PERFORMANCE EVALUATION OF SELECTIVE CONTROL MEASURES OF FOUNDATION SEEPAGE FOR EMBANKMENT DAMS OVER PERMEABLE STRATA
Submitted by
ZAHEER MUHAMMAD MALIK (2005-Ph.D-CEWRE-02)
FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY IN WATER RESOURCES ENGINEERING
CENTRE OF EXCELLENCE IN WATER RESOURCES ENGINEERING
University of Engineering and Technology LAHORE, PAKISTAN
(2012)
i PERFORMANCE EVALUATION OF SELECTIVE CONTROL MEASURES OF FOUNDATION SEEPAGE FOR EMBANKMENT DAMS OVER PERMEABLE STRATA
By:
Zaheer Muhammad Malik (2005-Ph.D-CEWRE-02)
A thesis submitted in partial fulfilment of the requirements for the Degree of
DOCTOR OF PHILOSOPHY
IN
WATER RESOURCES ENGINEERING
Thesis Examination Date:
______Prof. Dr. Ata-ur-Rehman Tariq Dr. Tahir Masood, Research Advisor and External Examiner / Project Engineer Internal Examiner Berkeley Associates Consulting Engineers, Lahore
______DIRECTOR
Thesis submitted on:
CENTRE OF EXCELLENCE IN WATER RESOURCES ENGINEERING University of Engineering and Technology, Lahore, Pakistan
2012
ii This thesis was evaluated by the following Examiners:
External Examiners:
From Abroad: 1) Dr. Krishna R. Reddy, Professor of Civil and Environmental Engineering, and Director, Geotechnical and Geoenvironmental Engineering Laboratory, University of Illinois at Chicago 842 West Taylor Street, Chicago, Illinois, USA. E-mail: [email protected], 2) Dr. Deanna S. Durnford, Professor Emeritus, Civil and Environmental Engineering Department, Colorado State University, Fort Collins, CO, USA. E-mail: [email protected], [email protected]
From Pakistan: Dr. Tahir Masood, Project Engineer, Berkeley Associates Consulting Engineers, Lahore. E-mail: [email protected]
Internal Examiner: Prof. Dr. Ata-ur-Rehman Tariq, Professor, Center of Excellence in Water Resources Engineering, University of Engineering & Technology, Lahore. E-Mail: [email protected]
iii
I dedicate this work to my parents Their prayers, love and encouragement enabled me to do it all
iv
ABSTRACT
This research has addressed typical geological complexities through performance evaluation of selective foundation seepage control features for embankment dams on deep permeable strata. The study is based on an intricate case history of Satpara Dam Project founded over moraines. Seepage control measures for
Satpara dam foundations included upstream extension of the dam core as an impervious blanket with a partial cutoff at its upstream end.
Seepage modelling was carried for ‘as-designed’ seepage mitigation measures under 2-D sectional flow model assumptions for different case scenarios using
SEEP/W. In-situ permeability test results from pre-construction exploratory investigations of the project, ranging over several logarithmic orders of magnitude, were scrutinised for identification of a representative dataset for assignment of K- values while characterising spatial variability. Relative seepage sensitivity analyses were made using ‘percentile’ values from the two-layered ‘pervious zone in homogeneous’ foundation representation approach and computed values from the multi-zoned ‘ROCKWORKS modelled’ multi-zoned foundation representation approach. Comparative inferences considered theoretically acceptable limits in averting piping initiation for the adopted seepage control measures. The refinement in
K-assumptions from the layered to the multi-zoned approach influenced and enhanced scale of magnitudes and rate of change of computed gradients and hydraulic heads along their distributions and trends at different points of consideration along the flow direction. The adopted seepage control scheme reduced the computed hydraulic gradients to 73.5% and the head potential to 72% at toe of core relative to no seepage control measures.
v
Available project data from five years of consecutive operation (for five impoundings during 2007-2011) showed that conservation level and consequently a steady state condition was not achieved. Effectiveness of the adopted seepage control measures at Satpara Dam was evaluated through comparative response of instruments installed across the cutoff wall, for upstream and downstream observations at two different depth zones, and along the flow path in the foundations upstream and downstream of the main dam axis. Piezometeric data from 27 selected responsive foundation piezometers covering the project area was used to develop percent potential contour plot plans, corresponding to selected pseudo-steady state reservoir levels and for a reservoir level common to the last three impounding stages.
Evaluation of head differentials and percent potential distributions also provided evidence of cross flow influences, indicating 3-D flow in the dam foundation domain.
Pseudo-steady state reservoir levels defined new loading conditions, which were used to extend the 2-D ‘as designed’ SEEP/W model and additionally simulate a
3-D seepage model using FEFLOW. Simulated results based on the multi-zoned pre- construction foundation representation could not replicate the observed ‘pseudo- steady’ foundation response. It was assessed that applied heads had likely caused redistribution of unsupported fines, at probable locations associated with higher potential drops and concentrated potential contours. The cutoff wall proximity and foundations underlying the drainage blanket were accordingly identified to induce higher gradients susceptible for subsequently dislodging fines. An inverse 2-D modelling approach was implied to correspondingly re-adjust K-values in the sectional multi-zoned pre-construction foundation representation. This helped improve related subsurface perceptions of post-construction foundation behaviour.
vi
ACKNOWLEDGEMENTS
I feel privileged to thank my research advisor, Professor Dr. Ata-ur-Rehman
Tariq, for his guidance and mentorship with this dissertation and during my doctoral studies.
I would also like to thank Prof. Dr. Muhammad Latif, Director Centre of
Excellence in Water Resources Engineering (CEWRE), University of Engineering and Technology (UET), Lahore, on providing excellent research facilities.
I specially acknowledge the DHI-WASY team for efficient technical support and providing me the opportunity to use their 3-D software package FEFLOW for this research.
I would like to thank Dr. Muhammad Tausif Bhatti (CEWRE) particularly for his help in officially acquiring FEFLOW under critical time constraints, and all other faculty members and staff of CEWRE for their continued help and cooperation.
My professional association with Satpara Dam Consultants (SDC), and the lead firm Pakistan Engineering Services (Pvt.) Ltd. (PES) provided the main progress and data source for this project. The patronage and bearing of Mr. Jamil Anwer, Chief
Executive PES and Mr. Sohail Anwer, Director PES, throughout this work and for allowing me to use the facilities of the company for data collection and frequent site visits is highly acknowledged. I am also indebted to the organization and all my fellow colleagues and staff members both at the head office in Lahore and at the field office in Skardu, for dependable support in a number of ways.
vii
This thesis would not have been possible without the explicit permission from
Brig. (R) Muhammad Zareen, Authority Advisor Projects (Northern Area), on behalf of Water and Power Development Authority (WAPDA), Pakistan, to use the project data and information for this research.
My profound gratitude goes to my parents for their prayers, belief, encouragement and motivation throughout this time. I appreciate my wife for her understanding and support throughout this endeavour. In the end, I must thank Rehan, my little son who had to bear with my mood shifts while studying and writing during evenings and weekends.
Zaheer M. Malik.
viii
TABLE OF CONTENTS
Chapter Description Page No.
