge capacity. and from embankment into foundation.4 – Factors which influence likelihood of breaching by instability of the downstream slope – Slide Initiates for internal erosion in the embankment.8H:1V OR For dams with a free draining rockfill downstream zone (b) Flatter than 1. in soil foundations. concrete face rockfill and rockfill with core wall. 20-50% fines Notes: (a) (b) Applies to the following dam types. central core earth and rockfill. Fine grained rockfill Silty sand. zoned earth and rockfill. earthfill with rock toe. Applies to the following dam types.

003 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-8 .Estimation of the probability of breach by slope instability – slide initiates for internal erosion in the embankment.8 and Table 13.0 if pore pressures do increase.00001 0.7. in soil foundations.001 0.0005 negligible 0.005 0.5 0. EE = Excessive Erosion branch. • The assessment considers whether the internal drainage measures in the dam are able to prevent pore pressures rising in the dam and/or foundation and whether the factor of safety of the dam falls below 1. CE = Continuing Erosion branch. 13.3.1 0.001 0.5 .9 0.002 0.9.Probability of Breach SECTION 13 Table 13.05 0. Table 13. Estimate the probability of slope instability initiating due to the increased leakage flows (Psi-i) using Table 13.3 Estimation of the probability of Slope Instability of the embankment initiates for internal erosion in a rock foundation • Estimate the probability that leakage through the rock foundations exits into the downstream shell PS using Table 13. and SE = Some Erosion branch.02 0.05 (CE) (EE) (SE) 6 9 11 13 18 24 Note: Select the probability scale corresponding to the filter erosion condition being considered on the event tree.6.0001 negligible 0. and from embankment into (Relative importance factor (RF)) x (Likelihood factor(LF)) foundation versus ∑ 0.005 0.01 0.

A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-9 .0 (a) Note (a) The geometry of the leakage flow path affects the flow rate. and the defect is in direct contact with downstream shell 0. and the defect is not in direct contact with downstream shell. and the defect is likely to be in direct contact with downstream shell The open or in-filled defect in the rock foundation does not daylight downstream of the dam.1 to 0. Evaluate the relative discharge capacity of the foundation drains and downstream zone compared to the size of the defect in the rock foundation using Table 13. and the defect is not in direct contact with the downstream shell. Estimate the probability of a downstream slide initiating using Table 13.6 – Probability of seepage exits from defects or solution features in a rock foundation into the downstream shell (PS) Probability of the Seepage Path Exiting into the Downstream Shell Negligible.7. and there is very limited interconnectivity through other joint sets The open or in-filled defect in the rock foundation daylights downstream of the dam. If the leakage path daylights downstream of the dam it is likely there will be less flow into the downstream shell of the dam.0 (a) 1. but there is a likely connection into the downstream shell via an interconnected open joint set The open or in-filled defect in the rock foundation daylights downstream of the dam.Probability of Breach SECTION 13 Table 13.5 to 1.5 0. Adopt PS = 0 Scenarios The open or in-filled defect in the rock foundation daylights downstream of the dam.9.8 and Table 13.

Probability of Breach

SECTION 13

Table 13.7 – Assessment of size of leak in defect or solution feature in a rock foundation relative to discharge capacity of foundation drains and downstream shell

Discharge capacity of foundation drain and downstream zone Limited Foundation filter drain with excessive fines, poor discharge capacity, Or silty sandy gravel, or high fines content weathered rockfill in the downstream zones Moderate Single stage foundation filter zone Or sandy gravel, or moderate fines rockfill in the downstream zones Good Foundation filter drains with good discharge capacity Or free draining rockfill or clean sandy gravel in the downstream zones

Width of Defect or solution feature exiting into the Downstream Zone

Poor No or limited zoning of materials and no foundation filter drains

<5mm 5 – 25mm 25 – 100mm 100 – 300mm >300mm

N ML MML MML MML

N ML MML MML MML

LL N ML MML MML

Negligible LL N ML MML

Note: LL = Less Likely; N = Neutral; ML = More Likely; MML = Much More Likely

13-10

Probability of Breach

SECTION 13

Table 13.8 – Factors which influence likelihood of breaching by instability of the downstream slope –- Slide Initiates for internal erosion in rock foundation

Likelihood Factor (LF)

Factor Relative Importance Factor (RF) (3) Less Likely (1) LL Refer to Table 13.7 N Refer to Table 13.7 Neutral (2) More Likely (3) ML Refer to Table 13.7 Much More Likely (4) MML Refer to Table 13.7

Size of leak relative to discharge capacity of the foundation drains and downstream shell (from Table 13.7) Downstream Slope For dams with an earthfill downstream zone (a)

(2) 3H:1V or flatter Flatter than 1.75H: 1V 2.5H:1V 2H:1V Steeper than 1.8H:1V Steeper than 1.3H: 1V

OR For dams with a free draining rockfill downstream zone (b) Downstream shell materials (1)

1.5H: 1V

Steeper than 1.4H: 1V

Sandy gravel <5% fines, Coarse grained, free draining rockfill

Sandy gravel 520% fines ‘Dirty’ rockfill

Cohesive soils, Fine grained rockfill

Silty sand, silty sandy gravel, 20-50% fines

Notes:

(a) (b)

Applies to the following dam types; homogeneous earthfill, earthfill with filter, earthfill with rock toe, zoned earthfill, concrete face earthfill, puddle core earthfill, earthfill with core wall and hydraulic fill. Applies to the following dam types; zoned earth and rockfill, central core earth and rockfill, concrete face rockfill and rockfill with core wall.

Table 13.9 - Estimation of the probability of breach by slope instability – slide initiates (Psi-i) (Relative importance factor (RF)) x for internal erosion in rock foundations versus

∑

(Likelihood factor (LF))

0.001

0.003

0.006

0.02

0.1

0.9

6

9

11

13

18

24

13-11

Probability of Breach

SECTION 13

**13.3.4 Loss of Freeboard due to Slope Instability
**

The assessment considers whether the resulting sliding deformations are sufficient to result in loss of freeboard so the reservoir overtops the dam crest. Estimate the probability of loss of freeboard using Table 13.10 and Table 13.11.

**Table 13.10 – Factors which influence likelihood of breaching by instability of the downstream slope – loss of freeboard
**

Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) > 7% Neutral (2) ≈ 5% More Likely (3) < 3% Much More Likely (4) < 1%

Freeboard compared to dam height at the time of incident Presence of strain weakening soils in the embankment and foundation

(2)

Sandy clays, low to medium plasticity, clay size content <20%, or medium dense to dense dilative non cohesive soils or rockfill. > 30 ft (9m)

Clays, sandy clays, clay size content 20% to 40%, or medium dense non cohesive soils.

Clays, sandy clays, high plasticity; clay size content (% passing 0.002 mm) > 40% or/and saturated, very loose sand, or loose silty sand contractive on shearing < 13 ft (4m)

As for more likely, but with very high clay size content or very loose contractive granular soil

Crest width

(1)

≈ 20 ft (6m)

≤ 10 ft (3m)

Table 13.11 - Estimation of the probability of breach by loss of freeboard (Relative importance factor (RF)) x (Likelihood factor(LF)) versus

∑

0.001

0.005

0.02

0.1

0.5

1.0

6

9

11

13

18

24

13-12

Probability of Breach

SECTION 13

13.4

Estimation of the Probability of Breach by Sloughing or Unravelling

13.4.1 Approach

For sloughing to occur, the downstream face would have to be relatively steep, and the shoulder material a cohesionless soil, probably sandy gravel, or gravely sand, possibly with some silty fines. The process would have to be allowed to continue until it gradually eroded away the crest and allowed the reservoir to overtop the embankment. Unravelling usually relates to the progressive removal of individual rocks by fairly large seepage flows flowing through the downstream rockfill. The approach is: • For internal erosion in rock foundations, Estimate the probability that seepage through the rock foundations exits into the downstream shell PS using Table 13.6. For internal erosion in the embankment, soil foundation and embankment into foundation, assume that the seepage will emerge into the downstream shell of the embankment, so PS = 1.0. For dams with a downstream zone of earthfill (i.e. clay, silt, sand or gravel) use Tables 13.12, 13.13 and 13.14 to estimate the probability of breach by sloughing Psl. For dams with a downstream zone of rockfill, use Tables 13.15, 13.16 and 13.17 to estimate the probability of breach by unravelling Pun. The probability of breach by sloughing or unravelling is equal to (PS) x (Psl) for dams with a downstream zone of earthfill or (PS) x (Pun) for dams with a rockfill shell.

•

•

•

•

For cases where a very coarse rockfill toe is present, the probabilities for breach by unravelling can be reduced from those in given in Table 13.16 if calculations indicate that unravelling is very unlikely to occur.

13-13

Probability of Breach

SECTION 13

**Table 13.12 – Factors which influence likelihood of breaching by unravelling – dams with an earthfill downstream zone
**

Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Cohesive soils Neutral (2) Sandy gravel <20% fines, More Likely (3) Silty sand, silty sandy gravel, 20%-50% non plastic fines. < 6 ft (2 m) 2H: 1V Much More Likely (4) As for more likely, but uncompacted materials < 3 ft (1 m) Steeper than 1.8H: 1V

Material in downstream zone Freeboard at the time of incident Downstream slope of the embankment

(2) (1)

> 13 ft (4 m) 3H:1V or flatter

≈ 10 ft (3 m) 2.5H: 1V

Table 13.12 applies to the following dam types; homogeneous earthfill, earthfill with filter, earthfill with rock toe, zoned earthfill, concrete face earthfill, puddle core earthfill, earthfill with core wall and hydraulic fill.

Table 13.13 - Estimation of the probability of breach by sloughing (earthfill) versus (Relative importance factor (RF)) x (Likelihood factor(LF)) for internal erosion in

∑

the embankment, in soil foundations, and from embankment into foundation

1.0 0.01 0.001

1.0 0.05 0.003

1.0 0.1 0.01

1.0 0.5 0.05

1.0 0.9 0.1

1.0 1.0 0.5

(CE) (EE) (SE)

6

9

11

13

18

24

Note: Select the probability scale corresponding to the filter erosion condition being considered on the event tree. CE = Continuing Erosion branch, EE = Excessive Erosion branch, and SE = Some Erosion branch.

13-14

Probability of Breach

SECTION 13

Table 13.14 - Estimation of the probability of breach by sloughing (earthfill) versus (Relative importance factor (RF)) x (Likelihood factor (LF)) for internal

∑

erosion in rock foundation

1.0 0.01 0.001

1.0 0.05 0.003

1.0 0.1 0.005

1.0 0.5 0.05

1.0 0.9 0.1

1.0 1.0 0.5

(>25mm) (5-25mm) (<5mm)

6

9

11

13

18

24

Note: Select the probability scale corresponding to the width of the open or in filled defect that exits into the downstream shell. Table 13.15 – Factors which influence likelihood of breaching by unravelling – dams with a rockfill downstream zone

Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) Coarse grained free draining rockfill. Flatter than 1.75H: 1V > 13 ft (4 m) Neutral (2) Medium grained “dirty” rockfill 1.5H: 1V More Likely (3) Fine grained rockfill Much More Likely (4) As for more likely, but uncompacted materials Steeper than 1.3H: 1V < 3 ft (1 m)

Material in downstream zone Downstream slope of the embankment Freeboard at the time of incident

(2)

Steeper than 1.4H: 1V < 6 ft (2 m)

(1)

≈ 10 ft (3 m)

This table applies to the following dam types; zoned earth and rockfill, central core earth and rockfill, concrete face rockfill and rockfill with core wall.

13-15

0005 0. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-16 .6 0. EE = Excessive Erosion branch. Table 13.Estimation of the probability of breach by unravelling (rockfill) versus (Relative importance factor (RF)) x (Likelihood factor (LF)) for internal erosion ∑ in rock foundation 0.5 0.1 0. zoned earth and rockfill.6 0. and from embankment into foundation 0.005 0.7 0.005 0.5 0.01 0.7 0.05 (5-25mm) 0.9 0. Note: Select the probability scale corresponding to the width of the open or in filled defect that exits into the downstream shell.001 0.17 .05 (CE) (EE) (SE) 6 9 11 13 18 24 Note: Select the probability scale corresponding to the filter erosion condition being considered on the event tree. CE = Continuing Erosion branch.0001 0.8 0.01 0.9 0.8 0. concrete face rockfill and rockfill with core wall.6 0.7 0.05 (<5mm) 6 9 11 13 18 24 This table applies to the following dam types.001 0.0 (100-300mm) 0.5 (25-100mm) 0.0001 0.0005 0.01 1.5 0.005 0.0001 0.8 0.Probability of Breach SECTION 13 Table 13.5 0.001 0. central core earth and rockfill.Estimation of the probability of breach by unravelling (rockfill) (Relative importance factor (RF)) x (Likelihood factor(LF)) for internal erosion in versus ∑ the embankment.001 0.05 0.0005 0. and SE = Some Erosion branch.0 0.1 0.005 0.001 0.05 0.01 0. in soil foundations.01 1.0 (>300mm) 1.9 0.005 0.16 .

5.5.6 0.3 Mode of Internal Erosion Internal erosion in the embankment and into the foundation Internal erosion in the foundation 13. Estimate the probability of loss of freeboard due to sinkhole formation using Table 13.Probability of Breach SECTION 13 13. Table 13. For internal erosion in the foundation.18 – Probability of a sinkhole or crest settlement developing (Ps-f) Probability of Sinkhole or Crest Settlement Developing given Internal Erosion has Initiated 0. loss of freeboard can also occur by excessive settlement of the embankment induced by the loss of foundation materials. Estimate the probability that the sinkhole causes loss of freeboard (Ps-lf).20. The probability of breach by sinkhole development = probability of sinkhole (Ps-f) x probability of loss of freeboard due to the sinkhole (Ps-lf).2 Probability of Sinkhole Formation Estimate the probability of a sinkhole developing as a result of the internal erosion (Ps-f) using Table 13.5.5 Estimation of the Probability of Breach by Sinkhole Development 13. Assume the sinkhole develops on the crest unless there is a specific reason to expect it to develop elsewhere on the embankment. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-17 .1 Approach • • Estimate the probability of a sinkhole developing as a result of the internal erosion (Ps-f).19 and Table 13.3 Probability of Loss of Freeboard due to Sinkhole Formation For breach to occur by sinkhole development into an erosion pipe in the embankment.18. • 13. the sinkhole or crest settlement would need to be sufficiently large to settle the crest to below reservoir level.

20 .1 0.Estimation of the probability of breach by sinkhole development – loss of (Relative importance factor (RF)) x freeboard given sinkhole develops versus ∑ (Likelihood factor (LF)) 0. and sandy clays < 13 ft (4 m) Non-cohesive.19 – Factors which influence likelihood of breaching by sinkhole development – loss of freeboard given sinkhole develops Likelihood Factor (LF) Factor Relative Importance Factor (RF) (3) Less Likely (1) > 13 ft (4 m) Neutral (2) ≈ 10 ft (3 m) More Likely (3) < 6 ft (2 m) Much More Likely (4) < 3 ft (1 m) Freeboard at the time of the incident Width of crest Material in the core of the embankment (2) (1) > 30 ft (9 m) High plasticity clay. poorly compacted/loose.0005 0.001 0. well compacted ≈ 20 ft (6 m) Low to medium plasticity clays.01 (CE) (EE) (SE) 6 9 11 13 18 24 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 13-18 .005 0.2 0.0002 0.Probability of Breach SECTION 13 Table 13.005 0. silty sand or silty sandy gravel ≤ 10 ft (3 m) As for more likely. Table 13.00005 0.0005 0.001 0.02 0.0001 0.0001 0.002 0.0002 0.0002 0.01 0.

Best practices for design. R-377. M. Y. Cyganiewicz. Majors. Foster. No. The University of New South Wales. R. Foster. 307-314.. C. Cyganiewicz. Beijing. (2003). FEMA (2005) Conduits through Embankment Dams. Kulkarni. Methods for estimating the probability of failure of embankment dams by internal erosion and piping – piping through the embankment. ASCE Journal of Geotechnical and GeoEnvironmental Engineering. Assessing embankment dams.. M. R. (1999). The University of New South Wales.F. School of Civil and Environmental Engineering. M.. R. (2005) Methods for estimating the probability of failure of embankment dams by internal erosion and piping in the foundation and from embankment to foundation.F. R. R-428. Geotechnical and Geoenvironmental Engineering. and Wan. ICOLD.A. inspection. ISBN 85841 395 7. ASCE. R. Vol.. J. UNICIV Report No. M. and Fell. 398-407. L-266. R. J. ISBN: 84841 366 3. 1. No. Fell. Sydney. C.. Foster. 129. construction. The University of New South Wales. ISBN: 85841 343 4. (2001). ISSN 0077-880X. renovation and repair. Assessing Embankment Dam Filters Which Do Not Satisfy Design Criteria.A. and Fell. Paris. Time for development of internal erosion and piping in embankment dams. problem identification and evaluation. School of Civil and Environmental Engineering.References SECTION 14 14 Barneich.ISBN: 85841 403 1. Vol. M. maintenance.. University of New South Wales. School of Civil and Environmental Engineering. and Foster. Application of reliability analysis in the Environmental Impact Report (EIR) and design of a major dam project.4.F. and Fell. M. (1999b). Fell. Proceedings of Uncertainty 1996. ISBN: 85841 343 4. Vol. A Framework for Estimating the Probability of Failure of Embankment Dams by Piping Using Event Tree Methods. Foster. UNICIV Report No. UNICIV Report No R-436. C. (1999a). Federal Emergency Management Agency. J.. M. Wan. Fell. The University of New South Wales. The University of New South Wales. and Foster. J. 237-260. Australia 2052. C. D.A. and Fell. M.. (2000).F. ASCE. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 References 14-1 . R. 20th Congress on Large Dams.. Use of Event Trees to Estimate the Probability of Failure of Embankment Dams by Internal Erosion and Piping. Fell. UNICIV Report No. (2004). PhD thesis. R. filters which do not satisfy design criteria. R. and Foster. R. Wan. May 2001. R-399. The time for development and detectability of internal erosion and piping on embankment dams and their foundations. (2001).4.127. Moriwaki. UNICIV Report No. R-376. The probability of failure of embankment dams by internal erosion and piping. Sydney. Australia. Wan. Foster. and Davidson. Geotechnical Engineering Division. School of Civil and Environmental Engineering.