ABSTRACT ...... v
ACKNOWLEDGEMENTS ...... vii
TABLE OF CONTENTS ...... ix
LIST OF TABLES ...... xiv
LIST OF FIGURES ...... xvi
LIST OF ABBREVIATIONS ...... xxiii
1. INTRODUCTION...... 1
1.1 GENERAL ...... 1
1.2 CURRENT TRENDS IN PAKISTAN ...... 3
1.3 PROBLEM STATEMENT ...... 5
1.4 STUDY OBJECTIVES ...... 8
1.5 SCOPE OF THE STUDY ...... 8
2. REVIEW OF LITERATURE ...... 10
2.1 GENERAL ...... 10
2.2 THEORETICAL CONCEPTS ...... 10
2.2.1 Risk Analysis ...... 10
2.2.2 Remediation of Seepage Problems ...... 11
2.2.3 Foundation Heterogeneity ...... 13
2.2.4 Uncertainities in Analyses ...... 16
2.2.5 Appraisal of Piping ...... 17
2.2.6 Typical Cutoff Considerations ...... 20
2.2.7 Typical Blanket Considerations ...... 23
ix
Table of Contents (Continued) Chapter Description Page No.
2.3 ASSOCIATED EXPERIENCES ...... 25
2.3.1 Upstream Impervious Blanket ...... 25
2.3.2 Cutoff Wall ...... 29
2.3.3 Combination of Blanket and Cutoff Wall ...... 32
2.3.4 Long Term Performance ...... 34
2.4 SUMMARY ...... 40
3. DESCRIPTION OF CASE STUDY ...... 42
3.1 PROJECT DESCRIPTION ...... 42
3.1.1 General ...... 42
3.1.2 History of Development and Present Conception ...... 42
3.2 PHYSICAL INVESTIGATIONS ...... 47
3.2.1 Exploratory Efforts ...... 47
3.2.2 Surface Geological Mapping ...... 49
3.2.3 In-situ Testing ...... 52
3.3 DETAILED SUBSURFACE EVALUATION ...... 53
3.3.1 Phase-I (Feasibility Stage - 1988) ...... 53
3.3.2 Phase-II (Design/Tender Stage - 2003) ...... 54
3.3.3 Phase-III (Construction Stage – 2004) ...... 58
3.4 DESIGN AND MONITORING PRIOR TO STUDY ...... 60
4. METHODOLOGY ...... 62
4.1 INTRODUCTION ...... 62
4.2 DATA COLLECTION ...... 62
x
Table of Contents (Continued) Chapter Description Page No.
4.3 ASSESSMENT OF FOUNDATION PERMEABILITY ...... 63
4.3.1 Foundation Permeability Profiles ...... 63
4.3.2 Foundation K-Modelling ...... 64
4.4 MULTIVARIATE SENSITIVITY ANALYSES ...... 68
4.4.1 2D Seepage Modelling ...... 68
4.4.2 Foundation Seepage Control Scenarios ...... 74
4.4.3 Foundation Representations ...... 76
4.4.4 Sensitivity Criteria ...... 78
4.5 INSTRUMENTED OBSERVATIONS ...... 80
4.5.1 Project Instrumentation ...... 80
4.5.2 Impounding Status and Data Acquisition ...... 89
4.5.3 Data Processing and Graphical Plots ...... 90
4.6 POST-CONSTRUCTION SEEPAGE MODELLING ...... 95
4.6.1 Comparative 3-D Seepage Modelling ...... 95
4.6.2 Post-Construction Loading Conditions ...... 105
4.7 UNCERTAINITY PROPAGATION ANALYSES ...... 107
4.7.1 General Analyses Procedure ...... 107
4.7.2 Statistical Evaluation Parameters ...... 109
5. RESULTS AND DISCUSSIONS ...... 112
5.1 INFERENCES ON SUBSURFACE CONDITIONS ...... 112
5.1.1 Summary Findings from Exploratory Efforts ...... 112
5.1.2 Geological Evolution of the Depositional Environment ...... 116
xi
Table of Contents (Continued) Chapter Description Page No.
5.2 ASSESSMENT OF FOUNDATION PERMEABILITY ...... 118
5.2.1 Initial Inferences ...... 118
5.2.2 Statistical Inferences ...... 119
5.2.3 K-Value Estimates for Modeling Dam Foundations ...... 126
5.3 MULTIVARIATE SENSITIVITY ANALYSES ...... 129
5.3.1 General ...... 129
5.3.2 Upstream Impervious Blanket versus Partial Cutoff Wall ...... 129
5.3.3 Upstream Cutoff Wall versus Downstream Cutoff Wall ...... 134
5.3.4 Selected Seepage Control Combination ...... 138
5.4 PERFORMANCE EVALUATION ...... 141
5.4.1 Impounding ...... 141
5.4.2 Seepage Observations ...... 143
5.4.3 Piezometeric Observations ...... 146
5.4.4 Cutoff Wall and Corresponding Piezometeric Data ...... 146
5.4.5 Upstream Impervious Blanket and Corresponding Piezometeric Data ...165
5.4.6 Piezometeric Potential Plots for Adopted Seepage Control Combination
Measures ...... 180
5.4.7 Inferences from Performance Evaluation ...... 191
5.5 UNCERTAINTY PROPAGATION ANALYSES ...... 193
5.5.1 Basic Hypothesis ...... 193
5.5.2 Conditions for Comparative Analysis ...... 196
5.5.3 Comparative modelled results...... 211
5.5.4 “Uncertainty Propagation” Inferences ...... 217
xii
Table of Contents (Continued) Chapter Description Page No.
6. CONCLUSIONS AND RECOMMENDATIONS ...... 219
6.1 GENERAL ...... 219
6.2 CONCLUSIONS...... 219
6.3 RECOMMENDATIONS ...... 220
REFERENCES ...... 221
ANNEXURE - I: SUBSURFACE LOG REPORTS ...... 231
ANNEXURE - II: SEEP/W INPUT FUNCTIONS ...... 251
VITA...... 255
xiii
LIST OF TABLES
Table Description Page No.
4.1 Selected functions from material database file ...... 71
4.2 Individual material WC functions and Ksat values for each K-Fn ...... 72
4.3 Tabulated list of modeled case scenarios ...... 74
4.4 Vibrating Wire Piezometer Details ...... 84
4.5 Standpipe Piezometer Details ...... 86
4.6 Seepage Measurement Station Details ...... 88
5.1 Phase wise summary - Logs ...... 113
5.2 Descriptive Statistics for the permeability data set ...... 121
5.3 Quantitative description of permeability data ...... 125
5.4 Computed Gradients and Total Heads at Selected Points for Case
Scenarios of Impervious Upstream Blanket ...... 130
5.5 Computed Gradients and Total Heads at Selected Points for Case
Scenarios of Partial Cutoff Wall under Core ...... 132
5.6 Computed Gradients and Total Heads at Selected Points for
Combination Case Scenarios of a Partial Cutoff Wall with 3 m Thick
and Full Length Impervious Upstream Blanket ...... 136
5.7 Computed Percent Differences for Gradients and Head Potentials
Corresponding to Rockworks Modelled (Multi-Zoned) Foundation
Representation for Relative Effectiveness of the Selected Seepage
Control Measures ...... 139
5.8 Piezometeric data corresponding to cutoff wall instrumented sections ...150
5.9 Average percent piezometeric potential drops across cutoff wall
instrumented sections ...... 150
xiv
List of Tables (Continued) Table Description Page No.
5.10 Piezometeric data corresponding to pseudo-steady reservoir levels
from last three partial impounding stages for selected instruments
installed in the foundation underlying upstream impervious blanket ...... 176
5.11 Model evaluation statistics for resultant total heads ...... 212
xv
LIST OF FIGURES
Figure Description Page No.