The non-filter factor of safety against piping through sands. The University of New South Wales. A new model to deal with the piping mechanism on “Filters in Geotechnical and Hydraulic Engineering. Geotechnical Testing Journal. 4. Montreal. and Fell. and Dunnigan. Geotechnique 44. S. UNICIV Report No.H. and Brogan. Judgment and innovation. J. No.F. Investigation of internal erosion and piping of soils in embankment dams by the slot erosion test and the hole erosion test. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 14-2 . The University of New South Wales. Weijers. R. ASCE Geotechnical Special Publication No. Wan.F. Critical seepage gradients beneath embankment dams. Wan. (2004). 111. 130.3. International Commission on Large Dams.429. and Fell. R. Sherard. Vol.A and Sellmeijer.F.L. (2004a). No. Sydney. Experiments on piping in sandy gravels. (2003). ISBN: 85841 375 2. (2000). ISBN: 85841 379 5. Experimental investigation of internal instability of soils in embankment dams and their foundations. J.B. 927-947. C.P. R. (2004). Bachelor of Civil Engineering thesis.B. Critical filters for impervious soils. J. 295-303. Brauns. 373-380. Damage and Cracking of Embankment Dams by Earthquakes.115. R.F.M. Paris. Reston. School of Civil and Environmental Engineering. R. 449-460. Balkema. R. C. School of Civil and Environmental Engineering. Paris Q83-R17. Kavazanjian. (2004). J.F. L. Eng. Rotterdam. ASCE Journal of Geotechnical and GeoEnvironmental Engineering. and Fell. The University of New South Wales. S. R-412. and Fell.7. Pells. and Fell. UNICIV Report No. ASCE. Wan. No. C. The University of New South Wales. Laboratory tests on the rate of piping erosion of soils in embankment dams. (2002). School of Civil and Environmental Engineering. Geotech.References SECTION 14 Maniam. Silva and E. and Fell. Sydney. J. ISBN 85841 396 5. Wan. Experimental investigation of internal instability of soils in embankment dams and their foundations. (1994). Herbaum and Schuler (editors). (1993).27.3. Edited by F. Damage and Cracking of Embankment Dams by Earthquake and the Implications for Internal Erosion and Piping. Sydney.429. UNICIV Report No. C. Wan. School of Civil and Environmental Engineering. R-406. Investigation of rate of erosion of soils in embankment dams. No. M. Schmertmann. The University of New South Wales. (2002). and Fell. C. UNICIV Report No. Pells.W. School of Civil and Environmental Engineering. (2004b). vol. J. R. Skempton A. Vol. ISBN 85841 396 5. and the Implications for Internal Erosion and Piping. ICOLD. ASCE. (1989). Proceedings 21st Internal Congress on Large Dams.

4 Section 7.2 Section 6.5 Section 5.1.4 Section 6. Figure 5. Figure 5.2.3 Section 7.3 Section 6.2 Section 12.2 Section 11.6 Section 5.3 Section 6.2 Section 11.6 Section 4.6.5 Section 7.1 Section 7.1 Section 3.2. Figure 5.6.2 Section 5.3.4.4 Section 6.2 Section 9 Section 12.4.1 Section 6. Figure 5.4 Section 5.2.1 Section 6.1.6 α β AEP CE CF CLSM CPT EE POR G H Hw iav iavf icr IHET ipmt LF NE OWC Pc PCE Pcl Pdetect Pe Pfe PH Phf A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 15-1 .2 Section 6.2.1 Section 4.3 Section 5.2 Section 7.3.List of Acronyms & Symbols SECTION 15 15 List of Acronyms & Symbols List of Acronyms & Symbols used in this Report Acronym/Symbol Meaning Slope of sides of compressible zone abutment slope Annual Exceedance Probability Continuing Erosion Amount of collapse compression as a proportion of the layer thickness Controlled low strength material Cone Penetration Test Excessive Erosion Pool of record Height of the gap/crack formed by collapse compression Height of dam Height of wall/cliff Average seepage gradient Average seepage gradient in the foundation Critical seepage gradient Erosion Rate Index Average gradient required to initiate backward erosion Likelihood factor No Erosion boundary Optimum Water Content Probability of a transverse crack Probability of continuing erosion Probability of continuous layer of cohesionless soil Probability of detection Probability of initiation of backward erosion or suffusion at the core-foundation contact Probability of filtered exit Probability of heave Probability of hydraulic fracture across the cutoff Reference Section 5.6 Section 11.

4 Section 5.2 Section 5.6. Figure 5.1.4 Section 11.2. Figure 5.1 Section 6.2.6 Section 5.3 Section 5.4 Section 7.6 Section 12.2 Section 4.2.3 Section 11.2 Section 7.1.List of Acronyms & Symbols SECTION 15 trench PI PIC Pic PIH PINH Pintervene PIP PIUS Ppath Punf RF SE SPT(N1)60 SMDD TP W/D W/H Wb Wv Probability of initiation of erosion Probability of initiation of erosion in a crack Probability of scour erosion at the core-foundation contact Probability of initiation and progression of backward erosion given heave Probability of initiation and progression of backward erosion given no heave Probability of intervention Probability of initiation and progression of backward erosion Probability soil is internally unstable Probability of a continuous pathway of open joints in rock or coarse grained soil Probability of unfiltered exit Relative importance factor Some Erosion Standard Penetration Test N value.1 Section 5.4 Section 12.3 Section 9 Section 6.2 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 15-2 . Figure 5. corrected to 60% energy Standard Maximum Dry Density Thickness of layer of poorly compacted soil Width to depth ratio of the cut-off trench Width to height ratio of the core Width of bench at base of wall/cliff Width of base of valley Section 6.1 Section 12.6 Section 6.3.

Appendix A Navigation Table for Internal Erosion Through the Embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 .

PCE Evaluate the probabilities for No.Table A1. (5) Probability of Breach Pbreach Initiating Mechanism Sketch (1) Failure Path Identification and Screening Use Table A2 to identify and screen potential crack mechanisms (6) Calculate the Probability of Failure Initiation of Erosion in Transverse Cracks in Upper Part of the Embankment Crack in Upper Part Cross Section Upper 1/3rd Estimate the probabilities for forming a roof (PPR). PSE. Excessive and Continuing Erosion for the failure path under consideration using Table A7. Probability of Failure by Internal Erosion Through the Embankment (Sheet 1) (2) Evaluate the Probability of Initiation of Erosion PI Use Table A2 to evaluate the probability of initiation for each crack initiating mechanism. Pbreach-NE = 0. Pfail = PI (IMx) x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes PI (IMx) Yes Transverse Cracks in Upper Part of Embankment PP Intervention Fails No Some Erosion Continuation No Erosion PSE Progression No Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table A8. Excessive and Continuing Erosion PNE. PSE PEE PCE (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi Estimate the probability for not detect and intervene using Table A9. Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure for each IM using the event tree. Some. Excessive and Continuing Erosion branches using Table A10.Guidance Document July 2008 Delta Version A-1 . PP = PPR x PPC x PPL Estimate the probabilities of breach for the Some. PI (IMx) (3) Probabilities for No. Some. PEE.

crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table A8. Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure for each IM using the event tree.Table A1. Excessive and Continuing Erosion for the failure path under consideration using Table A7. Pbreach-NE = 0. PP = PPR x PPC x PPL Estimate the probabilities of breach for the Some. Pfail = PI (IMx) x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes Some Erosion Yes Transverse Cracks in Middle and Lower Part of Embankment PP Intervention Fails No PSE Progression No PI (IMx) Continuation No Erosion Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Some. PI (IMx) (3) Probabilities for No. Excessive and Continuing Erosion PNE. Some.Guidance Document July 2008 Delta Version A-2 . PSE PEE PCE (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi Estimate the probability for not detect and intervene using Table A9. Excessive and Continuing Erosion branches using Table A10. PCE Evaluate the probabilities for No. (5) Probability of Breach Pbreach (6) Calculate the Probability of Failure Initiation of Erosion in Transverse Cracks in the Middle and Lower Parts of the Embankment Crack in Middle and Lower Part Cross Section Lower 2/3rd Use Table A3 to identify and screen potential crack mechanisms Estimate the probabilities for forming a roof (PPR). PSE. Probability of Failure by Internal Erosion Through the Embankment (Sheet 2) Initiating Mechanism Sketch (1) Failure Path Identification and Screening (2) Evaluate the Probability of Initiation of Erosion PI Use Table A3 to evaluate the probability of initiation for each crack initiating mechanism. PEE.

Some. PCE Evaluate the probabilities for No. Excessive and Continuing Erosion PNE. Probability of Failure by Internal Erosion Through the Embankment (Sheet 3) Initiating Mechanism Sketch (1) Failure Path Identification and Screening (2) Evaluate the Probability of Initiation of Erosion PI Use Table A4 to evaluate the probability of initiation for each initiating mechanism.Guidance Document July 2008 Delta Version A-3 . Pfail = PI (IMx) x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes Some Erosion Yes Poorly Compacted or High Permeability Zones in the Embankment PP Intervention Fails No PSE Progression No PI (IMx) Continuation No Erosion Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure for each IM using the event tree. PEE. PSE PEE PCE (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi Estimate the probability for not detect and intervene using Table A9. PP = PPR x PPC x PPL Estimate the probabilities of breach for the Some. PSE. Excessive and Continuing Erosion for the failure path under consideration using Table A7. Some.Table A1. PI (IMx) (3) Probabilities for No. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table A8. Excessive and Continuing Erosion branches using Table A10. (5) Probability of Breach Pbreach (6) Calculate the Probability of Failure Initiation of Erosion in Poorly Compacted or High Permeability Zones in the Embankment Poorly Compacted or High Permeability Zone Cross Section Use Table A4 to identify and screen potential crack mechanisms Estimate the probabilities for forming a roof (PPR). Pbreach-NE = 0.

Probability of Failure by Internal Erosion Through the Embankment (Sheet 4) Initiating Mechanism Sketch (1) Failure Path Identification and Screening (2) Evaluate the Probability of Initiation of Erosion PI Use Table A5 to evaluate the probability of initiation for each initiating mechanism. Excessive and Continuing Erosion PNE. (5) Probability of Breach Pbreach (6) Calculate the Probability of Failure Initiation of Erosion in Poorly Compacted or High Permeability Zones Along a Conduit Cross Section Conduit Use Table A5 to identify and screen potential crack mechanisms Estimate the probabilities for forming a roof (PPR). Excessive and Continuing Erosion branches using Table A10. Pbreach-NE = 0. Some. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table A8. PCE Evaluate the probabilities for No. PSE. Excessive and Continuing Erosion for the failure path under consideration using Table A7.Guidance Document July 2008 Delta Version A-4 . PI (IMx) (3) Probabilities for No. PSE PEE PCE (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi Estimate the probability for not detect and intervene using Table A9. Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure for each IM using the event tree. PP = PPR x PPC x PPL Estimate the probabilities of breach for the Some.Table A1. Pfail = PI (IMx) x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes Some Erosion Yes Poorly Compacted or High Permeability Zones Along a Conduit PP Intervention Fails No PSE Progression No PI (IMx) Continuation No Erosion Σ Pfail Initiation Along Conduit No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . PEE. Some.

Pbreach-NE = 0. PI (IMx) (3) Probability for Continuing Erosion PCE Evaluate the probability for Continuing Erosion for erosion into the conduit using Table A7 (Scenario 4).0 Pudi Estimate the probabilities of breach for the Continuing Erosion branch using Table A10 (sinkhole/crest settlement breach mechanism only). Sheet 4) Yes Continuing Erosion Continuation No Erosion PCE Develops into IE Along Conduit No Yes Yes PP Pudi Breach No Pbreach-CE Pfail-CE Yes Erosion into a Conduit PI (IMx) Intervention Fails No Initiation Into Conduit No Progression No Σ Pfail A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . PP = 1. PALONG (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi Estimate the probability for not detect and intervene using Table A9.16 with weighted score from Table 6.11. PCE (3) Probability for Erosion into the Conduit to Develop into Internal Erosion Along the Conduit P Evaluate the probability for erosion into the conduit to develop into internal erosion along the conduit using Table 6.Guidance Document July 2008 Delta Version A-5 . (5) Probability of Breach Pbreach (6) Calculate the Probability of Failure Initiation of Erosion into a Conduit Leading to Sinkhole or Crest Settlement Not applicable to this failure path. Pbreach-CE Calculate the probability of failure for each IM using the event tree.Table A1. Probability of Failure by Internal Erosion Through the Embankment (Sheet 5) Initiating Mechanism (1) Failure Path Identification and Screening Use Table A5 to identify and screen potential crack mechanisms (2) Evaluate the Probability of Initiation of Erosion PI Use Table A5 to evaluate the probability of initiation into the conduit. Pfail = PI (IMx) x PP x Pudi x (PCE x Pbreach-CE)] EVENT TREE STRUCTURE Yes PALONG This branch leads to Internal Erosion Along the Conduit (refer Table A1.

PP = PPR x PPC x PPL Estimate the probabilities of breach for the Some. PI (IMx) (3) Probabilities for No. Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure for each IM using the event tree. PEE. Excessive and Continuing Erosion for the failure path under consideration using Table A7. Some. Excessive and Continuing Erosion branches using Table A10.Table A1. PCE Evaluate the probabilities for No. Pfail = PI (IMx) x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes PI (IMx) Yes Poorly Compacted or High Permeability Zones Adjacent to a Spillway or Abutment Wall PP Intervention Fails No Some Erosion Continuation No Erosion PSE Progression No Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . PSE. Excessive and Continuing Erosion PNE. PSE PEE PCE (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi Estimate the probability for not detect and intervene using Table A9. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table A8. Some. Probability of Failure by Internal Erosion Through the Embankment (Sheet 6) Initiating Mechanism Sketch (1) Failure Path Identification and Screening (2) Evaluate the Probability of Initiation of Erosion PI Use Table A6 to evaluate the probability of initiation for each initiating mechanism. (5) Probability of Breach Pbreach (6) Calculate the Probability of Failure Initiation of Erosion in Poorly Compacted or High Permeability Zones Adjacent to a Spillway or Abutment Wall Use Table A6 to identify and screen potential crack mechanisms Estimate the probabilities for forming a roof (PPR).Guidance Document July 2008 Delta Version A-6 . Pbreach-NE = 0.

Assess weighted score from Table 5.4 IM5 – Differential settlements in the foundation beneath the core Long Section Check if applicable using Table 3.23.1.25 and width for hydraulic fracture from Table 5.29 to 5.5 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Use the larger of the two values WC1 (6) Estimate probability of erosion (PIC) IM1 – Cross Valley Differential Settlement Crack No Exclusions Apply – Always Include Obtain from Row 1 of Table 5.4 = PC4 x PIC.5 and obtain probability from Table 5. PC4 = PC4 (un-factored) x MOU. PC2 (un-factored) Multiplication factor same as for FM1 (MOU.1).3 Crack Long Section IM4 – Cross section Settlement due to Poorly Compacted Shoulders Cross Section Cracking Check if applicable using Table 3.26(A).24.6. PC4 (un-factored) Multiplication factor same as for FM1 (MOU.4.1 (4) Estimate maximum likely crack width at the dam crest (5) Estimate likely crack width at reservoir level stage being considered (WC) Obtain width for cracking from Table 5.1.29 to 5. WC3 PI.1. using weighted score from Step (2) Use crack width from Step (4).2.29 to 5. PC5 = PC5 (un-factored) x MOU.5 PI.24. PC1 (un-factored) (3) Multiplication Factor for Observations and Measured Settlement (MOU) and calculate PCU Obtain multiplication factors from Table 5.9 and obtain probability from Table 5.22 and Table 5.Guidance Document July 2008 Delta Version A-7 .4 PI. PIC.Table A2. WC5 Obtain probability of erosion from the most appropriate table from Tables 5.3 PI. using weighted score from Step (2) Obtain probability of erosion from the most appropriate table from Tables 5.1 Obtain from Row 2 of Table 5.35 based on the core soil type and using the crack width (WC1) determined in Step (5). Select the maximum factor (MOU.1 Obtain probability of erosion from the most appropriate table from Tables 5.1 Obtain from Table 5.24.26. Assess weighted score from Table 5.5 = PC5 x PIC.8. using weighted score from Step (2) Obtain width of crack from Table 5.1. PIC. PC3 = PC3 (un-factored) x MOU.25 and width of hydraulic fracture from Table 5.1 and obtain probability from Table 5.1 Obtain from Row 4 of Table 5.3 = PC3 x PIC.1). WC4 Settlement of shoulders Obtain probability of erosion from the most appropriate table from Tables 5.26(B). Assess weighted score from Table 5. Probability of Initiation of Erosion in Transverse Cracks – Upper Parts of the Embankment Probability of Transverse Cracking in Upper Part of Embankment Probability of Erosion in the Crack (7) Calculate the Probability Initiation of Erosion (PI) Initiating Mechanism (1) Screening of Failure Path (2) Assess the weighted score (WS) and estimate Probability of a Crack PC (un-factored) Assess weighted score from Table 5. Assess weighted score from Table 5.1 Obtain from Row 3 of Table 5. PC5 (un-factored) Multiplication factor same as for FM1 (MOU. using weighted score from Step (2) Obtain width of crack from Table 5.35 based on the core soil type and using the crack width (WC3) determined in Step (5) PIC. Use the larger of the two values.3 and obtain probability from Table 5.29 to 5.24.35 based on the core soil type and using the crack width (WC5) determined in Step (5).1).7 obtain probability from Table 5.1 = PC1 x PIC. PC3 (un-factored) Multiplication factor same as for FM1 (MOU.1) PC2 = PC2 (un-factored) x MOU.2 Long Section IM3 – Cross Valley Arching Check if applicable using Table 3.2 Obtain probability of erosion from the most appropriate table from Tables 5.10.25.29 to 5.35 based on the core soil type and using the crack width (WC4) determined in Step (5).2 = PC2 x PIC.25.35 based on the core soil type and using the crack width (WC2) determined in Step (5) PIC. PC1 = PC1 (un-factored) x MOU.1). WC2 PI.1 Long Section IM2 – Differential Settlement Adjacent a Cliff Crack/Gap Check if applicable using Table 3. PIC. using weighted score from Step (2) Obtain width of crack from Table 5.