3.1 Project location map ...... 44
3.2 Project layout plan ...... 45
3.3 Main project features ...... 46
3.4 Subsurface investigation plan – Phase wise distribution ...... 48
4.1 Typical K-D (permeability vs. depth) plot output ...... 64
4.2 Dialog Box “P-Data Section Options” ...... 66
4.3 Typical outputs from Rockworks in RockPlot2D ...... 67
4.4 Typical model domain setup for SEEP/W sectional, i.e. 2-D seepage
analysis ...... 70
4.5 Typical SEEP/W resultant outputs showing equipotential total head
contours and flow paths ...... 73
4.6 Typical ‘Node Information’ dialog box in SEEP/W Contour window ...... 73
4.7 Typical “Define” window for PZ foundation approximation showing
the layered hydraulic conductivity K assignment ...... 79
4.8 Typical “Define” window for RW foundation approximation
showing the multi-zoned hydraulic conductivity K assignment ...... 79
4.9 Pertinent case scenarios and sensitivity criteria for the multivariate
sensitivity analysis ...... 81
4.10 Instrumentation layout plan ...... 83
4.11 Seepage Monitoring Stations (SMS) ...... 88
4.12 Typical Time Plot ...... 93
4.13 Typical Head Plot ...... 93
4.14 Percent Potential Contour Plot Plan ...... 94
xvi
List of Figures (Continued) Figure Description Page No.
4.15 Model domain described as ‘Supermesh’ components in FEFLOW...... 96
4.16 Finite element mesh in FEFLOW...... 96
4.17 3D Layer Configuration dialog in FEFLOW...... 97
4.18 Typical “.DAT” format extension table file for importing K-values in
FEFLOW...... 98
4.19 Typical Parameter Association dialog linking K-values in FEFLOW. ...100
4.20 Typical layered cross-sectional profiles extracted through
RockWorks ...... 101
4.21 Typical manually developed layer map on project grid for Layer 2...... 101
4.22 K-data assignment for the cutoff wall (blowup) within the FEFLOW
K-map for Layer 2...... 102
4.23 FEFLOW Snapshot View for 3D Mesh and Digital Elevations Model ...103
4.24 Typical assigned boundary conditions in FEFLOW...... 104
4.25 Problem Settings dialog in FEFLOW for “phreatic” mode
assignment to the top slice, i.e. Slice 1...... 104
4.26 Typical resultant head distribution for the top slice (FEFLOW
Snapshot View from 3D model)...... 105
4.27 Layout plan for orientation of section X-X and section K-K ...... 109
5.1 Permeability distribution and the corresponding investigative drilling ...115
5.2 Phasewise distribution of permeability test attempts ...... 120
5.3 Distribution of permeability with reference to Nullah Proximity ...... 120
5.4 Measures of central tendency for the permeability dataset ...... 122
5.5 Representative permeability data set ...... 123
5.6 Frequency Histogram for the permeability test results ...... 124
xvii
List of Figures (Continued) Figure Description Page No.
5.7 Box plots for representative permeability data set ...... 126
5.8 Foundation profiles for SCM sensitivity analyses ...... 128
5.9 Computed Gradients and Total Heads at Selected Points for Case
Scenarios of Upstream Impervious Blanket ...... 130
5.10 Computed Gradients and Total Heads at Selected Points for Case
Scenarios of Partial Cutoff Wall under Core ...... 132
5.11 Computed Gradients and Total Heads at Selected Points for
Combination Case Scenarios of a Partial Cutoff Wall at the upstream
end of a 3 m Thick and Full Length Impervious Upstream Blanket ...... 137
5.12 Computed Gradients and Total Heads at Selected Points for
Combination Case Scenarios of a Partial Cutoff Wall at the
downstream end (Repositioned under Core) of a 3 m Thick and Full
Length Impervious Upstream Blanket ...... 137
5.13 Impounding Status of Satpara Dam ...... 142
5.14 Measured Seepage Record ...... 144
5.15 Location of cutoff wall piezometers ...... 149
5.16 Time plots for upper level piezometers installed in depth zone 1
(>2630 m) in cut-off wall instrumented section K-K ...... 151
5.17 Response plots for upper level piezometers installed in depth zone 1
(>2630 m) in cut-off wall instrumented section K-K ...... 151
5.18 Time plots for lower level piezometers installed in depth zone 2
(<2630 m) in cut-off wall instrumented section K-K ...... 152
5.19 Response plots for lower level piezometers installed in depth zone 2
(<2630 m) in cut-off wall instrumented section K-K ...... 152
xviii
List of Figures (Continued) Figure Description Page No.
5.20 Time plots for upper level piezometers installed in depth zone 1
(>2630 m) in cut-off wall instrumented section I-I ...... 153
5.21 Response plots for upper level piezometers installed in depth zone 1
(>2630 m) in cut-off wall instrumented section I-I ...... 153
5.22 Time plots for lower level piezometers installed in depth zone 2
(<2630 m) in cut-off wall instrumented section I-I ...... 154
5.23 Response plots for lower level piezometers installed in depth zone 1
(<2630 m) in cut-off wall instrumented section I-I ...... 154
5.24 Time plots for upper level piezometers installed in depth zone 1
(>2630 m) in cut-off wall instrumented section H-H ...... 155
5.25 Response plots for upper level piezometers installed in depth zone 1
(>2630 m) in cut-off wall instrumented section H-H ...... 155
5.26 Time plots for lower level piezometers installed in depth zone 2
(<2630 m) in cut-off wall instrumented section H-H ...... 156
5.27 Response plots for lower level piezometers installed in depth zone 2
(<2630 m) in cut-off wall instrumented section H-H ...... 156
5.28 Time plots for all upper level piezometers installed in depth zone 1
(>2630 m) in cut-off wall instrumented sections K-K, I-I and H-H ...... 157
5.29 Response plots for all upper level piezometers installed in depth zone
1 (>2630 m) in cut-off wall instrumented sections K-K, I-I and H-H .....157
5.30 Time plots for all lower level piezometers installed in depth zone 2
(<2630 m) in cut-off wall instrumented sections K-K, I-I and H-H ...... 158
5.31 Response plots for all lower level piezometers installed in depth zone
2 (<2630 m) in cut-off wall instrumented sections K-K, I-I and H-H .....158
xix
List of Figures (Continued) Figure Description Page No.
5.32 Cutoff wall longitudinal profile and typical cross-section ...... 165
5.33 Layout of selected foundation piezometers for performance
evaluation of impervious upstream blanket ...... 167
5.34 Time and response piezometeric data plots for VWP-49 ...... 168
5.35 Time and response piezometeric data plots for VWP-40 ...... 169
5.36 Time and response piezometeric data plots for VWP-17 ...... 170
5.37 Time and response piezometeric data plots for VWP-21 ...... 171
5.38 Time and response piezometeric data plots for SP-11 ...... 172
5.39 Time and response piezometeric data plots for SP-13 ...... 173
5.40 Comparative time plots of selected foundation piezometers for
performance evaluation of the upstream impervious blanket ...... 174
5.41 Comparative response plots of selected foundation piezometers for
performance evaluation of the upstream impervious blanket ...... 174
5.42 Comparative average potential drops at pseudo-steady reservoir
levels from last three partial impounding stages along preferential
flow path across main dam axis ...... 177
5.43 Comparative average potential drops at pseudo-steady reservoir
levels from last three partial impounding stages for identifying
constricted valley effects ...... 178
5.44 Layout of foundation piezometers for performance evaluation ...... 181
5.45 Individual percent potential contour plot plans for the last three
partial impounding stages ...... 182
5.46 Comparative percent potential contour plot plans for the last three
partial impounding stages ...... 183
xx
List of Figures (Continued) Figure Description Page No.
5.47 Graphical percent potential plots for instruments installed along cross
valley profiles corresponding to the last three partial impounding
stages ...... 189
5.48 Permeability Assignment Plan for Slice 1 ...... 197
5.49 Permeability Assignment Plan for Slice 2 ...... 197
5.50 Permeability Assignment Plan for Slice 3 ...... 198
5.51 Permeability Assignment Plan for Slice 4 ...... 198
5.52 Permeability Assignment Plan for Slice 5 ...... 199
5.53 Permeability Assignment Plan for Slice 6 ...... 199
5.54 Permeability assignment plans for all slices (top left – slice 1 to
bottom right – slice 6) from 3D FEFLOW model...... 200
5.55 Head Distributions Plots for Slice 1 at Stage 4 Pseudo-Steady
Reservoir Level i.e. El. 2656 m.s.m.s.l...... 204
5.56 Head Distributions Plots for Slice 2 at Stage 4 Pseudo-Steady
Reservoir Level i.e. El. 2656 m.s.m.s.l...... 204
5.57 Head Distributions Plots for Slice 3 at Stage 4 Pseudo-Steady
Reservoir Level i.e. El. 2656 m.s.m.s.l...... 205
5.58 Head Distributions Plots for Slice 4 at Stage 4 Pseudo-Steady
Reservoir Level i.e. El. 2656 m.s.m.s.l...... 205
5.59 Head Distributions Plots for Slice 5 at Stage 4 Pseudo-Steady
Reservoir Level i.e. El. 2656 m.s.m.s.l...... 206
5.60 Head Distributions Plots for Slice 6 at Stage 4 Pseudo-Steady
Reservoir Level i.e. El. 2656 m.s.m.s.l...... 206
xxi
List of Figures (Continued) Figure Description Page No.
5.61 Head Distributions Plots for Slice 1 at Stage 3 Pseudo-Steady
Reservoir Level i.e. El. 2647 m.s.m.s.l...... 207
5.62 Head Distributions Plots for Slice 2 at Stage 3 Pseudo-Steady
Reservoir Level i.e. El. 2647 m.s.m.s.l...... 207
5.63 Head Distributions Plots for Slice 3 at Stage 3 Pseudo-Steady
Reservoir Level i.e. El. 2647 m.s.m.s.l...... 208
5.64 Head Distributions Plots for Slice 4 at Stage 3 Pseudo-Steady
Reservoir Level i.e. El. 2647 m.s.m.s.l...... 208
5.65 Head Distributions Plots for Slice 5 at Stage 3 Pseudo-Steady
Reservoir Level i.e. El. 2647 m.s.m.s.l...... 209
5.66 Head Distributions Plots for Slice 6 at Stage 3 Pseudo-Steady
Reservoir Level i.e. El. 2647 m.s.m.s.l...... 209
5.67 Head Distribution plan plots for all slices (top left – slice 1 to bottom
right – slice 6) from 3D FEFLOW model...... 210
5.68 Comparative graphical plots at Section X-X for head distribution at
pseudo-steady state impounding reservoir levels ...... 213
5.69 Comparative graphical plots at Section K-K for head distribution at
pseudo-steady state impounding reservoir levels ...... 214
5.70 Foundation profiles for uncertainty propagation analyses ...... 215
5.71 2-D (SEEP/W) Seepage analyses output showing equipotential total
head contours and flow paths for Section X-X and Section K-K ...... 216
xxii
LIST OF ABBREVIATIONS
2-D Two Dimensional
3-D Three Dimensional amsl above mean sea level
ASCE American Society for Civil Engineers
ASTM American Standarda for Testing Materials
CAD Computer Aided Design
CDO Central Design Office cfs Cubic Feet per Second cm/s or cm/sec Centimeters per Second
CMTL Central Material Testing Laboratories
COV Co-efficient of Variance
CPA Closest Point Algorithm
DHI Danish Hydraulic Institute
ECRD Earth Core Rockfill Dam
El. Elevation
FEM Finite Element Method
Fn. Function ft Foot (feet) gpm gallons per minute
HEPO Hydro-Electic Power Organization
K Permeability / Hydraulic Conductivity km Kilometer (s)
Ksat Permeability / Hydraulic Conductivity at saturation
xxiii
Kx Permeability / Hydraulic Conductivity in the horizontal direction
Ky Permeability / Hydraulic Conductivity in the vertical direction lbs pounds m meter (s) m2/s or m2/sec Square Meters per Second m3/d Cubic Meters per Day m3/s or m3/sec Cubic Meters per Second
MCM Million Cubic Meter
MED Main Embankment Dam
NAWO Northern Area Works Organization
No. Number
NSL Natural Surface Level
P&I Planning and Investigation, WAPDA
PZ Pervious Zone in foundation assumption
RW RockWorks modelled foundation
SDC Satpara Dam Consultants
SMS Seepage Monitoring Station
SP Standpipe Piezometer
STDEV Standard Deviation
Vol. W.C. Volumetric Water Content
VWP Vibrating Wire Piezometer
W.C. Water Content by Weight (Volumetric Water Content)
WAPDA Water and Power Development Authority
xxiv
CHAPTER 1.
INTRODUCTION
1.1 GENERAL
Embankment dams are complicated structures constrained by physical conditions of the site. These structures are founded on and surrounded by naturally contrived environments and are engineered for compatibility at the base and flanks.
The natural arrangement (frequent disarrangement) of foundation materials sometimes leads to unforeseen and difficult consequences requiring extensive remediation of seepage problems through the foundations.
The most important aspects of designing and building safe dams and serving their intended purpose of storing water are methods of analysing seepage through the foundations or under seepage, which is the conventional terminology to this effect.
The preferred deterministic approach for under seepage analysis includes developing foundation profiles from site investigations, followed by assigning of hydraulic conductivity values. Finite-element seepage analyses are then conducted considering multiple cross-sections, to compare predicted gradients to piping criteria. The results of such analyses are interpretations of computerized flow nets. Usually, flow nets
(manual or computerised) are based on the assumption that the soil within a given stratum through which the water percolates is uniformly permeable.
The greatest uncertainty of an embankment dam lies in the condition and effectiveness of foundation treatments for seepage control. Foundation seepage through deep natural foundations faces the propensity for serious design and safety problems. This aspect is aggravated with heterogeneity of the foundations in the
1
absence of bed rock or any suitable impervious stratum at shallow depths. Piping incidents in fairly common glacial, alluvial and fluvial depositional environments are reported in literature to occur with increased frequency compared to other foundation types. Multiple foundation strata of different soil media, complex geometries and intricate geological conditions complicate the realistic scenario.
In general practice, theoretical solution to such physical anomalies is addressed assuming the foundations being composed of one or more homogeneous strata with well-defined ‘layered’ boundaries. Scaled transformations and/or variable equivalent conductivity assumptions are used in succession as a matter of conjecture.
This understanding accentuates ‘qualification’ over ‘quantification’ of seepage, in designing, assessing, and monitoring of control measures for foundation seepage. In the words of Terzaghi and Peck (1967), “The safety of a dam with respect to a failure by piping has no relationship to the amount of water that escapes from the reservoir.
Large losses of water may be associated with a high degree of safety against piping.”
Foundation seepage is a direct response of the hydraulic pressure ‘head’ exerted by the depth of the reservoir water behind the embankment, consistent over a time scale. Concentrated seepage flows have the ability to cause catastrophic events.
The associated pore water pressures and hydraulic gradients act as a catalyst for initiating internal erosion or piping, along susceptible sub-surface flow paths.
Foundation seepage problems are generally based on assumptions of a ‘steady state’ flow, which exists when the reservoir level has been at the same elevation (the conservation water level) long enough for the highest/maximum pore water pressures to develop everywhere in the zone of seepage. In reality, the reservoir behind the majority of earth dams is never continuously full, and there is a time lag before
2
changes in the reservoir elevation are reflected by changes in the pore pressures of the foundation. This time lag also depends on the permeability of the soil, and hence carries over an inherent uncertainty of estimation. This effect is neglected in general practice with an assumption that the reservoir remains full, long enough for the steady-state pore pressures to develop. As such, the pre-construction foundation seepage analyses are not usually maintained or extended for comparison to the actual foundation response during and after the dam construction.