25.1.0). PC6 = PC6 (un-factored) x MOU.8 IM13 – Cracking due to earthquake loading.24. Long Section Check if applicable using Table 3.29 to 5. Obtain from Row 1 of Table 5.29 to 5.7 = PC7 x PIC.39. PC13 = PC13 (un-factored) Obtain from Table 5. PC6 (un-factored) (3) Multiplication Factor for Observations and Measured Settlement (MOU) and calculate PCU Obtain multiplication factor from Table 5.35 based on the core soil type and using the crack width (WC6) determined in Step (5). Assess weighted score from Table 5.23 (MOU.1 and obtain probability from Table 5.1.13 = 1.6 Stage 2 Stage 1 IM7 – Cracking in the crest due to desiccation by drying.6).8 = PC8 x PIC. Assess damage class from Figure 5.13 PI.12. using weighted score from Step (2) Use crack width from Step (4).7 Obtain from Row 5 of Table 5.14 and obtain probability from Table 5. PIC.6 PI. PIC. using weighted score from Step (2) Crack Obtain probability of erosion from the most appropriate table from Tables 5.29 to 5.29 to 5.13 Long Section A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . PC8 (un-factored) Multiplication factor not applicable (MOU.35 based on the core soil type and using the crack width (WC7) determined in Step (5). using weighted score from Step (2) Obtain width of crack from Table 5.11 and obtain probability from Table 5.0).24. WC6 (6) Estimate probability of erosion (PIC) IM6 – Cracking due to differential settlements due to embankment staging.1.7).25.1.25.7 IM8 – Cracking on seasonal shutdown layers during construction and staged construction surfaces due to desiccation by drying Long Section Crack Stage 2 Stage 1 Check if applicable using Table 3. WC8 Obtain probability of erosion from the most appropriate table from Tables 5.Probability of Transverse Cracking in Upper Part of Embankment Probability of Erosion in the Crack (7) Calculate the Probability Initiation of Erosion (PI) Initiating Mechanism (1) Screening of Failure Path (2) Assess the weighted score (WS) and estimate Probability of a Crack PC (un-factored) Assess weighted score from Table 5.Guidance Document July 2008 Delta Version A-8 . Assess weighted score from Table 5. Crack Check if applicable using Table 3.2.15.35 based on the core soil type and using the crack width (WC13) determined in Step (5). PC8 = PC8 (un-factored) Obtain from Row 6 of Table 5. PC6 = PC6 (un-factored) x MOU.8 PI. Obtain probability from Table 5.6 (4) Estimate maximum likely crack width at the dam crest (5) Estimate likely crack width at reservoir level stage being considered (WC) Obtain width of crack from Table 5.6 = PC6 x PIC. PC7 (un-factored) Obtain multiplication factor from Table 5. WC7 Obtain probability of erosion from the most appropriate table from Tables 5.2 (refer to Section 5.13 = PC13 x PIC.4 on how to apply). using damage class from Step (2) Obtain width of crack from Table 5.39.8 = 1.24.8 for earthfill dams or Figure 5.35 based on the core soil type and using the crack width (WC7) determined in Step (5). WC13 Obtain probability of erosion from the most appropriate table from Tables 5.6 PI. PC13 (un-factored) Multiplication factor not applicable (MOU.23 (MOU. PIC. Long Section Check if applicable using Table 3. PIC.9 for earth and rockfill dams.

28.35 based on the core soil type and using the crack width (WC12) determined in Step (5). PC12 = PC12 (un-factored) Obtain from Row 3 of Table 5.29 to 5.12 = PC12 x PIC. WC9 (6) Estimate probability of erosion (PIC) IM9 – Cross Valley Settlement Crack Check if applicable using Table 3.17.29 to 5.0). Assess weighted score from Table 5.35 based on the core soil type and using the crack width (WC9) determined in Step (5).21.10 PI. using weighted score from Step (2) Use crack width from Step (4).29 to 5.2. Obtain probability of erosion from the most appropriate table from Tables 5.12 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . (4) Estimate maximum likely crack width (includes hydraulic fracture) Obtain from Row 1 of Table 5.28.35 based on the core soil type and using the crack width (WC10) determined in Step (5) PIC.16 and obtain probability from Table 5.19.2.9 Long Section IM10 – Differential settlement causing arching of the core onto the shoulders of the embankment.18 and obtain probability from Table 5.Table A3. PIC.). PC12 (un-factored) Multiplication factor not applicable (MOL = 1.9 = PC9 x PIC. WC12 Obtain probability of erosion from the most appropriate table from Tables 5.12 IM12 – Small scale irregularities in the foundation profiles under the core Long Section PI.Guidance Document July 2008 Delta Version A-9 . Check if applicable using Table 3.9 PI. PC10 (un-factored) Multiplication factor same as for FM9 (MOL) PC10 = PC10 (un-factored) x MOL Obtain from Row 2 of Table 5. PC9 = PC9 (un-factored) x MOL. using weighted score from Step (2) Use crack width from Step (4).22 (MOL.28. using weighted score from Step (2) (5) Estimate likely crack width at reservoir level stage being considered (WC) Use crack width from Step (4). PIC. Assess weighted score from Table 5. PC9 (un-factored) (3) Multiplication Factor for Observations and Measured Settlement (MOL) and calculate PC Obtain multiplication factor from Table 5. Probability of Initiation of Erosion in Transverse Cracks – Middle and Lower Parts of the Embankment Probability of Transverse Cracking in Middle and Lower Parts of the Embankment Probability of Erosion in the Crack (7) Calculate the Probability Initiation of Erosion (PI) Initiating Mechanism (1) Screening of Failure Path (2) Assess the weighted score (WS) and estimate Probability of a Crack PC (un-factored) Assess weighted score from Table 5.10 IM11 – Differential settlement in the foundation under the core Already considered in IM5 Check if applicable using Table 3. WC10 Obtain probability of erosion from the most appropriate table from Tables 5.10 = PC10 x PIC.20 obtain probability from Table 5.2.

use Table 6. PP14 = PP14 (unfactored) x MOS.24 (MOS. Obtain probability of erosion from Table 6. Assess weighted score from Table 6.3.14 Cohesive soils: PI.14 IM15 – Poorly compacted or high permeability layer on the corefoundation contact Long Section Poorly compacted layer Check if applicable using Table 3. Estimate the height of the gap G14 = TP. PP15 (un-factored) Obtain multiplication factor from Table 6. Not applicable – consider as for cohesive soils Not applicable – consider as for cohesive soils Not applicable .). PIP.35 based on the core soil type and using the gap height (G16) determined in Step (7).3.1 for cohesive soils or Table 6. For reservoir levels above the normal operating pool level.10.7.refer to Steps (7) and (8). PP16 (un-factored) Obtain multiplication factor from Table 6.17 Stage 2 Stage 1 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Check depth of freezing induced flaws using Table 6.3. estimate the critical gradient (iCR) (6) Estimate probability of erosion (PIP) IM14 – Poorly compacted or high high permeability layer in the embankment Cross Section High Permeability Zone Check if applicable using Table 3.14 = PP14 x PIC.4 – see Section 6.25 to estimate time for seepage gradient to develop in layer. If Cu>6.).29 to 5. PIC.6.16 IM17 – Seasonal shutdown layers during construction and staged construction surfaces due to freezing.29 to 5.3.30 (G17) Cohesionless soils: PI. PP14 (un-factored) (3) Multiplication Factor for Observations of Seepage (MOS) and calculate PP Obtain multiplication factor from Table 6.8.16 Obtain probability of erosion from the most appropriate table from Tables 5. PIC. Long Section Check if applicable using Table 3.2 for cohesionless soils.14 Obtain probability of erosion from Table 6.26 for well compacted layers and Table 6.15 Cohesive soils: PI.35 based on the core soil type and using the gap height (G15) determined in Step (7). Exclude if the reservoir level rise is insufficient for seepage gradient to develop.15 = PP15 x PIC.17 Cohesive soils: PI. PP17 = PP17 (unfactored) x MOS.26 for well compacted layers and Table 6.15 Not applicable .27 for poorly compacted layers based on the average seepage gradient across the embankment core (iave) and iPMT (or iCR). Repeat the above step for each of the postulated high permeability layer types Same as IM14 Repeat the above step for each of the postulated high permeability layer types Obtain probability of erosion from the most appropriate table from Tables 5. PP15 = PP15 (unfactored) x MOS.29 to 5.Table A4.14 = PP14 x PIP.24 (MOS. PIP.17 Cohesionless soils: PI.24 (MOS.4 and obtain probability from Table 6.17= PP17x PIC. Not applicable – consider as for cohesive soils Not applicable – consider as for cohesive soils Assess the width of the of the frost induced flaw using Table 6. Assess the thickness of the poorly compacted layer (TP. PIC.).35 based on the core soil type and using the gap height (G14) determined in Step (7).Guidance Document July 2008 Delta Version A-10 .6 and obtain probability from Table 6.15 IM16 – Cracking in the crest due to desiccation by freezing Long Section Check if applicable using Table 3. (4) Assess whether there is time for seepage gradient to develop (5) Estimate average seepage gradient required to initiate and progress backward erosion (iPMT or iCR) Estimate the average gradient (iPMT)corrected required to initiate and progress erosion from Figure 6.16 = PP16 x PIP. PIC.3.17 = PP17 x PIP. Assess the width of the of the frost induced flaw using Table 6.15 Cohesionless soils: PI.9 and obtain probability from Table 6.5. Probability of Initiation of Erosion in a Poorly Compacted or High Permeability Zones in the Embankment Probability of High Permeability Zone in the Embankment Probability of Erosion in the High Permeability Zone (cohesionless soils) Probability of Erosion in the High Permeability Zone (cohesive soils) (7) Estimate the height of the gap due to collapse settlement (8) Estimate probability of erosion (PIC) (7) Calculate the Probability Initiation of Erosion (PI) Initiating Mechanism (1) Screening of Failure Path (2) Assess the weighted score (WS) and estimate Probability of a High Permeability Zone PP (un-factored) Assess weighted score from Table 6.29.29 to 5.14 x CF Obtain probability of erosion from the most appropriate table from Tables 5.2. PP16 = PP16 (unfactored) x MOS.30 (G16) Obtain probability of erosion from the most appropriate table from Tables 5.24 (MOS. Assess weighted score from Table 6.35 based on the core soil type and using the gap height (G17) determined in Step (7).refer to Steps (7) and (8).16 Cohesive soils: PI. PP17 (un-factored) Obtain multiplication factor from Table 6.14 Cohesionless soils: PI.16 = PP16 x PIC.).27 for poorly compacted layers based on the average seepage gradient across the embankment core (iave) and iPMT (or iCR). Assess weighted score from Table 6. Obtain the probability from Table 6. Estimate the amount by which the layer may collapse (CF) from Table 6.15 = PP15 x PIP.14).

PIC.27 for poorly compacted layers based on the average seepage gradient across the embankment core (iave) and iPMT (or iCR). assess the weighted score from Table 6. Same as FM18 This step is not applicable to this situation.11 and obtain probability from Table 6.26 for well compacted layers and Table 6. This step is not applicable to this situation.3.15.18 Cohesive soils: PI.18 = PP18 x PIC. This step is not applicable to this situation.13.3.19 = PP19 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . estimate the critical gradient (iCR) – see Section 6.24 (MOS. (6) Estimate probability of erosion (PIP) IM18 – Poorly compacted or high permeability layer around a conduit through the embankment Long Section High Permeability Zone Check if applicable using Table 3.18 Cohesionless soils: PI.). Obtain probability of erosion from Table 6. PI.29 to 5.18 IM19 – Erosion into a (nonpressurised) conduit Check if applicable using Table 3. When the internal condition of the conduit is known. Probability of Initiation of Erosion in a Poorly Compacted or High Permeability Zone Around or into a Conduit Probability of High Permeability Zone in the Embankment Probability of Erosion in the High Permeability Zone (cohesionless soils) Probability of Erosion in the High Permeability Zone (cohesive soils) (7) Estimate the height of the gap due to collapse settlement (8) Estimate probability of erosion (PIC) (7) Calculate the Probability Initiation of Erosion (PI) Initiating Mechanism (1) Screening of Failure Path (2) Assess the weighted score (WS) and estimate Probability of a High Permeability Zone PP (un-factored) Assess weighted score from Table 6.2. Estimate the amount by which the layer may collapse (CF) from Table 6.18). PP19 = PP19 (unfactored).35 based on the core soil type and using the gap height (G14) determined in Step (7).Table A5. This step is not applicable to this situation. obtain probability from Table 6. (4) Assess whether there is time for seepage gradient to develop (5) Estimate average seepage gradient required to initiate and progress backward erosion (iPMT or iCR) Estimate the average gradient (iPMT) required to initiate and progress erosion from Figure 6. PP18 (un-factored) (3) Multiplication Factor for Observations of Seepage (MOS) and calculate PP Obtain multiplication factor from Table 6.18 x CF Obtain probability of erosion from the most appropriate table from Tables 5.12.18 This step is not applicable to this situation.29. Estimate the height of the gap G18 = TP. PP19 (un-factored) Not applicable. When the internal condition of the conduit is not known. PIP.14 and obtain probability from Table 6. PP18 = PP18 (unfactored) x MOS.6. If Cu>6.Guidance Document July 2008 Delta Version A-11 .4. Assess the thickness of the poorly compacted layer (TP.18 = PP18 x PIP.

20 = PP20 x PIP.18. Assess weighted score from Table 6.1).3.26(B).27 for poorly compacted layers based on the average seepage gradient across the embankment core (iave) and iPMT (or iCR).1 Not applicable to this failure path Not applicable to this failure path Refer to Steps (7) and (8).18).3.25 to estimate time for seepage gradient to develop in layer. Select the maximum factor (MOU.29 to 5. PP20 = PP20 (unfactored) x MOS. Assess weighted score from Table 6.29. PC21 = PC21 (unfactored) x MOU. PIC.35 based on the core soil type and using the crack width (WC21) determined in Step (7). Estimate the amount by which the layer may collapse (CF) from Table 6.26 for well compacted layers and Table 6. PIP.20 Assess the thickness of the poorly compacted layer (TP.1). PIC.19 and obtain probability from Table 6.35 based on the core soil type and using the gap height (G20) determined in Step (7).29 to 5. PIC.25 and width for hydraulic fracture from Table 5. PP20 (un-factored) (3) Multiplication Factor for Observations of Seepage (MOS) and calculate PP Obtain multiplication factor from Table 6.22. PC21 (un-factored) Obtain multiplication factors from Table 5.20. Obtain width of crack from Table 5.20 Cohesive soils: PI.35 based on the core soil type and using the crack width (WC22) determined in Step (7).20 IM21 – Crack/gap adjacent to a spillway or abutment wall Check if applicable using Table 3. Estimate the height of the gap G20 = TP.20 Cohesionless soils: PI.22 = PC22 x PIC.22 PI. Exclude if the reservoir level rise is insufficient for seepage gradient to develop.21 IM22 – Differential settlement adjacent to a spillway or abutment wall Crack/Gap Check if applicable using Table 3. Probability of Initiation of Erosion in a Poorly Compacted or High Permeability Zone Adjacent to a Spillway or Abutment Wall Probability of High Permeability Zone in the Embankment Probability of Erosion in the High Permeability Zone (cohesionless soils) Probability of Erosion in the High Permeability Zone (cohesive soils) (7) Estimate the crack width or height of the gap due to collapse settlement (8) Estimate probability of erosion (PIC) (7) Calculate the Probability Initiation of Erosion (PI) Initiating Mechanism (1) Screening of Failure Path (2) Assess the weighted score (WS) and estimate Probability of a High Permeability Zone PP (un-factored) Assess weighted score from Table 6.29 to 5. PC22 (un-factored) Obtain multiplication factors from Table 5. (4) Assess whether there is time for seepage gradient to develop (5) Estimate average seepage gradient required to initiate and progress backward erosion (iPMT or iCR) Estimate the average gradient (iPMT) required to initiate and progress erosion from Figure 6. Not applicable to this failure path Obtain probability of erosion from Table 6. Select the maximum factor (MOU.26(B). Obtain width of crack at the level of interest from Table 5.20 = PP20 x PIC. WC21 Obtain probability of erosion from the most appropriate table from Tables 5. estimate the critical gradient (iCR).20 x CF Obtain probability of erosion from the most appropriate table from Tables 5. (6) Estimate probability of erosion (PIP) IM20 – Poorly compacted or high permeability zone associated with a spillway or abutment wall Check if applicable using Table 3. using weighted score from Step (2).3.21 and obtain probability from Table 6. Obtain max likely width of crack at dam crest from Row 2 of Table 5. For reservoir levels above the normal operating pool level.4. PC22 = PC22 (unfactored) x MOU.).22 Long Section A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .16 and Table 5.21 = PC21 x PIC. use Table 6.24 (MOS.Guidance Document July 2008 Delta Version A-12 . If Cu>6.Table A6.17 and obtain probability from Table 6.16 and Table 5.21 PI.24.17.1 Not applicable to this failure path Refer to Steps (7) and (8). Use the larger of the two values WC22 Obtain probability of erosion from the most appropriate table from Tables 5.17.

INTERNAL EROSION THROUGH THE EMBANKMENT INTO A TOE DRAIN INTERNAL EROSION THROUGH THE FOUNDATION INTO A TOE DRAIN A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . PCE. SOIL SOIL CONDUIT Step 1: Evaluate the joint/defect opening size that would allow Continuing Erosion (JOSCE) using Table 10. Step 2: Estimate the conditional probability for Continuing Erosion (PCE) by estimating the proportion of soils that are coarser than JOSCE and using Table 10. ZONED EARTHFILL WITH COHESIVE SHELLS INTERNAL EROSION ABOVE FILTER ZONE. Use Table 10.3 for more details. joint or crack. COHESIVE DOWNSTREAM SHELL Calculate PNE = 1.0 – PCE PSE = PEE = 0 Scenario 3: Filter/transition zone is present downstream of the core or a downstream shoulder zone which is not capable of holding a crack/pipe.12.10. PCE) No potential for filtering.1 to estimate the probability for continuing erosion. Probability of Continuation for Internal Erosion in the Embankment Assess which Scenario is Applicable to the Failure Path under Consideration Scenario 1: Homogeneous zoning with no fully intercepting filter 1 HOMOGENEOUS EARTHFILL Examples Probability of Continuation (PNE . PSE .1. ZONED EARTHFILL WITH CHIMNEY FILTER 3 1 3 ZONED EARTHFILL WITH GRANULAR SHELLS Scenario 4: Piping into an open defect.0 PNE = PSE = PEE = 0 EARTHFILL WITH TOE DRAIN. See Section 7. Excessive Erosion (PEE) and Continuing Erosion (PCE). 1A 1 1A 1A 1A 1 2 Filtering does not occur if the crack/high permeability zone persists through the downstream shoulder zone.Table A7.Guidance Document July 2008 Delta Version A-13 .11. 1 Probability for continuing erosion. 2 1 1 3 1 EARTHFILL WITH CHIMNEY FILTER 3 2 Follow the procedure outlined in Section 10.4 to estimate the probabilities of No Erosion (PNE). Figure A1 shows a flow chart which summarizes the procedure. Some Erosion (PSE). OPEN JOINTED ROCK EROSION INTO AN OPEN CRACK OR JOINT IN A CONDUIT OR WALL EROSION INTO OPEN JOINTS IN ROCK FOUNDATION Scenario 5: Erosion into a toe drain 1 1 Estimate the probability of continuing erosion for erosion into a toe drain using Table 10. The assessment of erosion into a toe drain considers the observed condition of the toe drain (from video or external inspections) and the design and construction details of the toe drain. PEE .1. INTERNAL EROSION ABOVE TOE DRAIN 2 3 1 3 INTERNAL EROSION ABOVE CORE AND FILTERS Scenario 2: Downstream shoulder of fine grained cohesive material which is capable of holding a crack/pipe. PCE=1.