1.2 CURRENT TRENDS IN PAKISTAN
In Pakistan, two major dams were located on deep alluvial filled valleys.
These include the gigantic Tarbela Dam on Indus River and the Khanpur Dam on
Haro River. An upstream impervious blanket was the prime seepage control measure resorted to in both these dams. The alternate solutions of providing a positive vertical cutoff or a partial cutoff were technically impracticable and economically prohibitive, because the impervious stratum was very deep.
The design of an upstream blanket was pioneered at Tarbela, in the late fifties to early sixties. At that time, computer modelling of foundation seepage problems was not common. The under-seepage design was done through manual estimation of flow nets, using assumptions much simplified compared to the actual situation. The different schemes considered included the foundations as being homogeneous and layered, with an average equivalent foundation permeability of 10-2 cm/sec. Honjo et al. (1992) later on presented a comparison of the theoretical results with the actual monitoring data. Their assumption for assigning the value for foundation permeability was the same 10-2 cm/sec. Their results have shown a good
3
agreement with the observed values. However the known incidents of sinkhole formation (referred in detail in the literature review) was not addressed nor depicted from their results. Regular monitoring at Tarbela relies on an observational method of comparative evaluation. The project is inspected annually by DSO and periodically at intervals of about four years by specially selected engineers from outside organisations. The first periodic inspection was carried out by the project designers in
1986-87, followed by the second, third and fourth inspections in 1991, 1995 and 1998 respectively. The most recent inspection was carried out in August 2006 at high reservoir level and then in April 2007 at low reservoir level. In the latest inspection, comparisons were made with the instrumental readings observed during the fourth inspection carried out in 1998. Data for intervening years was also examined, as percent piezometeric potentials for individual piezometers along the flow direction, to analyse the trends. The rise in piezometeric levels in the alluvial foundations was attributed to the ponding at Ghazi Barotha barrage (recently constructed downstream of Tarbela dam), and improvement due to strengthening of the blanket by sediment deposition and healing of sinkholes formed in early years of the project.
At Khanpur Dam, the experience gained from Tarbela was first utilized for design and construction. Comparative modelling was again done at later stages, a 3-D model by Faiz-ul-Hassan (2001) and a 2-D model by Jamil et al. (2010). Both of these modelling attempts involved homogeneous foundation assumptions, with inconsistent results when compared with the actual observations. The Khanpur dam has a three- tier monitoring and surveillance plan put in place by WAPDA, to ensure the project safety. The first tier inspection is conducted by the project site surveillance section through regular visual and instrumental observations. The second tier inspection is
4
carried out by DSO WAPDA, while the third tier inspections are carried out by an independent team of experts. Four periodic inspections have been carried out under the third tier, in chronological order 1985, 1988, 1993 and the latest in 2007. Seepage through the foundations has remained the focus of attention. Several interventions have been made for its reduction and control. Unfortunately, out of the 22 piezometers installed below the blanket only four were in working order during the third inspection. The fourth inspection has reported that they have ceased to work.
The available records (in the form of individual piezometeric elevation plots against reservoir heads and the change in annual potentials contoured on the project plan) however showed progressive behaviour reflecting a decrease in the piezometeric levels corresponding to the blanket, as the sediment load brought in by the floods strengthened the existing blanket.
1.3 PROBLEM STATEMENT
This research has availed Satpara Dam Project as an intricate case history, founded over moraines. The foundation moraines at Satpara Dam Project are characterised with a material composition constituting marked open works of gravels, cobbles, boulders of variable sizes and little binding material like silt and clay. In consequence of this diversified subsurface structure, these deposits also comprise of highly permeable zones of unknown extents. The absence of bed rock or any suitable impervious stratum at considerable depths, poses difficulties in a positive seepage cut- off, whereas space restrictions due to the presence of natural Satpara Lake at the upstream end under play the effectiveness of an impervious blanket in terms of the required extents (length) of placement. These typical site conditions necessitate the need for a flexible design to adjust to the unforeseen site conditions (e.g. buried
5
boulders, unknown extent of openwork zones and depth of bedrock). As such the adopted seepage control measure at Satpara Dam was a combination of the upstream impervious blanket with an additional partial cutoff wall at its upstream end.
The project design was reviewed in 2004 prior to commencement of the construction works, for adequacy checks in regards to the foundation seepage control.
2-D seepage analyses were conducted with a conservative and safer but relatively limited approach, specifically governed by site conditions and serious time constraints. The heterogeneous foundations at Satpara, as modelled for the design review, used a two-layered approach. The conventional and established concept of
‘exit gradients’ at dam toe was considered in a sensitivity analysis for acceptance of the foundation seepage control measures.
The basis for all seepage analysis and related concerns are dependent on reliable estimates of permeability (K). The complex heterogeneity of the Satpara foundations presents significant uncertainties in this regard. Selection of representative hydraulic conductivity ‘K’ values for modelling the natural foundations for seepage sensitivity analyses is therefore realized as a core issue in this research.
The applicability of an average permeability used for ‘as designed’ analyses shall be re-considered keeping in mind the range and pattern of variation throughout the foundations. This research addresses difficulty of quantifying actual spatial stochastic variation for seepage modelling and sensitivity analyses scenarios in detail. The proposed approach was intended to indicate the possible range of behaviour corresponding to representative foundation models, with different assumed distributions of the hydraulic conductivity. For the proposed research approach, that combines statistical and finite-element techniques, analyses with a range of
6
permeability patterns would also enable estimation of maximum possible critical gradients. The presence of highly pervious but ill-defined strata shall complicate the seepage study and as such, statistical analyses of the influence of material and permeability variations shall be used to estimate a critical nature of the computed hydraulic gradient distributions at different points along the flow direction.
The selected case study has also reached its completion stage with valuable instrumented information gathered through its construction period. Performance evaluation at Satpara Dam was independently done by spot checks on the observed instrumented data from the field staff. Different teams of experienced representative personnel from a Panel of Experts (PoE), Dam Safety Organization (DSO), and
Monitoring and Surveillance (M&S), the latter two being WAPDA subsidiary organisations, were frequently involved in this process.
Standard performance evaluation procedures in practice are usually aimed at visual or instrumented observations from the operational phases of the projects and pre-construction modelling efforts are generally disregarded. This research was anticipated to investigate the confidence level for acceptability of pre-construction seepage analysis in considerations of post construction monitoring and performance evaluation. The research approach thereby evolved a three dimensional foundation permeability data management based on in-situ field measurements, first to establish seepage sensitivity with respect to critical parameters, and then the model simulations were adapted to corroborate with piezometeric response from observed data.
Information and findings from the first phase of the research extended general conditions for optimal utilization of selective control measures related to foundation seepage. Whereas the second phase distilled site specific instrumented observations
7
from the selected case history through meticulous understanding, to present an improved construal of performance evaluation alongside a consistent framework of logical foundation seepage analysis.
1.4 STUDY OBJECTIVES
The specific objectives incorporated within this study cover the following aspects:
1. Assessment of representative foundation permeability, adequately
apprehending the subsurface complexity (heterogeneity), with relevance to
seepage modelling.
2. Multivariate sensitivity analyses for selection of adequate foundation seepage
control measures, covering the possible range of expected field conditions.
3. Performance evaluation of adopted foundation seepage control measures
through instrumented observations.