FIGURE A1 – FLOW CHART FOR EVALUATING THE PROBABILITIES FOR NO.1 to 0. Yes (PFC ≥ 0. EE and CE for each representative base soil grading (refer to Step 8 in Section 10. Plot the particle size distributions for the base material.1) Evaluate the remaining steps by considering the ‘cracked’ filter/transition zone as the base soil. If FSb = 0. and calculate PNE = 1. Estimate the proportion of filter gradings that fall into the No Erosion. Some Erosion. SOME. Use Tables 10.(PSE + PEE + PCE) 2. then assign a probability of continuing erosion PCE between 0. Yes No Is the filter/transition material susceptible to suffusion? Evaluate the probability the filter or transition zone materials are internally unstable (PIUS) using Figure 6.1.e. Make an initial estimate of the probabilities for the No Erosion. then assume the filter/transition materials are susceptible to suffusion.5? No (FSB ≥ 0. and Continuing Erosion boundaries for the fine limit. then regrade the base grading on the particle size that is missing. Some Erosion.75 mm. and the zone downstream of the cracked filter as the filter material.1 and 0. average and coarse limit of the regraded core materials. Is the filter/transition material susceptible to segregation? Evaluate weighted score using Tables 10.4.5. If there are no gradations that fall into the Continuing Erosion category. Some. If the base soil is gap graded. The suggested approach is to estimate the proportions for Some. Plot the DF15 values for these boundaries on the grading curve limits of the filter/transition material.4a for the calculations). Estimate the probabilities for No. then regrade the base grading such that the maximum size is 4.75 mm. Excessive and Continuing Erosion OUTPUT. Yes (PIUS ≥ 0.3) Use the adjusted filter/transition gradings in the remaining steps. Yes (FSB < 0.5. If the maximum particle size of the base soil is >4. FSB = (total stress of material over the crack exit) / (reservoir head at the crack exit). No (PFC < 0.Guidance Document July 2008 Delta Version A-14 .6 for materials with <10% fines. average base and coarse base). Do this for each of the representative base soil gradings (fine base. and assess whether segregation needs to be considered using Table 10.5 for materials with >10% fines or Figure 6.8 to aid judgement in assigning the probability for Continuing Erosion (this allows for the possibility of gradations being coarser than indicated by the available data).0. for suffusion or segregation). 3. Use the adjusted grading curves for the filter/transition if required by the preceding steps (i. Is the FSB value < 0. Excessive Erosion and Continuing Erosion branches by calculating the sum-product of the % of base soil gradings and the % of NE. Assume the filter holds a crack if the probability of holding a crack PFC is ≥ 0.9. SE.7 to evaluate the DF15 values. then use Table 10. PSE PEE PCE A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . No (PIUS < 0. EXCESSIVE AND CONTINUING EROSION FOR FILTER/TRANSITION ZONES Will the filter/transition zone blow out due to a limited depth of cover? Calculate the factor of safety of blow out.5) If FSb < 0.3) Evaluate the DF15 values for the No Erosion. Excessive Erosion and Continuing Erosion categories. If the probability of internal instability (PIUS) is ≥ 0. Use judgement to adjust the calculated percentages to take into account the effects of other factors such as the distribution of the core and filter gradations in the fill.1. the average grading (average base) and the coarser 5% of the base soils (coarser base). 1. Excessive and Continuing Erosion (PSE. borrow area variability and selective placement of materials. Adjust the gradation of the filter/transition materials using the graphical method in Figure 10.3.3 and 10.1. Select representative gradings of the regraded base soil which are indicative of the finer 5% of the base soil gradings (fine base).1) Adjust base soil gradations. Some Erosion.5) Will the filter/transition material hold a crack? Evaluate the probability the filter/transition zone materials holding a crack PFC using Table 10. PEE and PCE).1 then probability of continuing erosion PCE = 1.6 and 10.2.

3.Guidance Document July 2008 Delta Version A-15 . PPR Not applicable to this failure path PPR = 1. PPL Not applicable to this failure path PPL = 1. Probability of Progression for Internal Erosion in the Embankment Applicable Initiating Mechanism All IE Embankment Failure Paths except IM19 (IM1 to IM18) IM19 (Erosion into a conduit causing crest settlement/sinkhole) Probability of Forming a Roof (PPR) Assess the probability of the core material to form a roof of a pipe using Table 11.Table A8.0 Probability that Crack Filling Action is Not Effective (PPC) Assess the probability of crack filling action not stopping pipe enlargement using Table 11. PPC Not applicable to this failure path PPC = 1.2.0 Probability that Upstream Zone Fails to Limit Flows (PPL) Assess the probability that the upstream zone or concrete elements fail to limit flows using Table 11.0 A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .1.

8.775 Example sub-event tree for calculating the probability of not intervening A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .225 Probability of Not Intervening = 0. 12.3.5 0.2 and 12.1 0.Table A9. that it is not detected Pnd Assess the probability of not detecting the leak using Table 12.6. Pui (6) Probability of Not Detect and Not Intervene Pudi (2) Assess the probability of the concentrated leak not being observable Pnol Assess the weighted score (WS) from Table 12. Pudi = Pnol + [(1-Pnol) x Pnd x Pui] (Refer to sub-event tree diagram below) Yes Not Intervene Tree 0.225 Intervention is Suc cessful 0.5 0.45 Yes 0.5 Interve ntion and Repair is Not Succ essful No 0.1 No Intervention Probability of Not Intervening 0.9 0.1.5 and obtain probability from Table 12. Pnd All Failure Paths/Initiating Mechanisms Estimate the approximate time for progression of piping and development of a breach using Tables 12.45 No Intervention Not Detected in Time 0.5 0.225 No Intervention No 0.7. Calculate the probability of unsuccessful detection and intervention using Pnol from Step (2).1 Internal Erosion is Not Able to be Detected Yes No 0. Pnd from Step (3) and Pui from Step (5).Guidance Document July 2008 Delta Version A-16 . Pnol (3) Assess the probability that given the leak is observable. = Pnol + [(1-Pnol) x Pnd] (5) Assess the probability that intervention and repair is not successful Pui Assess the probability that intervention and repair is not successful using Table 12. Probability of Unsuccessful Detection and Intervention for Internal Erosion in the Embankment Probability of Not Detecting Applicable Failure Path/Location (1) Assess the Rate of Internal Erosion and Piping Probability of Not Intervening (4) Calculate the probability of not detecting the internal erosion Pndi Calculate the probability of not detecting the internal erosion using Pnol from Step (2) and Pnd from Step (3).

16. Pge (a) Estimate the probability of slope instability occurring due to the increased seepage flows. (a) Estimate the probability of a sinkhole developing as a result of the internal erosion (Ps-f) from Table 13.12 and obtain the probability for each branch of the tree from Table 13.18. Excessive and Continuing Erosion branches) Assess which breach mechanisms are relevant to the type of dam zoning and failure path being considered using Table 13. This is done for each branch of the event tree [Some Erosion (Psi-i SE).Psd-CE)]. (c) Calculate the probability of breach by slope instability for each branch of the tree. Assess the weighted score from Table 13. (c) Calculate the probability of breach by sinkhole development.Psu-SE) x (1 . Excluded breach mechanisms are assigned a probability of breach = 0.2. Assess the weighted score from Table 13.10 and obtain the probability from Table 13.15 and obtain the probability for each branch of the tree from Table 13. Some. (c) For dams with a downstream zone of rockfill.Guidance Document July 2008 Delta Version A-17 . Calculate the probability of breach for each branch of the event tree (i.e.11. No Erosion Branch (NE) Some Erosion Branch (SE) Excessive Erosion Branch (EE) Continuing Erosion Branch (CE) E Pbreach = 0 for NE Pge-SE = 0 Pge-EE = 0 Pge-CE from Table 10.19 and obtain the probability for each branch of the event tree from Table 13. Some Erosion.e. estimate the probability of breach by gross enlargement using Table 13. (a) Probability seepage will emerge into the downstream shell PS = 1.3 (Psi-SE) = (Psi-i SE) x (Psi-lf). (b) Estimate the probability of loss of freeboard due to instability (Psi-lf).Psd-SE)].Pge-SE) x (1 Psi-SE) x (1 . If this breach mechanism is applicable. Excessive Erosion and Continuing Erosion branches) by summing the probabilities using de Morgan’s rule as follows. (b) Estimate the probability that the sinkhole causes loss of freeboard.Pge-EE) x (1 Psi-EE) x (1 . The probability of breach by gross enlargement Pge= 0 for the No. Excessive Erosion (Psu-EE) and Continuing Erosion (Psu-CE).0. Pbreach-EE= 1 – [(1 . (Psi-EE) = (Psi-i EE) x (Psi-lf). Breach by gross enlargement is negligible in cases where the downstream shell is unable to support a roof of a pipe. Excessive Erosion (Psi-i EE) and Continuing Erosion (Psi-i CE)]. Some Erosion (Psu-SE).Pge-CE) x (1 Psi-CE) x (1 . (Psu-SE) (Psu-EE) (Pun-CE) (Psd-SE) = (Ps-f) x (Ps-lf SE) (Psd-EE) = (Ps-f) x (Ps-lf EE) (Psd-CE) = (Ps-f) x (Ps-lf CE) Pbreach-NE = 0 Pbreach-SE= 1 – [(1 . assess the weighted score from Table 13.4 and obtain the probability from Table 13. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .Psd-EE)]. Assess the weighted score from Table 13. Some Erosion (Ps-lf SE). (b) For dams with a downstream zone of earthfill.20. No. Excessive Erosion (Psu-EE) and Continuing Erosion (Psu-CE). Some Erosion (Psu-SE). Pbreach-CE = 1 – [(1 .13. Evaluate whether this is applicable using Table 13.1. Some and Excessive Erosion branches.Table A10. (Psi-CE) = (Psi-i CE) x (Psi-lf).3. assess the weighted score from Table 13.Psu-EE) x (1 .5. Probability of Breach for Internal Erosion in the Embankment Applicable Failure Path/Location (1) Screening of breach mechanisms (2) Probability of Breach by Gross Enlargement Pge (3) Probability of Breach by Slope Instability Psi (4) Probability of Breach by Sloughing or Unravelling Psu (5) Probability of Breach by Sinkhole Psd (6) Calculate the Probability of Breach Branch on Event Tree All Failure Paths/Initiating Mechanisms Evaluate the probabilities of breach for each branch of the event tree (i. Only consider the breach mechanisms that are included for estimating the probability of breach.Psu-CE) x (1 . Excessive Erosion (Ps-lf EE) and Continuing Erosion (Ps-lf CE).

Appendix B Navigation Table for Internal Erosion Through a Soil Foundation A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 .

PEE. Pbreach-NE = 0. Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure using the event tree. Pfail = PI BEP x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes PI (IMx) Yes Backward Erosion in a Layer of Cohesionless Soil in the Foundation PP Intervention Fails No Some Erosion Continuation No Erosion PSE Progression No Σ Pfail Initiation & progression No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . (3) Probabilities for No. PSE PEE PCE (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi Estimate the probability for not detect and intervene using Table B7. PI BEP Backward erosion piping Failure Path/Location Sketch (1) Failure Path Identification and Screening Use Table B2 to identify and screen potential failure paths. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table B6. Some. PCE Evaluate the probabilities for No. PSE. (5) Probability of Breach Pbreach (6) Calculate the Probability of Failure Initiation of Backward Erosion in a Layer of Cohesionless Soil in the Foundation 3 1 3 Estimate the probabilities for forming a roof (PPR). Excessive and Continuing Erosion for the failure path under consideration using Table B5. Excessive and Continuing Erosion branches using Table B8. Some. PP = PPR x PPC x PPL Estimate the probabilities of breach for the Some.Guidance Document July 2008 Delta Version B-1 .Table B1. Probability of Failure by Internal Erosion Through a Soil Foundation (Sheet 1) (2) Evaluate the Probability of Initiation of Erosion PI Use Table B2 to evaluate the probability of initiation and progression of backward erosion for each failure path. Excessive and Continuing Erosion PNE.

PSE PEE PCE Estimate the probabilities for forming a roof (PPR). Excessive and Continuing Erosion for the failure path under consideration using Table B5. Estimate the probabilities of breach for the Some. Pfail = PI suffusion x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] 3 1 3 Pudi Internally unstable soil Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes PI (IMx) Yes Suffusion in a Layer of Cohesionless Soil in the Foundation PP Intervention Fails No Some Erosion Continuation No Erosion PSE Progression No Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Pbreach-SE Pbreach-EE Pbreach-CE (Pge-CE = 0) Calculate the probability of failure using the event tree.Guidance Document July 2008 Delta Version B-2 .Table B1. Some. PI suffusion Evaluate the probabilities for No. Excessive and Continuing Erosion branches using Table B8. Use Table B3 to evaluate the probability of initiation of suffusion for each failure path. Probability of Failure by Internal Erosion Through a Soil Foundation (Sheet 2) Initiation of Suffusion in a Layer of Cohesionless Soil in the Foundation Use Table B3 to identify and screen potential failure paths. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table B6. Pbreach-NE = 0. PP = PPR x PPC x PPL Estimate the probability for not detect and intervene using Table B7.

Excessive and Continuing Erosion for the failure path under consideration using Table B5.Table B1. Estimate the probabilities of breach for the Some. PP = PPR x PPC x PPL Estimate the probability for not detect and intervene using Table B7. Pbreach-NE = 0. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table B6. PSE PEE PCE Estimate the probabilities for forming a roof (PPR). Excessive and Continuing Erosion branches using Table B8. 3 Use Table B4 to evaluate the probability of initiation for each failure path.Guidance Document July 2008 Delta Version B-3 . Pfail = PI crack x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] Pudi Desiccation cracks in clay Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes PI (IMx) Yes Erosion in a Crack in Cohesive Soil in the Foundation PP Intervention Fails No Some Erosion Continuation No Erosion PSE Progression No Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Probability of Failure by Internal Erosion Through a Soil Foundation (Sheet 3) Initiation of Erosion in a Crack in Cohesive Soil in the Foundation Use Table B4 to identify and screen potential failure paths. Some. PI crack 3 1 Evaluate the probabilities for No. Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure using the event tree.

1. • Estimate the average seepage gradient (i avf ) through the cohesionless soil layer in the foundation beneath the central part of the dam (not at the toe where there are likely to be locally higher gradients) for the level for the reservoir stage under consideration. For each failure path.2. (P I BEP) = (P cl ) x [(P H ) x (P IH ) + (1-(P H ) x (P INH )]. Correct this average gradient for the geometry. Use i av and (i pmt ) Corrected as inputs.3. Correct this average gradient for the geometry. and grain size as detailed in the Supporting Information Section C6.0. also estimate the critical gradient (i cr ) from i cr = ( γ sat - Estimate the probability of backward erosion given heave has occurred as follows. use the approximate method described in Section 7. Identify potential failure paths for backward erosion in the soil foundations. Adopt this gradient if it is smaller than (i pmt ) Corrected . • From the particle size distribution of the foundation material. and grain size as detailed in the Supporting Information Section C6.4 from Schmertmann (2000). • From the particle size distribution of the foundation material estimate the representative uniformity coefficient Cu = D 60 /D 10 .4. PCL Backward erosion piping Assess the Probability of Heave PH Estimate the Probability of Initiation and Progression of Backward Erosion Given Heave has occurred P IH Estimate Probability of Initiation and Progression of Backward Erosion where Heave is not predicted P INH Calculate the Total Probability of Initiation and Progression of Backward Erosion Estimate the probability of initiation and progression of backward erosion (P I BEP) from the results of the assessments in the preceding steps. PH Estimate the probability of backward erosion given heave has occurred as follows.4. the probability of initiation and progression of backward erosion is assessed as follows. (2) If sand boils have not been observed. pmt γ w ) / γ w .3.5. • Estimate the average seepage gradient (i avf ) through the cohesionless soil layer in the foundation beneath the centre of the dam at the level for the reservoir stage under consideration. • Estimate the probability of initiation from Table 7. • For Cu> 6.2. For Cu> 6. For lower reservoir levels use the procedure in Section 7. horizontal to vertical permeability ratio of the zone subject to backward erosion. (1) If sand boils have been observed. Refer to Steps a) and b) in Section 7. check if is applicable using Table 3. to downstream of the embankment. Probability of Initiation of Backward Erosion in a Layer of Soil in the Foundation Failure Path/Location (1) Screening of Failure Path Probability of a Continuous Layer from Upstream to Downstream PCL Estimate the probability there is a continuous layer of cohesionless soil from upstream of the embankment. estimate the uniformity coefficient Cu = D 60 /D 10 .0. the probability of initiation and progression of backward erosion = 1. • Estimate the probability of initiation and progression from Table 7. This gives (i • pmt 3 1 3 • ) Corrected .Table B2.2. Use i av and (i inputs.2 where piezometer data and/or seepage flow net models are available.4 Assess the probability of heave for the reservoir stage under consideration.2.2. • Estimate the point seepage gradient required to progress backward erosion using Figure 6.4.2.0 for reservoir levels at or above the level at which sand boils have been observed. Use the method described in Section 7. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .6. This gives (i pmt ) Corrected . also estimate the critical gradient (i cr ) from i cr = ( γ sat - • γ w)/ γ w . horizontal to vertical permeability ratio of the zone subject to backward erosion.6.4. • Estimate the point seepage gradient required to initiate and progress backward erosion from Figure 6. Otherwise. Adopt this ) Corrected as gradient if it is smaller than (i pmt ) Corrected .Guidance Document July 2008 Delta Version B-4 .

From the particle size distribution for the soil. For example “what is the probability of a continuous crack or pattern of cracks 5mm wide.5 for soils with more than 10% fines passing 0.6.29 to 5. (P I crack) = PCL x PIC 3 1 3 Desiccation cracks in clay A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Obtain the probability of erosion from the most appropriate table from Tables 5.Table B3. The grading curve is not adjusted for this procedure. Probability of Initiation of Suffusion in a Layer of Soil in the Foundation Failure Path/Location (1) Screening of Failure Path Probability of a Continuous Layer from Upstream to Downstream PCL Estimate the probability there is a continuous layer of cohesionless soil from upstream of the embankment. and Figure 6. check if is applicable using Table 3. For each failure path. Examples. Then estimate the probability the soil is internally unstable (PIUS) from Figure 6.6 for soils with less than 10% fines passing 0. differential settlement in the foundation. Probability of Initiation of Erosion in a Crack in a Layer of Soil in the Foundation Failure Path/Location (1) Screening of Failure Path Probability of a Continuous Layer from Upstream to Downstream PCL Estimate the probability there is a layer of soil in which a continuous crack or interconnected pattern of cracks may exist.35 in Section 5. (P I suffusion) = PCL x PIUS 3 1 3 Internally unstable soil Table B4.075 mm PIUS Estimate the probability of initiation of suffusion (P I suffusion) from the results of the assessments in the preceding steps. check if is applicable using Table 3. 60% and 90% are finer). PCL Assess the Probability that Erosion will Initiate in the Crack PIC Calculate the Probability of Initiation of Erosion in a Crack PIcrack Identify potential failure paths for erosion in cracks in soil foundations.075 mm (#200 sieve). The question is best put in terms of the width of the crack. d60 and d90 sizes (the particle size for which 15%.2.1. PCL Assess the Probability the Soil is Internally Unstable PIUS Calculate the Probability of Initiation of Suffusion PIsuffusion Identify potential failure paths for suffusion in soil foundations.4.3.Guidance Document July 2008 Delta Version B-5 .2 based on the soil type containing the cracks in the foundation and for the crack width (WC) being considered. Check if soil is potentially internally unstable using Section 6. PIC Estimate the probability of initiation of erosion in a crack (P I crack) from the results of the assessments in the preceding steps.4.5 for further guidance. For each failure path.4.” Refer to Section 7. to downstream of the embankment. desiccation cracks in the foundation soil. Refer to Steps a) and b) in Section 7. determine the d15.

3 1 3 ZONED EARTHFILL WITH GRANULAR SHELLS Scenario 4 – Exits into an open defect. Some Erosion (PSE).10. The assessment of erosion into a toe drain considers the observed condition of the toe drain (from video or external inspections) and the design and construction details of the toe drain. PCE) The probability of continuation Assess the probability that the exit will be an unfiltered for an unfiltered exit PCE is.0 consideration using Table 10. Step 2: Estimate the conditional probabilities for No Erosion (PNE).4 to estimate the probabilities of No Erosion (PNE). exit for the failure path under PCE = Punf x 1. Some Erosion (PSE). PSE . Some Erosion. Estimate the probability of continuing erosion for erosion into a toe drain using Table 10. PSE . An unfiltered exit is implicit if a heave condition or a sand boil is present. Excessive Erosion (PEE) and Continuing Erosion (PCE).Guidance Document July 2008 Delta Version B-6 . PEE . SOIL SOIL CONDUIT Step 1: Evaluate the opening size for No Erosion. joint or crack. Punf Assess which filtered exit scenario is most applicable to the failure path under consideration. an unfiltered exit is implicit if a heave condition or a sand boil is present.13. Excessive Erosion and Continuing Erosion using Table 10. Probability of Continuation for Internal Erosion in the Foundation Assess the Probability that the Exit will be an Unfiltered Exit (Punf) Probability of Continuation for an Unfiltered Exit (PCE) Probability of Continuation for a Filtered Exit (PNE .11.1. Scenario 3 – Exits into a filter/transition zone or other granular material which is not capable of holding a crack/pipe. Pfe = 1 – Punf Backward erosion piping For backward erosion piping.12. 2 1 1 3 1 3 2 EARTHFILL WITH CHIMNEY FILTER ZONED EARTHFILL WITH CHIMNEY FILTER Follow the procedure outlined in Section 10.Table B5. PCE) Sketch Probability of Continuation for a Filtered Exit (PNE . Excessive Erosion (PEE) and Continuing Erosion (PCE) by estimating the proportion of soils falling within each erosion category and using Table 10. PEE . (Note: Scenarios 1 and 2 are not applicable to this failure mode) 3 1 3 Calculate the probability that there will be a filtered exit (Pfe). OPEN JOINTED ROCK EROSION INTO AN OPEN CRACK OR JOINT IN A CONDUIT OR WALL EROSION INTO OPEN JOINTS IN ROCK FOUNDATION Scenario 5 – Erosion into a toe drain 1 1 SION THROUGH THE INTO A TOE DRAIN INTERNAL EROSION THROUGH THE FOUNDATION INTO A TOE DRAIN A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . For all filtered exit scenarios. Figure A1 shows a flow chart which summarizes the procedure.

12. The filtering material may be a naturally occurring layer in the foundation. that it is not detected Pnd Assess the probability of not detecting the leak using Table 12.7. There needs to be a filtering material at the downstream end of the flow path for crack filling action to be effective.3.Guidance Document July 2008 Delta Version B-7 . then consider the potential for the overlying soil layers to wash into the developing pipe. If the eroding foundation soil layer is located further below the embankment. Pnd from Step (3) and Pui from Step (5).5 and obtain probability from Table 12. = Pnol + [(1-Pnol) x Pnd] (5) Assess the probability that intervention and repair is not successful Pui Assess the probability that intervention and repair is not successful using Table 12.8.Table B6. and the embankment forms the roof of the pipe. upstream zone or a concrete element fails to limit flows using Table 11.2.3. Calculate the probability of unsuccessful detection and intervention using Pnol from Step (2).1 as a guide to estimating the probability of the embankment and foundation materials in supporting the roof of a pipe in the foundation.2 and 12.1. Pndi = Pnol + [(1-Pnol) x Pnd x Pui] A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Probability of Unsuccessful Detection and Intervention for Internal Erosion in the Foundation Probability of Not Detecting Applicable Failure Path/Location (1) Assess the Rate of Internal Erosion and Piping Probability of Not Intervening (4) Calculate the probability of not detecting the internal erosion Pndi Calculate the probability of not detecting the internal erosion using Pnol from Step (2) and Pnd from Step (3). PPC Probability that Upstream Zone Fails to Limit Flows (PPL) Assess the probability that an upstream soil layer. Pnol (3) Assess the probability that given the leak is observable.6. PPL Table B7. PPR Probability that Crack Filling Action is Not Effective (PPC) If the eroding foundation soil layer is located immediately below the embankment. then evaluate the probability for crack filling action as for erosion in the dam using Table 11. or the embankment filter. Pnd All Failure Modes Estimate the approximate time for progression of piping and development of a breach using Tables 12. Probability of Progression for Internal Erosion in the Embankment Applicable Failure Path/Location All IE Foundation Failure Modes Probability of Forming a Roof (PPR) Use Table 11. Pui (6) Probability of Not Detect and Not Intervene Pudi (2) Assess the probability of the concentrated leak not being observable Pnol Assess the weighted score (WS) from Table 12.

20.Table B8. Excluded breach mechanisms are assigned a probability of breach = 0. (c) For dams with a downstream zone of rockfill.15 and obtain the probability for each branch of the tree from Table 13. Some Erosion (Psu-SE).11. estimate the probability of breach by gross enlargement using Table 13. Pge (a) Estimate the probability of slope instability occurring due to the increased seepage flows. Pbreach-EE= 1 – [(1 . A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .Psu-CE) x (1 . (Psi-CE) = (Psi-i CE) x (Psi-lf). assess the weighted score from Table 13. Excessive Erosion (Psi-i EE) and Continuing Erosion (Psi-i CE)]. The probability of breach by gross enlargement Pge= 0 where the mode of erosion is suffusion.13.e.18. No. Some Erosion.Psd-CE)]. (b) For dams with a downstream zone of earthfill. Excessive Erosion (Psu-EE) and Continuing Erosion (Psu-CE). (a) Estimate the probability of a sinkhole developing as a result of the internal erosion (Ps-f) from Table 13. Excessive Erosion (Ps-lf EE) and Continuing Erosion (Ps-lf CE). Pbreach-CE = 1 – [(1 .4 and obtain the probability from Table 13. Some Erosion (Ps-lf SE). No Erosion Branch (NE) 0 Some Erosion Branch (SE) Excessive Erosion Branch (EE) Continuing Erosion Branch (CE) Pbreach = 0 for NE Pge-SE = 0 Pge-EE = 0 Pge-CE from Table 10. (b) Estimate the probability that the sinkhole causes loss of freeboard. assess the weighted score from Table 13. (c) Calculate the probability of breach by slope instability for each branch of the tree (a) Probability seepage will emerge into the downstream shell PS = 1. Assess the weighted score from Table 13.16.1.Pge-CE) x (1 Psi-CE) x (1 . Excessive Erosion and Continuing Erosion branches) by summing the probabilities using de Morgan’s rule as follows. (c) Calculate the probability of breach by sinkhole development Calculate the probability of breach for each branch of the event tree (i. Assess the weighted score from Table 13.10 and obtain the probability from Table 13. Some and Excessive Erosion branches. Probability of Breach for Internal Erosion in the Foundation Applicable Failure Path/Location (1) Screening of breach mechanisms (2) Probability of Breach by Gross Enlargement Pge (3) Probability of Breach by Slope Instability Psi (4) Probability of Breach by Sloughing or Unravelling Psu (5) Probability of Breach by Sinkhole Psd (6) Calculate the Probability of Breach Branch on Event Tree All Failure Modes Evaluate the probabilities of breach for each branch of the event tree (i. Excessive Erosion (Psu-EE) and Continuing Erosion (Psu-CE). Some.Psd-SE)].19 and obtain the probability for each branch of the event tree from Table 13. The probability of breach by gross enlargement Pge= 0 for the No. (b) Estimate the probability of loss of freeboard due to instability (Psi-lf).12 and obtain the probability for each branch of the tree from Table 13. (Psu-SE) (Psu-EE) (Psu-CE) (Psd-SE) = (Ps-f) x (Ps-lf SE) (Psd-EE) = (Ps-f) x (Ps-lf EE) (Psd-CE) = (Ps-f) x (Ps-lf CE) Pbreach-NE = 0 Pbreach-SE= 1 – [(1 .3.e.Pge-SE) x (1 Psi-SE) x (1 .5. Assess the weighted score from Table 13. This is done for each branch of the event tree [Some Erosion (Psi-i SE). Some Erosion (Psu-SE).0. Excessive and Continuing Erosion branches) Assess which breach mechanisms are relevant to the type of dam zoning and failure path being considered using Table 13.Psu-EE) x (1 . (Psi-EE) = (Psi-i EE) x (Psi-lf).Pge-EE) x (1 Psi-EE) x (1 .Psu-SE) x (1 .3 (Psi-SE) = (Psi-i SE) x (Psi-lf).Guidance Document July 2008 Delta Version B-8 . If this breach mechanism is applicable.Psd-EE)].

Appendix C Navigation Table for Internal Erosion Through a Rock Foundation A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 .

Table C1. (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi (6) Probability of Breach Pbreach (7) Calculate the Probability of Failure Failure Path/Location Sketch (1) Failure Path Identification and Screening Use Table C2 to identify and screen potential failure paths. PSE.Guidance Document July 2008 Delta Version C-1 . Pbreach-25-100mm Pbreach->100mm Yes Breach No Calculate the probability of failure using the event tree. Excessive and Continuing Erosion PNE. Excessive and Continuing Erosion branches using Table C8. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table C6. PEE. Initiation of Erosion in Defects Related to Stress Relief Effects in the Valley Sides Stress Relief Defects Long Section Estimate the probabilities for forming a roof (PPR). Pd valley side 5-25mm. Pbreach-5-25mm. PP = PPR x PPC x PPL Estimate the probability for not detect and intervene using Table C7. 25-100mm and >100mm defects (3) Probabilities for No. Some. Probability of Failure by Internal Erosion Through a Rock Foundation (Sheet 1) (2) Evaluate the Probability of the Presence of Open Defects or In Filled Defects that Have Eroded Pd Use Table C2 to evaluate the probability of the presence of open defects or in filled defects that have eroded. Pfail = Pd-5-25mm x PP x Pudi x Pbreach-5-25mm + Pd-25-100mm x PP x Pudi x Pbreach-25100mm + Pd->100mm x PP x Pudi x Pbreach>100mm Pbreach->100mm EVENT TREE STRUCTURE Yes Continuing Erosion >100mm Continuation Progression No PP Yes Pudi Pfail->100mm Intervention Fails No No Erosion Yes Yes Continuing Erosion 25-100mm Continuation Progression No PP Pudi Pbreach-25-100mm Yes Breach No No Pbreach-5-25mm Pfail-25-100mm Intervention Fails No Erosion Yes Yes Continuing Erosion 5-25mm Pd Continuation Progression No PP Pudi Yes Breach No Pfail-5-25mm Intervention Fails No Erosion in defects in a rock foundation related to stress relief effects in the valley sides Presence of open or in filled defects <5mm No Erosion Σ Pfail A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Pudi Estimate the probabilities of breach for the Some. PCE The likelihood of continuation of the infill material is already considered in step (2).

Guidance Document July 2008 Delta Version C-2 . PP = PPR x PPC x PPL Estimate the probability for not detect and intervene using Table C7. Pd valley floor 5-25mm. Excessive and Continuing Erosion branches using Table C8. Estimate the probabilities for forming a roof (PPR). crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table C6.Table C1. Pfail = Pd-5-25mm x PP x Pudi x Pbreach-5-25mm + Pd-25-100mm x PP x Pudi x Pbreach-25100mm + Pd->100mm x PP x Pudi x Pbreach>100mm Valley Bulge Features Long Section Pudi Pbreach->100mm EVENT TREE STRUCTURE Yes Continuing Erosion >100mm Continuation Progression No PP Yes Pudi Yes Breach No Pfail->100mm Intervention Fails No No Erosion Yes Yes Continuing Erosion 25-100mm Continuation Progression No PP Pudi Pbreach-25-100mm Yes Breach No No Pbreach-5-25mm Pfail-25-100mm Intervention Fails No Erosion Yes Yes Continuing Erosion 5-25mm Pd Continuation Progression No PP Pudi Yes Breach No Pfail-5-25mm Intervention Fails No Erosion in defects in a rock foundation related to stress relief effects in the valley floor Presence of open or in filled defects <5mm No Erosion Σ Pfail A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . 25-100mm and >100mm defects The likelihood of continuation of the infill material is already considered in step (2). Use Table C3 to evaluate the probability of initiation for each failure path. Pbreach-25-100mm Pbreach->100mm Calculate the probability of failure using the event tree. Pbreach-5-25mm. Probability of Failure by Internal Erosion Through a Rock Foundation (Sheet 2) Initiation of Erosion in Defects Related to Stress Relief Effects in the Valley Floor – Valley Bulge and Rebound Use Table C3 to identify and screen potential failure paths. Estimate the probabilities of breach for the Some.

PP = PPR x PPC x PPL Estimate the probability for not detect and intervene using Table C7.Guidance Document July 2008 Delta Version C-3 . Pbreach-5-100mm. Estimate the probabilities for forming a roof (PPR). dolomite Pudi Pbreach->100mm EVENT TREE STRUCTURE Yes Continuing Erosion >300mm Continuation Progression No PP Yes Pudi Yes Breach No Pfail->300mm Intervention Fails No No Erosion Yes Yes Continuing Erosion 100-300mm Continuation Progression No PP Pudi Pbreach-25-100mm Yes Breach No No Pbreach-5-25mm Pfail-100-300mm Intervention Fails No Erosion Yes Yes Continuing Erosion 5-100mm Pd Continuation Progression No PP Pudi Yes Breach No Pfail-5-100mm Intervention Fails No Erosion in solution features in a rock foundation Presence of open or in filled defects <5mm No Erosion Σ Pfail A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Excessive and Continuing Erosion branches using Table C8. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table C6. 100-300mm and >300mm defects The likelihood of continuation of the infill material is already considered in step (2). Estimate the probabilities of breach for the Some. Pfail = Pd-5-100mm x PP x Pudi x Pbreach-5-100mm + Pd-100-300mm x PP x Pudi x Pbreach-100300mm + Pd->300mm x PP x Pudi x Pbreach>300mm Solution Features Long Section Limestone. Pd solution 5-100mm.Table C1. Probability of Failure by Internal Erosion Through a Rock Foundation (Sheet 3) Initiation of Erosion in Solution Features for Rock Subject to Solution Use Table C4 to identify and screen potential failure paths. Use Table C4 to evaluate the probability of initiation for each failure path. Pbreach-100-300mm Pbreach->300mm Calculate the probability of failure using the event tree.

Pbreach-5-25mm. Estimate the probabilities of breach for the Some. 25-100mm and >100mm defects The likelihood of continuation of the infill material is already considered in step (2). crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table C6. Pd other 5-25mm.Table C1. Probability of Failure by Internal Erosion Through a Rock Foundation (Sheet 4) Initiation of Erosion in Features for Associated with Landslides and Faults and Shears Use Table C5 to identify and screen potential failure paths. Use Table C5 to evaluate the probability of initiation for each failure path. PP = PPR x PPC x PPL Estimate the probability for not detect and intervene using Table C7. Estimate the probabilities for forming a roof (PPR). Excessive and Continuing Erosion branches using Table C8. Pbreach-25-100mm Pbreach->100mm Calculate the probability of failure using the event tree. Pfail = Pd-5-25mm x PP x Pudi x Pbreach-5-25mm + Pd-25-100mm x PP x Pudi x Pbreach-25100mm + Pd->100mm x PP x Pudi x Pbreach>100mm Defects associated with landslide Long Section Fault or Shear Zone Pudi Pbreach->100mm EVENT TREE STRUCTURE Yes Continuing Erosion >100mm Continuation Progression No PP Yes Pudi Yes Breach No Pfail->100mm Intervention Fails No No Erosion Yes Yes Continuing Erosion 25-100mm Continuation Progression No PP Pudi Pbreach-25-100mm Yes Breach No No Pbreach-5-25mm Pfail-25-100mm Intervention Fails No Erosion Yes Yes Continuing Erosion 5-25mm Pd Continuation Progression No PP Pudi Yes Breach No Pfail-5-25mm Intervention Fails No Erosion in defects in a rock foundation associated with landslides/faults/shears Presence of open or in filled defects <5mm No Erosion Σ Pfail A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .Guidance Document July 2008 Delta Version C-4 .