4. Comparative appraisal of the most reliable estimates to contemplate post-
construction behaviour of foundation and construe related implications.
1.5 SCOPE OF THE STUDY
The scope of this research is broadened for general applicability to all embankment dams on deep permeable heterogeneous foundations signified in context of constricted valley topography. The researchable issues put forth have considered the subsurface investigations data from conception, design and construction phases of the case study project of Satpara Dam. No additional field investigation or laboratory experimentation was conducted for the research.
8
Impounded reservoir levels from past five years of project operation from
2007 to 2011, during and after the dam construction, did not achieve the designed hydraulic loading. Performance evaluation of the selected seepage control measures of foundation seepage for was thereby limited to the available observation data corresponding to the actual impounded reservoir levels for the purpose of research.
9
CHAPTER 2.
REVIEW OF LITERATURE
2.1 GENERAL
This chapter summarizes previous research work conducted in areas related to the implications and performance evaluation of selective control measures of foundation seepage for embankment dams over permeable strata. The imminent literature review concentrates on the subject matter, and helps in the general problem perception. It also provides an overview of the related guidelines and standard norms of practice. The background information useful in assessments of representative parameters, performing the analyses, and making the selections are discussed herein.
This chapter broadly subdivides into two main sections. The first section describes theoretical concepts of foundation seepage, with specific relevance to foundation variability and piping. The subsequent section covers selected experiences associated with this topic regarding the design, construction, and performance of dams with emphasis on their foundation seepage control measures.
2.2 THEORETICAL CONCEPTS
2.2.1 Risk Analysis
Peck (2000) discussed the application of risk assessment to seepage and piping within the dam safety program of the Corps of Engineers as part of a two-day workshop. Dr. Peck has accentuated the importance of a site-specific experience, reflective of a critical study of precedents through case histories of successes and failures, accompanied by risk analysis for an engineered judgment on uncertain issues
10
of concern. He has emphasized this as the only adequate and realistic approach in dam safety assessment.
Peck (2000) approved the usefulness of statistical analysis considering that unwarranted conclusions are not drawn and that the analysis verifies actual observed or recorded data. Appreciation of the foundation variability introduces ‘oddball’ features, which differ from anomalies in their chances of predictability. Statistical analysis may fail to construe these occurrences whereas explicit familiarity and detailed knowledge of the physical concepts prove helpful in reasonable predictions.
Peck (2000) has referred sinkholes at Tarbela in this context, from the experience of John Lowe (1978). The formation of some 400 sinkholes did not lead to a piping failure. The specific ‘assemblage’ within these foundations, which consisted of lenses of sand and gravel, were not continuous, but erratic. Supply of additional material to the sinkholes did not cause failure but conversely, further migration of particles improved the performance and was able to induce self-healing properties in the foundations.
2.2.2 Remediation of Seepage Problems
Talbot et al. (2000) presented a “strawmen” state-of-the-practice white paper as part of a three-day workshop on Issues, Solutions, and Research Needs Related to
Seepage through Embankment Dams. During the presentations and the ensuing discussions, it became apparent that the state-of-the-practice in the area of seepage/piping problems and their control/remedy varies from region-to-region and organization-to-organization. Consequently, exclusive use of “cutoffs” (almost no
11
technique used for this purpose will actually completely cut off water flow) for seepage/piping problems is the current practice. The consensus of the workshop group supported the conjunctive use of cutoffs with adequate collection and control systems, the constituents of which are subject to engineering judgment.
Talbot et al. (2000) presented two principal problems associated with seepage through, around, and beneath an embankment dam i.e. piping and excessive water loss. Remedial measures for preventing piping control seepage, lessen chances of internal erosion in soil, but may not decrease the rate of seepage. Remedial measures for reducing water loss decrease the quantity, pressures and the rate of seepage through a dam, its foundation, or abutments. The best-suited solution to a particular dam depends on a variety of factors, as discussed by them in the white paper. Some of the more important factors in regards to the foundations include the foundation stratigraphy, seepage patterns, pressures and quantities.
Reducing the amount of seepage is beneficial in saving water as a resource and lowering the seepage pressures within the dam, foundation and abutments, thus reducing to some extent the probability of a piping failure. Methods available to reduce the amount of seepage include specialized constructions with low permeability materials (e.g., soil, soil and bentonite mixtures, soil-cement mixtures, asphalt, roller compacted concrete, concrete, metal, masonry, or a geomembrane). These specialized constructions may either include an upstream blanket, a “cutoff” or facing on the upstream slope of the dam and an internal “cutoff” within the dam and foundation.
Diaphragm wall, sheet piling, or grout curtain are sometimes referred in context of the latter and constructed with such methods as slurry trench excavation, deep soil mixing, or jet grouting.
12
Significant factors that affect the selection of the most appropriate solution for seepage reduction measures include the necessity for limiting the potential of hydraulic fracturing when constructing an internal cutoff, and the need to provide protection against erosion and animal damage for shallow elements such as upstream blankets and upstream facings. Finally, Talbot et al. (2000) are of the opinion that proper downstream drainage collection and safe discharge must always accompany any flow reduction technique.
2.2.3 Foundation Heterogeneity
Elkateb et al. (2003) reviewed recent advances in uncertainty assessment of foundation parameters. These uncertainties comprise lithological heterogeneity and inherent spatial variability of specific soil properties, such as permeability. Adverse predictions in up to 70% of the case histories by Morgenstern (2000) have been referred to aggravate the need of methods for quantitative assessment over general engineering judgement and reliance on factors of safety.
According to Elkateb et al. (2003), initial attempts to address complex ground profiles involved permeability up-scaling techniques as acquiescent to flow simulation and engineering calculations. They reviewed three general classifications i.e. empirical, semi-empirical and analytical methods or techniques. The power averaging technique (Deutsch, 1989) exemplifies under empirical methods. The more sophisticated semi-empirical methods include renormalization (King, 1989) and the representative elementary volume (REV) – renormalization (Norris et al., 1991). The geometric mean (Deutsch, 2002) as an estimate of the effective permeability of heterogeneous media was the premier analytical solution proposed by Warren and
13
Price (1961). They arrived at this conclusion by combining results of physical modelling with numerical simulation.
Elkateb et al. (2003) expressed assessment of the effects of inherent soil variability on engineering design through stochastic analyses. Phoon and Kulhawy
(1999) have summarized selected statistical characteristics for different soil types and field tests.
Elkateb et al. (2003) related that different studies, like Lumb (1970), Schultze
(1975), and Griffiths and Fenton (1993), have indicated that there is no generic pattern for soil properties and classical statistical properties of soil, such as the mean value, COV, and probability distributions are site and parameter specific. The authors have reviewed and criticized different approaches adopted for stochastic analyses with emphasis on the limitations of the current practice. Their discussions specifically include the application of reliability principles to limit equilibrium analyses, stochastic finite element analysis and the application of stochastic input soil parameters into deterministic numerical analysis.
In succession, Elkateb et al. (2003) have also discussed significant decision- making algorithms regarding the soil parameter for use in engineering analysis, together with comments regarding their practical applicability. These discussions have covered the worst-case and quasi worst-case approaches, reliability-based techniques, confidence interval approach, and Bayesian decision analyses. The major outcome from this paper is that estimates of representative soil parameters should consider heterogeneity aspects and the decision-making framework should also incorporate the project risk level.