Assess the filter materials and materials being eroded in terms of the Continuing Erosion criteria as described using the procedure in Section 10.Table C2.0% Does erosion initiate? No 50.18. The potential failure modes arising from the defects should be assessed and sketches prepared to show them so the risk analysis team have a clear picture of the failure modes and their relation to the defects.0% Estimated width below cutoff trench X mm Open feature of X mm width Does erosion continue? No 50.1 and 8.13 to aid judgment. Assess the weighted score (WS) from Tables 8. 25-100mm and >100mm. Pw(5-25mm) = w PTG(5-25mm) + (1-w) PSC(5-25mm) Pw(25-100mm) = w PTG(25100mm) + (1-w) PSC(25100mm) Estimate the width and extent of the open or in filled defects below the original ground surface using Section 8.1 where weighting is applied.4.0% repeat tree Sub-event tree structure to show computation of a continuous open defect below the embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Estimate the probability of erosion of infill initiating using first principles using Section 8.Punf.Guidance Document July 2008 Delta Version C-5 . Pd-(5-25mm) valley side Pd-(25-100mm) valley side Pd-(>100mm) valley side Describe the defects in relation to the embankment details (refer to Section 8. Compute the probability of a continuous open defect in rock foundation below the embankment using the event tree shown in Figure 8.0% 50. Calculate the weighted average of the two probability estimates. This is done for three defect sizes. grouting is assumed to be ineffective and assign a probability for grouting not being effective = 1.0% Yes 50. 5-25mm.0% 5-25mm 10. 25-100mm and >100mm.0% 50. For in filled defects.0% Is grouting ineffective? No 50.19 and 8.13). Pw(>100mm) = w PTG(>100mm ) + (1-w) PSC(>100mm ) Yes Open 50. Calculate the probability that there will be a filtered exit (Pfe) = 1.0) + (Pfe x PCE). PSC-(5-25mm) PSC-(25-100mm) PSC-(>100mm) Obtain the weighting factor (w) based on the quantity and quality of the investigation and construction data using Table 8.21 to estimate the probability of erosion initiating based on performance data.10.3. PTG-(5-25mm) PTG-(25-100mm) PTG-(>100mm) Estimate the probability of a continuous open or in filled defect in the rock foundation from upstream of the core to downstream of the core. Assess the probability that the defects below the level of the core are open or in filled using Table 8. This is to be done for each defect or solution feature width using the weighted probabilities from Section 8. Do this for each defect size. Assess the weighted score (WS) from Tables 8. 5-25mm. Estimate the probability of a continuous open or in filled defect in the rock foundation from upstream of the core to downstream of the core.5.12.16 Use Table 8. and obtain the probabilities for each defect size from Table 8. This is done for three defect sizes. Estimate the weighting factor based on the quality of the input data.17 and 8.0% Infilled 50.0% Open feature of X mm width Is grouting ineffective? No 50..20 where a cut-off has been excavated and backfilled in the rock foundation to intercept the continuous open defect or solution feature.4. Estimate the probability that there will an unfiltered exit (Punf) using Table 10.5 (see below). The probability of continuing erosion = (Punf x 1.0% Is feature open or infilled? Yes Yes 50. Calculate the weighted estimate of the probability of the presence of open or in filled defects.1. assess the likelihood of grouting not being effective using Tables 8.22.6.2.3.0% A butment stress relief features Continuous open or infilled feature present? 25-100mm >100mm 1. Use Tables 8.1. Probability of Initiation of Erosion in Defects Related to Stress Relief Effects in the Valley Sides in a Rock Foundation Failure Path/Location Probability of Continuous Open or in Filled Defects Based on Geology and Topography PTG Probability of Continuous Open or in Filled Defects Based on Investigation and Construction Data PSC Combine the Two Probability Estimates for Continuous Open or in Filled Defects Assess the Width and Extent of the Open or In Filled Defects Assess the Probability that the Defects are Open or In Filled Assess the Probability that Grouting or Cutoff Walls Has Not Been Effective in Cutting Off the Defects For open defects. Assess the Probability that Erosion of the Infill will Initiate Assess the Probability that Erosion of the Infill in Defects will Continue Combine the probabilities for open defects and in filled defects which will potentially erode Pd Describe the Defects and Failure Modes Identify potential failure paths for initiation of erosion in defects related to stress relief effects in the valley sides. and obtain the probabilities for each defect size from Table 8.0.0% 50.

Assess the weighted score (WS) from Tables 8.0% Yes 50. Pd-(5-25mm) valley floor Pd-(25-100mm) valley floor Pd-(>100mm) valley floor Describe the defects in relation to the embankment details (refer to Section 8.13 to aid judgment. Assess the probability that the defects below the level of the core are open or in filled using Table 8. This is to be done for each defect or solution feature width using the weighted probabilities from Section 8.5 (see below).13).9. 5-25mm.16. 25-100mm and >100mm. Assess the Probability that Erosion of the Infill will Initiate Assess the Probability that Erosion of the Infill in Defects will Continue Combine the probabilities for open defects and in filled defects which will potentially erode Pd Describe the Defects and Failure Modes Identify potential failure paths for initiation of erosion in defects related to stress relief effects in the valley floor.0% Infilled 50.4. assess the likelihood of grouting not being effective using Tables 8.7.20 where a cut-off has been excavated and backfilled in the rock foundation to intercept the continuous open defect or solution feature.0% repeat tree Sub-event tree structure to show computation of a continuous open defect below the embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .0% Open feature of X mm width Is grouting ineffective? No 50.0% A butment stress relief features Continuous open or infilled feature present? 25-100mm >100mm 1. Estimate the weighting factor based on the quality of the input data.Guidance Document July 2008 Delta Version C-6 . Probability of Initiation of Erosion in Defects Related to Stress Relief Effects in the Valley Floor – Valley Bulge and Rebound Failure Path/Location Probability of Continuous Open or in Filled Defects Based on Geology and Topography PTG Probability of Continuous Open or in Filled Defects Based on Investigation and Construction Data PSC Combine the Two Probability Estimates for Continuous Open or in Filled Defects Assess the Width and Extent of the Open or In Filled Defects Assess the Probability that the Defects are Open or In Filled Assess the Probability that Grouting or Cutoff Walls Has Not Been Effective in Cutting Off the Defects For open defects.Punf.17 and 8.0% 50. Calculate the probability that there will be a filtered exit (Pfe) = 1. Compute the probability of a continuous open defect in rock foundation below the embankment using the event tree shown in Figure 8. This is done for three defect sizes.0% Is feature open or infilled? Yes Yes 50. PTG-(5-25mm) PTG-(25-100mm) PTG-(>100mm) Estimate the probability of a continuous open or in filled defect in the rock foundation from upstream of the core to downstream of the core. Calculate the weighted average of the two probability estimates.8.18. 5-25mm.3. This is done for three defect sizes. and obtain the probabilities for each defect size from Table 8.6.0% 5-25mm 10. Pw(>100mm) = w PTG(>100mm ) + (1-w) PSC(>100mm ) Yes Open 50.12. The probability of continuing erosion = (Punf x 1. Calculate the weighted estimate of the probability of the presence of open or in filled defects.0% Does erosion initiate? No 50. Assess the weighted score (WS) from Table 8. grouting is assumed to be ineffective and assign a probability for grouting not being effective = 1.. Use Table 8.1.10.0% 50. Estimate the probability that there will an unfiltered exit (Punf) using Table 10.0% Is grouting ineffective? No 50. Pw(5-25mm) = w PTG(5-25mm) + (1-w) PSC(5-25mm) Pw(25-100mm) = w PTG(25100mm) + (1-w) PSC(25100mm) Estimate the width and extent of the open or in filled defects below the original ground surface using Section 8. Use Tables 8.Table C3. Do this for each defect size.22.0) + (Pfe x PCE). PSC-(5-25mm) PSC-(25-100mm) PSC-(>100mm) Obtain the weighting factor (w) based on the quantity and quality of the investigation and construction data using Table 8. Estimate the probability of erosion of infill initiating using first principles using Section 8.2. 25-100mm and >100mm.21 to estimate the probability of erosion initiating based on performance data.6 and obtain the probabilities for each defect size from Table 8.1 where weighting is applied. The potential failure modes arising from the defects should be assessed and sketches prepared to show them so the risk analysis team have a clear picture of the failure modes and their relation to the defects.0.19 and 8.0% 50. Assess the filter materials and materials being eroded in terms of the Continuing Erosion criteria as described using the procedure in Section 10.0% Estimated width below cutoff trench X mm Open feature of X mm width Does erosion continue? No 50. For in filled defects. Estimate the probability of a continuous open or in filled defect in the rock foundation from upstream of the core to downstream of the core.

Assess the weighted score (WS) from Table 8.16.0% Estimated width below cutoff trench X mm Open feature of X mm width Does erosion continue? No 50. Calculate the weighted estimate of the probability of the presence of open or in filled solution feature. 100-300mm and >300mm.Table C4.10 and obtain the probabilities for each defect size from Table 8.0% A butment stress relief features Continuous open or infilled feature present? 25-100mm >100mm 1. Use Tables 8.12. Estimate the weighting factor based on the quality of the input data. 100mm. 100-300mm and >300mm. Assess the Probability that Erosion of the Infill will Initiate Assess the Probability that Erosion of the Infill in Solution Features will Continue Combine the probabilities for open features and in filled features which will potentially erode Pd Describe the Defects and Failure Modes Identify potential failure paths for initiation of erosion in defects related to solution features.21 to estimate the probability of erosion initiating based on performance data.0% Does erosion initiate? No 50. Pd-(5-100mm) solution Pd-(100-300mm) solution Pd-(>300mm) solution Describe the solution features in relation to the embankment details (refer to Section 8.. For in filled defects. This is done for three sizes.18.0% 50. Probability of Initiation of Erosion in Defects Related to Solution Features for Rock Subject to Solution Failure Path/Location Probability of Continuous Open or in Filled Solution Feature Based on Geology and Topography PTG Estimate the probability of a continuous open or in filled solution feature in the rock foundation from upstream of the core to downstream of the core.0% repeat tree Sub-event tree structure to show computation of a continuous open defect below the embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .0% Open feature of X mm width Is grouting ineffective? No 50. assess the likelihood of grouting not being effective using Tables 8. Assess the probability that the solution features below the level of the core are open or in filled using Table 8.0% 50.0% Infilled 50.Punf.10.0% 5-25mm 10. Do this for each defect size.4.13). Estimate the probability that there will an unfiltered exit (Punf) using Table 10.17 and 8.12.0% 50. and obtain the probabilities for each defect size from Table 8.Guidance Document July 2008 Delta Version C-7 .0.0% Yes 50. 5100mm. This is done for three sizes.20 where a cut-off has been excavated and backfilled in the rock foundation to intercept the continuous open defect or solution feature.19 and 8.5 (below).0% Is grouting ineffective? No 50.3. Calculate the weighted average of the two probability estimates. Compute the probability of a continuous open solution feature in rock foundation below the embankment using the event tree shown in Figure 8. Pw(>300mm) = w PTG(>300mm ) + (1-w) PSC(>300mm ) Yes Open 50.13.11.1.13 to aid judgment.1 where weighting is applied. PTG-(5-100mm) PTG-(100-300mm) PTG-(>300mm) Probability of Continuous Open or in Filled Solution Feature Based on Investigation and Construction Data PSC Estimate the probability of a continuous open or in filled solution feature in the rock foundation from upstream of the core to downstream of the core. Use Table 8. The probability of continuing erosion = (Punf x 1.22. Estimate the probability of erosion of infill initiating using first principles using Section 8.3. The potential failure modes arising from the solution features should be assessed and sketches prepared to show them so the risk analysis team have a clear picture of the failure modes and their relation to the defects. Assess the filter materials and materials being eroded in terms of the Continuing Erosion criteria as described using the procedure in Section 10.6. Obtain the weighting factor (w) based on the quantity and quality of the investigation and construction data using Table 8. Assess the weighted score (WS) from Tables 8. Pw(5-100mm) = w PTG(5100mm) + (1-w) PSC(5-100mm) Pw(100-300mm) = w PTG(100300mm) + (1-w) PSC(100300mm) Estimate the width and extent of the open or in filled solution feature below the original ground surface using Section 8. This is to be done for each solution feature width using the weighted probabilities from Section 8. grouting is assumed to be ineffective and assign a probability for grouting not being effective = 1. PSC-(5-100mm) PSC-(100-300mm) PSC-(>300mm) Combine the Two Probability Estimates for Continuous Open or in Filled Solution Features Assess the Width and Extent of the Open or In Filled Solution Features Assess the Probability that the Solution Features are Open or In Filled Assess the Probability that Grouting or Cutoff Walls Has Not Been Effective in Cutting Off the Solution Features For open defects. Calculate the probability that there will be a filtered exit (Pfe) = 1.0% Is feature open or infilled? Yes Yes 50.0) + (Pfe x PCE).

The potential failure modes arising from the defects should be assessed and sketches prepared to show them so the risk analysis team have a clear picture of the failure modes and their relation to the defects.0% Is feature open or infilled? Yes Yes 50. grouting is assumed to be ineffective and assign a probability for grouting not being effective = 1.0% Estimated width below cutoff trench X mm Open feature of X mm width Does erosion continue? No 50.1 where weighting is applied. Calculate the weighted average of the two probability estimates.14.13). Some Erosion. Use Tables 8.Guidance Document July 2008 Delta Version C-8 . 5-25mm. Estimate the probability of erosion of infill initiating using first principles using Section 8.15.3 as a guide. Pd-(5-25mm) other Pd-(25-100mm) other Pd-(>100mm) other Describe the Defects and Failure Modes Identify potential failure paths for initiation of erosion associated with landslides.12.0% 5-25mm 10. This is to be done for each defect width using the weighted probabilities from Section 8.0% Open feature of X mm width Is grouting ineffective? No 50. Excessive Erosion and Continuing Erosion criteria as described using the procedure in Section 10. assess the likelihood of grouting not being effective using Tables 8..Punf.17 and 8.0% repeat tree Sub-event tree structure to show computation of a continuous open defect below the embankment A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Describe the defects in relation to the embankment details (refer to Section 8. Faults and Shears Failure Path/Location Probability of Continuous Open or in Filled Defects Based on Investigation and Construction Data PSC Estimate the probability of a continuous open or in filled defect in the rock foundation from upstream of the core to downstream of the core.3. Assess the weighted score (WS) from Tables 8.0% 50.19 and 8. For in filled defects.0% A butment stress relief features Continuous open or infilled feature present? 25-100mm >100mm 1.0% 50. Assess the Probability that Erosion of the Infill will Initiate Assess the Probability that Erosion of the Infill in Defects will Continue Combine the probabilities for open defects and in filled defects which will potentially erode Pd Compute the probability of a continuous open defect in rock foundation below the embankment using the event tree shown in Figure 8.4.0% Is grouting ineffective? No 50.6.5. and obtain the probabilities for each defect size from Table 8.18. Estimate the weighting factor based on the quality of the input data. Calculate the probability that there will be a filtered exit (Pfe) = 1.21 to estimate the probability of erosion initiating based on performance data.13 to aid judgment. The probability of continuing erosion = (Punf x 1. Pw-(5-25mm) Pw-(25-100mm) Pw-(>100mm) Assess the Width and Extent of the Open or In Filled Defects Assess the Probability that the Defects are Open or In Filled Assess the Probability that Grouting or Cut-off Walls Has Not Been Effective in Cutting Off the Defects For open defects. Assess the filter materials and materials being eroded in terms of the No Erosion. faults and shears. Yes Open 50. Use Table 8.0) + (Pfe x PCE).0% Yes 50.Table C5. Assess the probability that the defects below the level of the core are open or in filled using Table 8. This is done for three defect sizes.10.0% 50. Probability of Initiation of Erosion in Defects Associated with Other Geological Features such as Landslides.20 where a cut-off has been excavated and backfilled in the rock foundation to intercept the continuous open defect or solution feature.16.0.0% Infilled 50.1.0% Does erosion initiate? No 50. Estimate the probability that there will an unfiltered exit (Punf) using Table 10. 25-100mm and >100mm. Estimate the width and extent of the open or in filled defects below the original ground surface using Section 8.

2 and 12.6. There needs to be a filtering material at the downstream end of the flow path for crack filling action to be effective. Probability of Progression for Internal Erosion in a Rock Foundation Applicable Failure Path/Location All IE Foundation Failure Modes Probability of Forming a Roof (PPR) The probability of the rock foundation materials in supporting the roof of a pipe in the foundation is assumed to be 1.7. Pnd from Step (3) and Pui from Step (5). upstream zone or a concrete element fails to limit flows using Table 11. Pnd All Failure Modes Estimate the approximate time for progression of piping and development of a breach using Tables 12.Table C6. PPC Probability that Upstream Zone Fails to Limit Flows (PPL) Assess the probability that an upstream soil layer. Pndi = Pnol + [(1-Pnol) x Pnd x Pui] A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . = Pnol + [(1-Pnol) x Pnd] (5) Assess the probability that intervention and repair is not successful Pui Assess the probability that intervention and repair is not successful using Table 12.3. or the embankment filter.Guidance Document July 2008 Delta Version C-9 . Pui (6) Probability of Not Detect and Not Intervene Pudi (2) Assess the probability of the concentrated leak not being observable Pnol Assess the weighted score (WS) from Table 12.1. The filtering material may be a naturally occurring layer in the foundation.5 and obtain probability from Table 12. Calculate the probability of unsuccessful detection and intervention using Pnol from Step (2). PPR = 1.0 Probability of Crack Filling Action Not Effective (PPC) Consider the potential for the overlying embankment or foundation soils to wash into the developing pipe. that it is not detected Pnd Assess the probability of not detecting the leak using Table 12. 12. Pnol (3) Assess the probability that given the leak is observable.8.3.0. Probability of Unsuccessful Detection and Intervention for Internal Erosion in a Rock Foundation Probability of Not Detecting Applicable Failure Path/Location (1) Assess the Rate of Internal Erosion and Piping Probability of Not Intervening (4) Calculate the probability of not detecting the internal erosion Pndi Calculate the probability of not detecting the internal erosion using Pnol from Step (2) and Pnd from Step (3). PPL Table C7.

Psd-100-300mm)]. <5mm 5-25mm 25-100mm Pge = 0 Pge = 0 Pge = 0 (Psi<5mm) = PS x (Psi-i <5mm) x (Psi-lf).6. (b) Estimate the probability of slope instability due to the increased flows Psii. Assess the weighted score from Table 13.Psu->300mm) x (1 . Ignore this failure mode unless the size of the in filled defect or solution feature is very large (e.Guidance Document July 2008 Delta Version C-10 . (c) Calculate the probability of breach by slope instability for each defect size.Psi-5-25mm) x (1 .20.Psi-100-300mm) x (1 . for each of the defect sizes) by summing the probabilities using de Morgan’s rule as follows.12 and obtain the probability for each defect size from Table 13.18. (a) Estimate the probability that leakage through the rock foundations exits into the downstream shell PS using Table 13.9. 100-300mm Pge = 0 (Psi-100-300mm) = PS x (Psi-i-100-300mm) x (Psilf).Psi-25-100mm) x (1 .7 and 13.11. assess the weighted score from Table 13.g. Calculate the probability of breach for each branch of the event tree (i. (b) Estimate the probability of loss of freeboard due to instability (Psi-lf). in-filled caverns or caves in karst > 2m).6.Psd-25100mm)].Pge-100300mm) x (1 . Psus (a) Estimate the probability of a sinkhole developing as a result of the internal erosion (Ps-f) from Table 13.19 and obtain the probability from Table 13. (Psu-100-300mm) = PS x (Psus-100-300mm) >300mm Pge->300mm (Psu->300mm) = PS x (Psus->300mm) A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .Psi->300mm) x (1 .Psd-5-25mm)]. Pge (a) Estimate the probability that leakage through the rock foundations exits into the downstream shell PS using Table 13. (Psi-5-25mm) = PS x (Psi-i 5-25mm) x (Psi-lf). Pbreach-5-25mm = 1 – [(1 . (Psu-<5mm) = PS x (Psus-<5mm) (Psu-5-25mm) = PS x (Psus-5-25mm) (Psu-25-100mm) = PS x (Psus-25-100mm) (Psd-<5mm) = 0 (Psd-5-25mm) = (Ps-f) x (Ps-lf SE) (Psd-25-100mm) = (Ps-f) x (Pslf EE) (Psd-100-300mm) = (Ps-f) x (Ps-lf CE) (Psd->300mm) = (Ps-f) x (Ps-lf CE) Pbreach-<5mm = 1 – [(1 . Pbreach->300mm = 1 – [(1 . (Psi->300mm) = PS x (Psi-i->300mm) x (Psi-lf). Assess the weighted score from Tables 13. (c) Calculate the probability of breach by sinkhole development. Assess the weighted score from Table 13. Pbreach-100-300mm = 1 – [(1 .Psu-5-25mm) x (1 .Psi-<5mm) x (1 Psu-<5mm)]. This is done for each defect size.17. (Psi-25-100mm) = PS x (Psi-i-25-100mm) x (Psi-lf).15 and obtain the probability for defect size from Table 13.8 and obtain the probability from Table 13. Psus (c) For dams with a downstream zone of rockfill.Pge->300mm) x (1 . assess the weighted score from Table 13. (b) Estimate the probability that the sinkhole causes loss of freeboard.Psu-25-100mm) x (1 .e.Psd->300mm)].10 and obtain the probability from Table 13. (b) For dams with a downstream zone of earthfill. Probability of Breach for Internal Erosion in a Rock Foundation Applicable Failure Path/Location Size of Defects in the Rock Foundation (1) Screening of breach mechanisms (2) Probability of Breach by Gross Enlargement Pge (3) Probability of Breach by Slope Instability Psi (4) Probability of Breach by Sloughing or Unravelling Psu (5) Probability of Breach by Sinkhole Psd (6) Calculate the Probability of Breach All Failure Modes Evaluate the probabilities of breach for each defect size Not applicable to this mode of internal erosion.14. Pbreach-25-100mm = 1 – [(1 .Psu100-300mm) x (1 .Table C8.

Appendix D Navigation Table for Internal Erosion of the Embankment into or at the Foundation A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 .

Pfail = PI-BEP x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes PI-BEP Yes Backward Erosion or Suffusion from the Embankment into the Foundation PP Intervention Fails No Some Erosion Continuation No Erosion PSE Progression No Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Excessive and Continuing Erosion for the failure path under consideration using Table D5. Probability of Failure by Internal Erosion of the Embankment at or into the Foundation (Sheet 1) (2) Evaluate the Probability of Initiation of Erosion PI Use Table D2 to evaluate the probability of initiation.Table D1. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table D6. PCE Evaluate the probabilities for No. Pbreach-NE = 0. Excessive and Continuing Erosion PNE. PI –BEP (3) Probabilities for No. PSE PEE PCE (4) Probability of Progression PP (5) Probability of Unsuccessful Detection and Intervention Pudi Estimate the probability for not detect and intervene using Table D7. PP = PPR x PPC x PPL Estimate the probabilities of breach for the Some.Guidance Document July 2008 Delta Version D-1 . Excessive and Continuing Erosion branches using Table D8. Some. Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure using the event tree. Some. PSE. (5) Probability of Breach Pbreach Failure Path/Location Sketch (1) Failure Path Identification and Screening Use Table D2 to identify and screen potential crack mechanisms (6) Calculate the Probability of Failure Initiation of Erosion by Backward Erosion or Suffusion from a high permeability zone in the core or cut-off trench into the foundation 3 1 3 Backward erosion piping Estimate the probabilities for forming a roof (PPR). PEE.

Excessive and Continuing Erosion for the failure path under consideration using Table D5. Estimate the probabilities of breach for the Some.Guidance Document July 2008 Delta Version D-2 . Excessive and Continuing Erosion branches using Table D8. Probability of Failure by Internal Erosion of the Embankment at or into the Foundation (Sheet 2) Initiation of Erosion by Scour at the CoreFoundation Contact Use Table D3 to identify and screen potential crack mechanisms Use Table D3 to evaluate the probability of initiation. Some.Table D1. Pbreach-NE = 0. PI – scour Evaluate the probabilities for No. PP = PPR x PPC x PPL Estimate the probability for not detect and intervene using Table D7. Pfail = PI –scour x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] 3 1 3 Erosion of core by water flowing in open rock defects Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes PI -scour Yes Erosion by Scour at the Core-Foundation Contact PP Intervention Fails No Some Erosion Continuation No Erosion PSE Progression No Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure using the event tree. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table D6. PSE PEE PCE Estimate the probabilities for forming a roof (PPR).

PI-crack Evaluate the probabilities for No. Pbreach-NE = 0. Some. Pbreach-SE Pbreach-EE Pbreach-CE Calculate the probability of failure using the event tree. Pfail = PI -crack x PP x Pudi x [(PSE x Pbreach-SE) + (PEE x Pbreach-EE) + (PCE x Pbreach-CE)] 3 3 1 1 Hydraulic fracture in cut-off trench Pudi Yes Pbreach-CE Pfail-CE EVENT TREE STRUCTURE Yes Continuing Erosion PCE Progression No PP Yes Pudi Breach No Intervention Fails No Yes Yes Pudi Breach No Pbreach-EE Pfail-EE Yes Excessive Erosion PEE Progression No PP Intervention Fails No Yes Yes Pudi Breach No Pbreach-SE Pfail-SE Yes PI -crack Yes Erosion by in a Crack or Hydraulic Fracture Across the Cut-Off Trench PP Intervention Fails No Some Erosion Continuation No Erosion PSE Progression No Σ Pfail Initiation No A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . PP = PPR x PPC x PPL Estimate the probability for not detect and intervene using Table D7.Guidance Document July 2008 Delta Version D-3 . Estimate the probabilities of breach for the Some. Excessive and Continuing Erosion for the failure path under consideration using Table D5. Excessive and Continuing Erosion branches using Table D8.Table D1. PSE PEE PCE Estimate the probabilities for forming a roof (PPR). Probability of Failure by Internal Erosion of the Embankment at or into the Foundation (Sheet 3) Initiation of Erosion in a Crack or Hydraulic Fracture Across the Cut-Off Trench Use Table D4 to identify and screen potential crack mechanisms Use Table D4 to evaluate the probability of initiation of the embankment at or into a rock foundation. crack filling action not stopping pipe enlargement (PPC) and upstream zone fails to limit flows (PPL) for the failure path under consideration using Table D6.

use Table 6. Exclude if the reservoir level rise is insufficient for seepage gradient to develop. If Cu>6. assume they are present (refer to Section 9.2. (4) Estimate probability of erosion (PIP) (5) Calculate the Probability Initiation of Erosion (PI) Erosion into Open Joints in Rock Estimate the probability of a continuous pathway of open joints in rock in the base or sides of the core trench or core-foundation contact. Ppath Same as above Same as above Same as above Same as above A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . Probability of Initiation of Erosion by Backward Erosion from a high permeability zone in the core or cut-off trench into the foundation Probability of Erosion in the High Permeability Zone (cohesionless soils) Failure Path/Location (1) Assess the Probability of a Continuous Pathway Ppath (2) Assess whether there is time for seepage gradient to develop (3) Estimate average seepage gradient required to initiate and progress backward erosion (iPMT or iCR) Estimate the average gradient (iPMT) required to initiate and progress erosion from Figure 6. C3. C4 and C5).1). = Ppath x PP x PIP compacted layers and Table 6. (c) Calculate the probability of a continuous pathway for each defect size. For defects <5mm.4. Ppath = PCR x PTI For reservoir levels above the normal operating pool level.6.Table D2. Obtain probability of erosion Cohesionless soils: from Table 6. PIP Erosion into Coarse Grained Soil Estimate the probability of a continuous pathway of coarse grained soil in the base or sides of the core trench or core-foundation contact using Table 9.3. (b) Assess the probability the foundation treatment fails to prevent contact of the core with open defects or solution features (PTI) using Table 9.26 This is done for each defect size for poorly compacted layers based on the average seepage gradient across the embankment core at the core-foundation contact (iave) and iPMT (or iCR).Guidance Document July 2008 Delta Version D-4 .25 for well PI-BEP.1.25 to estimate time for seepage gradient to develop in layer. (a) Estimate the probability of a continuous pathway of open defects or solution features of different sizes >5mm from Section 8 (Tables C2. estimate the critical gradient (iCR) – see Section 6.2.

Do this for each PI-scour = Ppath x PIC defect size (i. The equivalent crack width can be assumed to be equal to the D15/4 value of the coarse grained soil.Table D3.2 for guidance) PIC Estimate the probability of erosion from the most appropriate table from Tables 5. The hydraulic gradient should be based on the seepage gradient on the core-foundation contact (refer to Section 9. 5-25mm. 25-100mm. Probability of Initiation of Erosion by Scour at the Core-Foundation Contact Failure Path/Location (1) Assess the Probability of a Continuous Pathway Ppath Same as for Backward Erosion (Step 1 of Table D2) Ppath (2) Estimate probability of erosion (PIC) (3) Calculate the Probability of Initiation of Erosion (PI) Scour Erosion due to Open Joints in Rock Estimate the probability of erosion from the most appropriate For each defect size. calculate the probability table from Tables 5.29 to 5. The hydraulic gradient should be based on the seepage gradient on the core-foundation contact PIC Same as above Scour Erosion due to Coarse Grained Soil Same as for Backward Erosion (Step 1 of Table D2) Ppath A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .Guidance Document July 2008 Delta Version D-5 .29 to 5. >100mm).35 based on the core soil type and of initiation by scour.6. using the defect size as the “crack width”.35 based on the core soil type and by representing the coarse grained soil as an equivalent crack width.e.

Guidance Document July 2008 Delta Version D-6 .Table D4.29 to 5.4.3 and obtain the probability from Table 9. Probability of Initiation of Erosion in a Crack or Hydraulic Fracture Across the Cut-Off Trench (1) Assess the Probability of a Continuous Pathway Ppath Same as for Backward Erosion (Step 1 of Table D2) Ppath (2) Estimate the probability of a hydraulic fracture occurring across the cut-off trench (Phf) Estimate the probability of a hydraulic fracture occurring across the cut-off trench (Phf). calculate the probability of initiation across the cut-off trench PI-crack = Ppath x Phf x PIC Erosion into Coarse Grained Soil Same as for Backward Erosion (Step 1 of Table D2) Ppath Same as above Same as above A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . (2) Estimate probability of erosion in the hydraulic fracture (PIC) Obtain probability of erosion from the most appropriate table from Tables 5. Assess the weighting score from Table 9. The hydraulic gradient should be based on the estimated seepage gradient across the cut-off trench.35 based on the core soil type and assuming the hydraulic fracture has a crack width of 5mm. PIC Same as above Failure Path/Location Erosion into Open Joints in Rock (3) Calculate the Probability of Initiation of Erosion (PI) For each defect size.

Excessive Erosion (PEE) and Continuing Erosion (PCE). 1A 1 1A ZONED EARTHFILL WITH COHESIVE SHELLS Calculate PNE = 1. PSE .Table D5.11. PCE) No potential for filtering.4 to estimate the probabilities of No Erosion (PNE). Some Erosion.12. PEE . The assessment of erosion into a toe drain considers the observed condition of the toe drain (from video or external inspections) and the design and construction details of the toe drain. Figure A1 (Appendix A) shows a flow chart which summarizes the procedure.10. PCE.1. HOMOGENEOUS EARTHFILL Scenario 2: Downstream shoulder of fine grained cohesive material which is capable of holding a crack/pipe. 1 Probability for continuing erosion. Use Table 10.1.0 PNE = PSE = PEE = 0 Filtering does not occur if the crack/high permeability zone persists through the downstream shoulder zone. EARTHFILL WITH CHIMNEY FILTER ZONED EARTHFILL WITH CHIMNEY FILTER 3 1 3 ZONED EARTHFILL WITH GRANULAR SHELLS Scenario 4: Piping into an open defect. Excessive Erosion (PEE) and Continuing Erosion (PCE) by estimating the proportion of soils falling within each erosion category and using Table 10. SOIL SOIL CONDUIT Step 1: Evaluate the opening size for No Erosion.0 – PCE PSE = PEE = 0 Scenario 3: Filter/transition zone is present downstream of the core or a downstream shoulder zone which is not capable of holding a crack/pipe. Some Erosion (PSE). Probability of Continuation for Internal Erosion of the Embankment at or into the Foundation Assess which Scenario is Applicable to the Failure Path under Consideration Scenario 1: Homogeneous zoning with no fully intercepting filter Examples Probability of Continuation (PNE . PCE=1. INTERNAL EROSION THROUGH THE EMBANKMENT INTO A TOE DRAIN INTERNAL EROSION THROUGH THE FOUNDATION INTO A TOE DRAIN A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .Guidance Document July 2008 Delta Version D-7 . See Section 10. joint or crack. Some Erosion (PSE). OPEN JOINTED ROCK EROSION INTO AN OPEN CRACK OR JOINT IN A CONDUIT OR WALL EROSION INTO OPEN JOINTS IN ROCK FOUNDATION Scenario 5: Erosion into a toe drain 1 1 Estimate the probability of continuing erosion for erosion into a toe drain using Table 10. Step 2: Estimate the conditional probabilities for No Erosion (PNE). Excessive Erosion and Continuing Erosion using Table 10.1 to determine the probability for continuing erosion. 2 1 1 3 1 3 2 Follow the procedure outlined in Section 10.3 for more details.

PPC Probability that Upstream Zone Fails to Limit Flows (PPL) For the case where erosion is occurring at the corefoundation contact.2 and 9. 13.6.2.7. Probability of Unsuccessful Detection and Intervention for Internal Erosion of the Embankment at or into the Foundation Probability of Not Detecting Applicable Failure Path/Location (1) Assess the Rate of Internal Erosion and Piping Probability of Not Intervening (4) Calculate the probability of not detecting the internal erosion Pndi Calculate the probability of not detecting the internal erosion using Pnol from Step (2) and Pnd from Step (3). that it is not detected Pnd Assess the probability of not detecting the leak using Table 13. PPL Table D7. Pnd All Failure Modes Estimate the approximate time for progression of piping and development of a breach using Tables 13.3.0 should be used. Pnol (3) Assess the probability that given the leak is observable. limitation of flows is embedded in the breach assessment and a probability for no flow limitation of 1.1.5 and obtain probability from Table 13.Table D6.Guidance Document July 2008 Delta Version D-8 . For the case where erosion initiates within a deep cut-off trench into open joints in a rock foundation. Pui (6) Probability of Not Detect and Not Intervene Pudi (2) Assess the probability of the concentrated leak not being observable Pnol Assess the weighted score (WS) from Table 13.8. Pnd from Step (3) and Pui from Step (5). assess the probability that the upstream zone or concrete elements fail to limit flows using Table 11. = Pnol + [(1-Pnol) x Pnd] (5) Assess the probability that intervention and repair is not successful Pui Assess the probability that intervention and repair is not successful using Table 13. PPR Probability of Crack Filling Action Not Effective (PPC) Assess the probability of crack filling action not stopping pipe enlargement using Table 11. Probability of Progression for Internal Erosion of the Embankment at or into the Foundation Applicable Failure Path/Location All IE Embankment into Foundation Failure Modes Probability of Forming a Roof (PPR) Assess the probability of the core material to form a roof of a pipe using Table 11.3. Calculate the probability of unsuccessful detection and intervention using Pnol from Step (2). Pudi = Pnol + [(1-Pnol) x Pnd x Pui] A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping .1.

Pbreach-EE= 1 – [(1 .Psd-EE)].Psd-CE)]. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping . estimate the probability of breach by gross enlargement using Table 13. No.2. Some and Excessive Erosion branches.20.3. Excessive Erosion (Ps-lf EE) and Continuing Erosion (Ps-lf CE).12 and obtain the probability for each branch of the tree from Table 13. No Erosion Branch (NE) 0 Some Erosion Branch (SE) Excessive Erosion Branch (EE) Continuing Erosion Branch (CE) Pbreach = 0 for NE Pge-SE = 0 Pge-EE = 0 Pge-CE from Table 13. Some. The probability of breach by gross enlargement Pge= 0 for the No. (Psi-CE) = (Psi-i CE) x (Psi-lf).4 and obtain the probability from Table 13. Excessive and Continuing Erosion branches) Assess which breach mechanisms are relevant to the type of dam zoning and failure path being considered using Table 13.Guidance Document July 2008 Delta Version D-9 .19 and obtain the probability for each branch of the event tree from Table 13. This is done for each branch of the event tree [Some Erosion (Psi-i SE). (Psu-SE) (Psu-EE) (Psu-CE) (Psd-SE) = (Ps-f) x (Ps-lf SE) (Psd-EE) = (Ps-f) x (Ps-lf EE) (Psd-CE) = (Ps-f) x (Ps-lf CE) Pbreach-NE = 0 Pbreach-SE= 1 – [(1 .Psu-EE) x (1 . Excessive Erosion (Psu-EE) and Continuing Erosion (Psu-CE).10 and obtain the probability from Table 13. (b) Estimate the probability that the sinkhole causes loss of freeboard. (Psi-EE) = (Psi-i EE) x (Psi-lf). Assess the weighted score from Table 13. (c) Calculate the probability of breach by slope instability for each branch of the tree. (b) Estimate the probability of loss of freeboard due to instability (Psi-lf). If this breach mechanism is applicable.0.15 and obtain the probability for each branch of the tree from Table 13.Table D8. Assess the weighted score from Table 13.e. Excessive Erosion (Psi-i EE) and Continuing Erosion (Psi-i CE)]. Evaluate whether this is applicable using Table 13.Pge-SE) x (1 Psi-SE) x (1 . Some Erosion (Ps-lf SE).16. Assess the weighted score from Table 13. assess the weighted score from Table 13. (c) Calculate the probability of breach by sinkhole development. Some Erosion (Psu-SE). (c) For dams with a downstream zone of rockfill. Pge (a) Estimate the probability of slope instability occurring due to the increased seepage flows.13.18.Psd-SE)]. Probability of Breach for Internal Erosion of the Embankment at or into the Foundation Applicable Failure Path/Location (1) Screening of breach mechanisms (2) Probability of Breach by Gross Enlargement Pge (3) Probability of Breach by Slope Instability Psi (4) Probability of Breach by Sloughing or Unravelling Psu (5) Probability of Breach by Sinkhole Psd (6) Calculate the Probability of Breach Branch on Event Tree All Failure Modes Evaluate the probabilities of breach for each branch of the event tree (i. Excessive Erosion (Psu-EE) and Continuing Erosion (Psu-CE). (a) Probability seepage will emerge into the downstream shell PS = 1. Some Erosion (Psu-SE).Pge-CE) x (1 Psi-CE) x (1 .Psu-SE) x (1 . Pbreach-CE = 1 – [(1 .3 (Psi-SE) = (Psi-i SE) x (Psi-lf).11. Breach by gross enlargement is negligible in cases where the downstream shell is unable to support a roof of a pipe.Pge-EE) x (1 Psi-EE) x (1 . Only consider the breach mechanisms that are included for estimating the probability of breach.1.5.Psu-CE) x (1 . (a) Estimate the probability of a sinkhole developing as a result of the internal erosion (Ps-f) from Table 13. Excluded breach mechanisms are assigned a probability of breach = 0. Calculate the probability of breach for each branch of the event tree (i.e. Some Erosion. assess the weighted score from Table 13. (b) For dams with a downstream zone of earthfill. Excessive Erosion and Continuing Erosion branches) by summing the probabilities using de Morgan’s rule as follows.