14
The most relevant study reviewed by Elkateb et al. (2003) is that of Griffiths and Fenton (1993), who have studied the effect of stochastic soil permeability on confined seepage occurring beneath water retaining structures. Their study combined finite element methods with random field concepts for the generation of soil permeability properties (with a fixed mean, standard deviation and spatial correlation structure), to perform Monte Carlo simulations of the seepage problem. Griffiths and
Fenton (1993) used the local average subdivision simulation technique to generate
1000 realizations of spatially variable hydraulic conductivity below a water retaining structure. The resulting field was then mapped onto a finite element mesh to perform numerical analysis of the problem under deterministic boundary conditions. Griffiths and Fenton (1993) assumed the hydraulic conductivity, K, to follow lognormal probability distribution, and accounted for the effect of spatial correlation structure through quantifying the influence of the scale of fluctuation on different response
(output) variables. They presented results of parametric studies to gauge the effect of standard deviation and correlation structure of permeability on the output statistics relating to seepage quantities, exit gradients and uplift pressures. They made a comparison for all cases with results that would be achieved on a deterministic basis and showed that gradients, flow rates and other quantities of interest were significantly affected by both the standard deviation and the correlation structure of soil permeability.
The limitations of the study by Griffiths and Fenton (1993) as identified by
Elkateb et al. (2003) lay in the use of an isotropic correlation structure where vertical and horizontal ranges were assumed equal, and the effect of different probability distributions and correlation structure models were not accounted for. Furthermore,
15
the sensitivity of the analysis to the number of realizations of the random variable, hydraulic conductivity, was not considered.
2.2.4 Uncertainities in Analyses
Jafarzadeh and Soleimanbeigi (2005) addressed the validity of seepage analysis results (i.e. downstream discharge rate, phreatic surface and downstream toe gradients) through their dependence on model geometry, uncertainty and non- homogeneity of the material properties. This study has shown that a finer mesh
(increased number of elements) reduces the magnitude of error, and higher estimates of the foundation permeability have a direct influence on the seepage analysis results.
Other outcomes of Jafarzadeh and Soleimanbeigi (2005) include sensitivity of abutment seepage to material properties and depth of grout curtain (the adopted seepage control measure for the case study). A change in these parameters affects discharge rates from the foundations and opposite banks. Variation of the grouting curtain depth within a layer will not result in a significant change in water flux, phreatic surface and the hydraulic gradients until it enters into a different underlying layer with a lower permeability coefficient. This also indicates the influence of lithological heterogeneity on foundation seepage treatments.
While addressing the topographical characteristics of a narrow valley,
Jafarzadeh and Soleimanbeigi (2005) have indicated requirements of higher margins of safety for a 2-D approach as compared to a 3-D model in a comparative scenario.
16
Jafarzadeh and Soleimanbeigi (2005) also support the use of curves and graphs as a comparative source in decisions regarding uncertainties arousing from seepage analysis inputs.
2.2.5 Appraisal of Piping
Richards and Reddy (2007) covered the progressive development of piping.
They have signified that ancient civilizations substantiate evidences of dams serving people for at least 5,000 years. However, the early methods of construction did not consider the effects of seepage in earth dams; and according to them probable piping failures might have occurred since the earliest dams were constructed around the period BC. They have also implied that experience evolved through successful construction of dams on a variety of foundation materials and empirically successful dam designs emerged by the first millennium AD. They refer Jansen (1983) who has contributed with elaborate historical information on this aspect.
Richards and Reddy (2007) briefly presented the historical awareness, evaluation and the subsequent design of defensive measures. They referenced different interpretations as the basis of various definitions to clarify the root cause of piping, including Backwards Erosion, Internal Erosion, Tunneling/Jugging, Suffosion and Heave.
Richards and Reddy (2007) supplement statistics of piping failures from previous studies of Foster et al (2000a), with 267 accumulated case histories.
Foundation related failures are comparative i.e. 15% from the detailed analysis of
Richards and Reddy (2007) and around 16.4% reported by Foster et al. (2000a). The
17
intended scope of Richards and Reddy (2007) was to evaluate failures due to piping; it thus included extended statistics pertaining to internal erosion (49.8%) and backwards erosion or suffusion (31.1%).
Richards and Reddy (2007) thereafter reported fundamental references in regards of piping awareness and evaluation, which include Clibborn (1902), Bligh
(1910), Lane (1934), Terzaghi (1922) and Casagrande (1937). The work of Col.
Clibborn subsequent to the 1898 failure of Narora Dam, highlights the importance of hydraulic gradient.
The Bligh’s creep theory, Lane’s weighted creep theory, Terzaghi’s theory of
‘heave’, and Casagrande’s formalization of the flow nets all provided an empirically derived basis for estimating the piping potential. Based on these theories and relationships, the increased piping potential is directly proportional to hydraulic head and inversely proportional to the length of seepage path. Terzaghi’s progressive work including Terzaghi (1939, 1943), Terzaghi and Peck (1948) and Terzaghi et al. (1996) in this regard has been accentuated by Richards and Reddy (2007), pertaining to initial piping failure mode perceptions.
In succession to the aforementioned, according to Richards and Reddy (2007), some other prominent early works include development of the filter design criteria as a defensive measure against piping failures, evaluation of internally unstable soils and the advent of the critical hydraulic gradient. Aberg (1993), Honjo and Veneziano
(1989), Sherard and Dunnigan (1989), Sherard et al. (1984 a, b), Arulanandan and
Perry (1983) are significant citations in regards to the advances in filter design.
18
Skempton and Brogan (1994), Adel et al. (1988), Kenney and Lau (1985), Kovacs
(1981), Kezdi (1979) are significant citations for evaluating internal stability of soils.
Richards and Reddy (2007) presented a follow-up from selected recent works like Foster et al. (2000b, 2002), Fell et al. (2003) and McCook (2004), to identify the importance of different piping failure modes, predictions of piping breach timing, prediction of clogging of filters and risk based assessment of piping failure. Richards and Reddy (2007) have indicated that engineering guidance on piping suggests safe hydraulic gradients of less than 1.0; although piping has also been reported to occur at lower gradients. In the views of Richards and Reddy (2007), generally ‘piping’ describes different phenomena and evaluation of its potential has to consider a range of different mechanisms. These aspects may be interrelated and conversely independent, subject to the progressive insight of the problem.
Specific conclusions from the review of Richards and Reddy (2007) indicated that to-date constitutive aspects of piping are unknown, the influence of seepage orientation has not been quantitatively studied, segregation piping (suffusion) may be a more common phenomenon in non-cohesive materials and currently there is no satisfactory method for evaluation of piping in cohesion-less soils.
Guidelines for evaluating the failure potential from seepage and piping related failure mode are enumerated by Von-Thun (1996) as five fundamental factors.
1. The physical conditions of the dam and foundation (geology and dam design
and construction features).
2. The nature of gradients through the dam and foundation (direction of flow
based on piezometric data).
19
3. The historical/operational performance of the dam in terms of observations of
direct evidence of potential seepage related problems (sinkholes, sand boils,
material transport).
4. The nature of seepage at site (always clear, does or does not change as a
function of the reservoir elevation, is constant for a given reservoir elevation,
is observable, etc.).
5. The presence and effectiveness of defensive design measures.
2.2.6 Typical Cutoff Considerations
Mukhopadhyay (2008) analyzed the seepage characteristics of a permeable foundation by combining the finite element method with flow nets. He initially presented a brief history of the development in theoretical backgrounds and numerical computations. He referred the Darcy’s Law of 1856, Casagrande (1961), Singh and
Punmia (1973) for the scientific and experimental approaches to the seepage theory.
He thereafter acknowledged the onset of numerical procedures from the works of
Desai (1975, 1976), Meek and Beer (1976), Akin and Gray (1977), Christian (1980),
Aalto (1984), Stelzer and Welzel (1987), Desai and Baseghi (1988), Tracy and
Radhakrishnan (1989), and Fan and Tompkins (1992).