Appendix E Guidance for Failure Paths Not Covered by the Unified Method A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 .

Each branch represents one possible outcome of the event or one possible state that a condition may assume. This may be due to an unusual type or configuration of the embankment dam or its foundations. In addition. E2 E2. starting with some initiator event and proceeding through events describing the response of the dam to each level of the initiator. Version 3.e. and finally to consequences that result.0). the structural response (failure) probability given that the load has occurred. The branches are mutually exclusive if each branch unambiguously describes one and only one possible outcome (i.Appendix E Guidance for Failure Paths Not Covered by the Unified Method E1 Introduction In some cases. all of the branches emanating from a node should represent the mutually exclusive and collectively exhaustive set of possible outcomes or states (this is typically not done in the load range branches). And perhaps most importantly.1 Developing Event Trees Principles Event trees are used to represent sequences or progressions of events that could result in adverse consequences when a dam or associated structure responds to various loading conditions. an event tree becomes the template for subsequent assignment of event probabilities and calculation of risk. there is no “overlap” among them). An event tree consists of a series of linked nodes and branches. Together. it fosters common knowledge and understanding of failure modes. and the magnitude of that consequence. the risk analyst or risk analysis team will need to develop their own event tree structure to portray the postulated sequence of events. probabilities add up to 1. Each node represents an uncertain event or condition. the adverse consequence given that the load and failure have both occurred. This section provides additional guidance for those cases which are not adequately covered by the initiating mechanisms or failure path locations described in Section 3 of the Guidance Document. The event tree is also a tool for evaluating changes in risk given certain actions and assumptions. May 2003. The event tree is constructed from left to right.3. it is a means for identifying where the greatest potential risks are. In these circumstances. The risk associated with one sequence in the event tree is the product of the load probability. By providing a graphical representation of the logic structure for the progression of each failure mode.1. and synergetic discussion of various issues associated with failure modes. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . Each event node is predicated on the occurrence of all directly-linked branches that precede it in the tree. and they are collectively exhaustive if together they describe all possible outcomes (i. it may not be possible to match a potential internal erosion failure path to the generic event tree framework described in Section 3. These event sequences are developed all the way to breach of the dam. The following guidance for developing event trees and estimating structural probabilities has been extracted directly from Reclamation’s “Dam Safety Risk Analysis Methodology”.1.e. The total risk for the load category is the sum of the products for all event tree paths.

These diagrams are less complex than the formally constructed event trees. Often it is useful to begin with “logic diagrams” that generally list the various sequential steps needed to take place during a given failure mode. etc. or drains. E2. This information provides the means for conceptualizing and specifying the occurrences. Case histories can provide additional insight for identifying failure modes and for breaking down the modes into sequences of events. Too much detail. Considerable effort should be devoted to determining atypical failure modes that might be unique to the dam in question. The potential for adverse consequences associated with improper operation of the facilities should be considered as one of these unique failure modes. Techniques for achieving an appropriate level of detail in the event trees include the following: A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . and the event tree becomes unmanageable or incomprehensible to a degree that important insights are lost.Appendix E Guidance for Failure Paths Not Covered by the Unified Method The best way to start creating an event tree is to establish failure modes through a failure mode screening process. and interventions that could be pertinent to the dam under consideration. For many dam types and applicable failure modes. and to some degree on the purpose of the risk analysis. The event tree must balance needs for comprehensiveness and detail against needs for consistency.. and communication. An example of this might be consideration of construction of an alternative(s) that would prevent the continued development of adverse consequences. or a berm. on the complexity of the failure modes considered. clarity. conditions. Failure and incident information provided in case history reports describe the progression and sequence of the events that have occurred for other dams. a process sometimes called “failure mode decomposition”. Incidents that have progressed nearly to failure but have stopped for some reason provide information that is as valuable as information regarding complete failures. on the number of load ranges needed. Once a failure mode has been identified. the event tree should be formulated to show the sequence of events and/or conditions which would have to take place or exist in order for the dam to respond in an adverse manner. For instance . Too little detail can reduce the ability to target specific risk contributors and can create problems in making reasonable structural response probability estimates. The potential failure modes should be identified and each event in the progression should be explicitly and unambiguously documented (such that all team members have a common understanding of the potential failure modes) for later use in the structural response probability estimation phase.2 Complexity The size and complexity of the event tree depend on what is known about the dam and its expected behavior under different loading conditions. The event tree should also identify possible interventions which could terminate the development of the adverse consequence. is considered.have an “intervention” node in an event tree for a seepage related failure mode where the probability of successfully constructing say filters. there are often one or more especially well-documented failure(s) or incident(s) that chart the progression of events in some detail. Successful intervention would terminate one path of the event tree.

Sometimes load ranges are selected to represent information available from related analyses. The first step is to be sure each team member has a clear understanding of each node of the event tree.1 Step 1.Bounds for load increments should be chosen specifically to bracket load ranges where it is expected that the structural response (or the consequences of dam failure) will be fundamentally different from the structure’s response (or the dam failure consequences) in other load ranges.There is no need to propagate event sequences once it becomes apparent that they cannot lead to an uncontrolled release of the reservoir. (An event tree node represents a choice at which the preceding event must be considered to have happened and two or more subsequent events could take place. a node description for “unfiltered exit” might be: A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . While any number of increments can be used. An open discussion usually takes place during this step where team members freely discuss their understandings of the event node and the wording being proposed. but constructing them separately and sequentially better organizes the process. Construct separate event trees for each load type. The facilitator should then capture the thoughts of the group into the description of the node.As with any other engineering analysis. The reasons why an event sequence branch is truncated are an important part of the risk analysis documentation.Appendix E Guidance for Failure Paths Not Covered by the Unified Method • Truncate non-failure branch pathways as early as possible . Limit the number of load increments for initiator events . and sometimes. E3.These trees will often be similar or identical. A comparatively simple initial effort can identify the key elements in the tree that need to be expanded and less important parts that can be pruned in subsequent iterations. For instance. • • • E3 Estimating Structural Response Probabilities Summarized below is a process for making structural response probability estimates that has been found to work well for various risk analyses. All steps described below are performed jointly by all the participants of the risk analysis team.) This is best done by having the facilitator(s) write out the description of the node at the top of a flip chart (or some other visual means that is readily accessible at any time). Use a staged approach . Dividing the full range of possible loading values into a few increments is usually sufficient for most problems. it is unreasonable to expect that everything can be fully captured in an event tree on the first pass through the problem. for each load increment . there must be sufficient reason to suspect that considering different load increments will lead to different structural responses or to some fundamental change in the adverse consequences.

Appendix E Guidance for Failure Paths Not Covered by the Unified Method “the soil particles that are being carried by seepage flow must exit from the dam at a location where there is no filter present to trap the soil. Considerable report writing time can be saved if this chart can be created on a computer as the discussion takes place.” It is perfectly acceptable to further decompose the node in the word description. “2 out of 95 gradation tests of as-constructed earthfill failed the limits and were left in place” [factors leading to a higher probability]. The purpose of this step in the process is to display all the information that will be used in making the estimate for all team members to see and discuss. “the zone 2 for ‘Dam X’ (the case history dam) was much less compatible for the zone 1 than is the dam under study” [factors leading to a lower probability]. For instance. For instance. Also to be listed are any similarities/dissimilarities with the case histories being used as a comparison. A filter is defined as a soil that reasonably meets Reclamation’s design standard for filters. The listing is usually done on the same chart immediately below the node description. The group then ‘brainstorms’ any and all information that is pertinent to the event node being discussed. the team members can judge for themselves the importance of the information being listed as they make their estimates.2 Step 2. “the specified gradation is likely to segregate during placement” [factors leading to a higher probability]. Each piece of information is listed on the flip chart in either a ‘factors leading to a higher probability’ or ‘factors leading to a lower probability’ column depending on whether the information is can be used as evidence to support or oppose belief in the event. Others might be “93 out of 95 gradation tests of as-constructed earthfill showed acceptable limits were achieved” [factors leading to a lower probability]. For instance. one team member might want to list his/her concerns as to the appropriateness of the filter criteria used in the listing of the above information and include this in the ‘factors leading to a higher probability ’ column. a node description as above might also add: “The zone 2 of the embankment must reasonably meet filter criteria for the zone 1. Nearly any type of information is permissible to be listed if it helps the team members make their estimates. An example showing a record of steps 1 and 2 is shown below. Even information of a general nature or member biases can be listed. As described below in step 3. For instance. E3. The zone 3 outer shell must reasonably meet filter criteria for the zone 2". as described. The team should agree that the information is being placed in the correct column. Disagreements are usually solved by using clear wording that describes the information or by adding an opposing view in the opposite column. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . “gradation limits in construction specification meet filter criteria for the zone 1" might be listed in the ‘factors leading to a lower probability ’ column for the ‘unfiltered exit’ description discussed above in step 1. actually happening. The terms ‘factors leading to a higher probability’ and ‘factors leading to a lower probability’ are used in terms of the event node.

Verbal descriptors can be used for assigning response probabilities when there is not a basis (i. If many of the distributions for events in the event tree are skewed like this. does the group feel that all values within the range are equally likely? If there is no single “most reasonable” or “popular value”. under these circumstances the team members can use the subjective information that was generated during step 2 (“factors leading to a higher probability ” versus “factors leading to a lower probability ” exercise) to judge if the event tree node designated “unfiltered exits” is more likely or unlikely relative to the scale of verbal descriptors as shown in Table E1. but the 50th percentile will often be some other value. if the group believes that an erroneous mean value is to be used about which the random simulation should pick values equally distributed.e. Stated another way. these reasons should be stated for the record and a triangular distribution should be used with the peak of the triangle placed at the value which would be expected to occur most often. it may result in the “most popular” estimate calculated for annualized life loss being off-center within the range estimated from the Monte Carlo simulation. The team should obtain “reasonable high” and “reasonable low” probability estimates. then a uniform distribution should be used. Once a clear understanding of what the node of the event tree represents has been established (step 1). An alternative subjective probability mapping scheme developed by Barneich et al (1996) is presented in Table E2. The mode will be the value randomly selected more often than any other during the simulation. then the group might reconsider if a triangular distribution should be used. 2003). However. Elicit a “reasonable low” probability estimate by selecting a trial value and asking “Is it unlikely that the actual probability value is less than this value?” Elicit a “reasonable high” estimate by selecting a trial value and asking “Is it likely that the actual probability is less than this value. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . During the simulation. This is not a technical problem.B of the Dam Safety Risk Analysis Methodology (Reclamation. then a probability estimate may be made for the node of interest. The mean of the triangular distribution is often not the same as the mode. but it may be difficult to communicate the reasons to those not well versed in probability and statistics. For the uniform distribution this should not be a problem.Appendix E Guidance for Failure Paths Not Covered by the Unified Method E3. The team should be told how the distributions will be used in the Monte Carlo analysis. If there are reasons to suspect one value is more likely. and all relevant issues by team members related to that node have been aired and summarized (step 2). For example. appropriate statistical information) for use of what can be termed the “known” failure frequency rate method. Determine if the group feels that any given value within the established range should be more likely than any other.3 Step 3. The expected value for the both the uniform distribution and the triangular distribution will be the mean value of all the random selections for each variable during the simulation. Related discussions on establishing estimate distributions are provided in Section V. values will be equally distributed about the mean.

999 0. several potential failure scenarios can be identified. a single scenario could be identified after considerable effort.Appendix E Guidance for Failure Paths Not Covered by the Unified Method Table E1 – Probability Mapping Scheme (Bureau of Reclamation. or is observed in one isolated instance. however.001 Table E2 – Mapping scheme linking description of likelihood to quantitative probability (adapted from Barneich et al 1996) Order of Magnitude of Probability Assigned 1 10-1 10-2 Description of Condition or Event Occurrence is virtually certain Occurrence of the condition or event are observed in the available database The occurrence of the condition or event is not observed. the team members might assign a verbal descriptor of “very unlikely” (probability of 0.99 0.9 0. The occurrence of the condition or event is not observed in the available database.5 0. in the available database.01) to the node described as “unfiltered exit” in step 1 above based on the available information: “93 of 95 gradation tests of as-constructed zone 3 earthfill materials generally met Reclamation filter criteria for the zone 2 earthfill material where seepage might exit” “Zone 3 earthfill materials are such that they are not likely to separate and segregate during placement” A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . even after considerable effort. and no plausible scenario could be identified. 10-3 10-4 In the example being used. It is difficult to think about any plausible failure scenario. The condition or event has not been observed. 2003) Verbal Descriptors Virtually Certain Very Likely Likely Neutral Unlikely Very Unlikely Virtually Impossible Descriptor Probability 0.1 0.01 0.

or about 25 percent. the probability that this event will occur in n independent trials. The probability of getting two sixes each time the dice are thrown is 1/36. is given by the following equation: pn = 1 . pn. p is the probability that the initiating event will trigger some other event. An example of this method for estimating a response probability for a node described as “unfiltered exit” might be: “Reclamation has about 150 dams that have clay tile drains” “22 of these clay tile drain systems have been shown to have defects or crushed zones that compromise the integrity of the drain” “While none of these 22 compromised clay tile drain systems have lead to failure of a Reclamation structure. and each time the event occurs there is the same probability that this event will trigger some other event.1/36)10. i.04 (6/150) for an “unfiltered exit” related to Reclamation dams with clay tile drain systems. Any available statistical information of this nature should be presented in establishing the likely ranges for the probability estimate. One situation is where a potential initiating event takes place many times over the life of a dam. It is appropriate to consider this equation in two situations where structural response probabilities are being estimated. Another useful way to incorporate performance based probability assessments is to consider certain repeated events or multiple examples of an identical condition as repeated Bernoulli trials. The probability of getting the two sixes at least once in 10 throws is 1 . there have been 6 incidences where material was piped through the compromised portions of the clay tile drain system. and n is the number of times the initiating event has occurred. The statistical information presented here for drains and piping incidents is only hypothetical. In this situation. If a random event has a probability of occurrence of p. and if this A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . Another situation is where many dams have the same component.e.(1 . Clark Canyon Dam” Based on the outlined information. but this type of information could be gathered in many cases to help make probability estimates. one could assign an estimated response probability of 0.(1 .p)n An example would be a pair of fair dice thrown 10 times.Appendix E Guidance for Failure Paths Not Covered by the Unified Method “As-built drawings indicate that zone 2 and zone 3 earthfill materials were placed to the lines and grades specified” Estimates of response probabilities can sometimes be made on a more quantitative basis by comparing known historical or statistical databases that are relevant to the node for which a response probability is being estimated..

the facilitator(s) should focus more on getting agreement on the possible range and characteristic probability distribution for the estimate (see Section V). One way this can be used is to check the reasonableness of a probability estimate.3 that material movement would begin should the reservoir reach elevation 5340 in any given year. the team should identify those items on the flip chart which were most important in arriving at the probability estimates. This information and discussion should be documented by the recorder. the divergent opinions must be accounted for in the analysis. In addition. the facilitator(s) could use the verbal descriptors by suggesting: “I sense the group feels this event is not very likely. E3. the team should indicate why it believes the most significant factors should receive more weight than others. n is the number of dams and p is the probability the condition will cause the other event to happen. “I sense there are more reasons to believe we are on the likely rather than the unlikely side of being neutral. The facilitator(s) should lead the discussion between the protagonists of the opposing views and identify the underlying premises or key evidence supporting each argument. The facilitator(s) may use words like “I’m sensing the group feels fairly neutral about this estimate. In this situation.3 probability estimate would seem unreasonable (unless other factors could be placed in the “factors leading to a higher probability” evidence column). This is a very fruitful area to obtain ideas that would suggest further exploration or analysis to resolve A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . there is a certain probability it will cause some other event or condition to happen. At this point. how about 0. Since the reservoir has been above that elevation fifteen times and no material has been observed. Rather.” If the discussion indicates the event is not very probable. suggest a reasonable starting place as a best estimate. and that no soil materials have appeared in seepage collection weirs during that time period. Returning to the flip chart containing the factors pertinent to the event.53 percent chance) that material movement should begin if the reservoir rises above 5340 fifteen times. E3.5?” Or. The above equation says it is nearly certain (a 99. This can include a discussion of what adverse situations actually exist versus what adverse situations only have the potential to occur. and canvass the group’s willingness to accept the estimate. This does not mean that the facilitator(s) must force all members to accept a single estimate. The risk analysis participants then identify the factors from step 2 that had the greatest effect on the probability estimate generated in step 3. the facilitator(s) must sense the group’s feeling as discussion takes place. The facilitator(s) should ensure the risk analysis participants have reached consensus on the probability and uncertainty estimates. While this process may result in debate among the participants. the team members estimate the probability is . Assume that when the team is considering piping. Assume a given reservoir has reached elevation 5340 fifteen times in the last forty years.Appendix E Guidance for Failure Paths Not Covered by the Unified Method component is present. this discussion can bring out additional information which was not previously available or readily understood. should this be very unlikely or virtually impossible?” If the group cannot agree on an estimate. the .5 Step 5.4 Step 4.

The use of the software “Precision Tree” and “@Risk” makes it very easy to carry a range or different distributions through the risk analysis calculations. The separate calculations for risk would then be reported along with the descriptions of the conflicting ways the group members saw the problem. A Unified Method for Estimating Probabilities of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document Delta Version July 2008 . If the group cannot agree that a range or distribution will adequately characterize their judgement. the process continues by repeating steps 1 through 5 for each remaining node of the event tree. Once consensus is reached on the specific response probability estimate and uncertainty. then the analysis can be conducted using each representative estimate in separate calculations.Appendix E Guidance for Failure Paths Not Covered by the Unified Method the differences.6 Step 6. and to examine “what if” scenarios to determine how a given piece of information might affect the outcome. E3.