Mukhopadhyay (2008) attempted to use the Finite Element Method is to determine the nodal potentials of four node quadrilaterals elements. Subsequently flow nets are drawn on the basis of the F.E.M. nodal potentials and bilinear shape functions. He has investigated the effect of cutoff penetration depth (d), width of a top horizontal impervious barrier (B), and the influence of foundation anisotropy with respect to its permeability on the ensued seepage. The model formulation being
20
somewhat relevant under the objectives of this research is described briefly. The permeability variation with foundation depth (D) has been assumed parabolic under alternate scenarios of increasing (to a maximum of n) and decreasing (from a maximum of m), both with a minimum limit of unity. The influence of ‘d’ and ‘B’ both have been considered separately as aspect ratios to the total depth ‘D’.
A constant d/D of 0.8 was considered with 5 cases of D/B (0.25, 0.5, 1, 1.5 and 2.5), five position alternates for the cutoff (upstream end, B/4, B/2, 3B/4 and downstream end) and nine permeability variation cases (m = 1, 2, 4, 10, 100 and n =
2, 4, 10, 100). As such a total of 225 cases were analyzed under a constant head differential from upstream to downstream. The sectional model developed comprised a zoned continuous medium assembled of discrete size elements interconnected at nodes. The individual zones were arranged depending on the individual D/B ratio and variable permeability assumptions.
The computer program developed by Mukhopadhyay (2008) was validated through manual computations. Comparative families of variation curves for different permeability approximations (m or n) were developed for evaluation of the results.
These variation curves typically included cutoff positions with discharge, foundation depth with discharge, and nodal potentials under the top horizontal impervious barrier with the distance from upstream end.
Salient outcomes of Mukhopadhyay (2008) revealed that permeability variations do not influence the discharge for the same position alternate and the same
D/B ratio, however comparative nodal potentials decrease from upstream to downstream and increase with the D/B ratio, relative to permeability variations. The
21
results also show that the discharge varies inversely with the D/B ratio and comparative discharge corresponding to same D/B ratio increases until the ¾th position alternate for the cutoff following which a sudden decrease is observed.
Similar theoretical cutoff wall performance has been studied by various other researchers. Rice (2007) has reviewed some relevant work of Telling et al. (1978a,
1978b and 1978c) and Lefebvre et al. (1981). Rice (2007) has also independently carried out finite element analyses for seepage through foundations to provide a better understanding of the performance of seepage barriers and the mechanisms that affect the performance.
Results of Rice (2007) indicate that a cutoff is effective in reducing the flow and consequently the gradient. Based on the analyses, he has established that the magnitude of the hydraulic gradients is a function of the following important factors;
1. The differential hydraulic head across the cutoff,
2. The depth of embedment into a low permeability layer,
3. The permeability ratio of the permeable layer being cut off and the low
permeability layer; and
4. The configuration of the seepage barrier and other soil layers in the dam cross
section.
Rice (2007) concluded that it is important to analyse each dam and seepage barrier individually to assess the potential for developing high hydraulic gradients. He also recommended specific considerations regarding the potential under analysed gradients to cause internal erosion.
22
2.2.7 Typical Blanket Considerations
Gohernejad et al. (2010) discussed the importance of study of seepage through earth dams and shows different preventive measures that have been used all around the world depending upon the type of reservoir and economic studies. One of these preventive measures includes a layer of low permeability placed on the upstream side used to control foundation seepage through deep alluvium layers.
Gohernejad et al. (2010) referred practical references in this regard, including the Fort-Randal and Gavins-Point dams by Lane and Wohlt (1961), Chief Joseph and
McNary dams by Brown (1961), the dam of Saskatchewan River in Canada by
Peterson (1968), Arrow and More-Falls dam by Wilson and Marsal (1979), and
Tarbela dam in Pakistan for one of the largest upstream impermeable blanket from the
WCD report by Asianics Agro-Dev. International (Pvt.) Ltd. (2000). Gohernejad et al.
(2010) have referenced Uginchus (1935) in presenting the differential equation of seepage in upstream blanket for unlimited and limited lengths and Bennett (1946) who resolved the same differential equation of seepage in upstream blanket dams by a different approach to additionally achieve an optimized blanket length.
Gohernejad et al. (2010) analysed the Farim-Sahra Dam using SEEP/W for an upstream clay blankets with various lengths and thicknesses. The deepest cross section was modeled with material divisions and permeability coefficients conforming to the geological studies. Initial analyses considered a clay blanket only on the surface of reservoir, variable lengths (50,100 and 150m) and thicknesses (0.15, 0.3, 0.5, 1.0,
1.5 and 2.0m). Their results were ineffective in reducing seepage, however, an
23
additional case with an extension of the clay blanket upto the surface of upstream shoulder reduced seepage quantities to about 75% in comparison with the first case.
Gohernejad et al. (2010) have graphically compared lengths and thicknesses of clay blanket and the volume of soil used with the seepage amount in percentage.
Their comparison shows that increasing thickness beyond 0.15m for blankets with 50 and 100m lengths has no appreciable effect to reduce seepage in case of upstream shoulder covered with clay blanket. Blanket length of 150m has maximum efficiency with a thickness of 0.6m. Their study revealed that increasing the blanket length is more effective to reduce seepage than to increase the thickness for equal soil volumes.
A comparison of the analyzed blanket thicknesses was made with those computed using Bennett’s equation for defined lengths in the case where upstream shoulder was covered by impermeable blanket.
Gohernejad et al. (2010) suggested an upstream clay blanket with 150m length and 0.75m thickness for execution at the Farim-Sahra Dam as a result of their comparison, corresponding to the maximum reduction in seepage. They have concluded that the effectiveness of an upstream blanket in increasing the creep length and reducing seepage was apparent. The best optimization is achieved through the maximum possible thickness and length, subjective to local and topographical situations. Extending an upstream blanket over the upstream shoulder of dam may not be practical for all dams and site specific considerations in this regard are imperative.
24
2.3 ASSOCIATED EXPERIENCES
2.3.1 Upstream Impervious Blanket
Tariq (2003) described the unusual experiences regarding one of the world’s largest impervious blankets placed, at Tarbela Dam in Pakistan. The main dam is a
143 m high and 2745 m long embankment dam founded on a dense deposit of river alluvium up to 220 m in depth, containing layers of openwork gravel. The design of an impervious upstream blanket with vertical drainage at the downstream toe was the solution resorted to, and at that time the profession had no previous experience of providing a blanket for a maximum head of 145 m. The impervious blanket extends about 1500 m upstream from the toe of the Main Embankment. Its width is 1980 m at the toe and about 2590 m in the vicinity of the upstream end. The thickness of the blanket varies from 1.5 m at the upstream end-which thickness is held for 300 m on a nearly parabolic curve to 12 m at the toe of the embankment. The blanket was well instrumented to monitor the blanket performance upon reservoir filling.
Following the first filling of the reservoir to 80% of its maximum depth in
1974 summer, an excessive underseepage of 8.5 m3/s was recorded downstream of the main dam. A failure of the upstream part of the diversion tunnel No. 2 occurred and the reservoir was depleted. This dramatically revealed that serious problems had already developed in the upstream blanket. A total of 362 sinkholes were detected together with about 140 cracks as well as several compression ridges. These sinkholes varied in size with the maximum sinkhole diameter 10m and maximum depth 2.5m.
Depleting the reservoir provided an excellent opportunity of direct observation of the blanket in addition to regular instrumentation monitoring.
25
Because of large extent of the project and the great depth of the alluvial foundation and the special problems posed by the presence of openwork and sand strata in the alluvium it was not feasible to investigate the foundation extensively enough to be sure beyond any reasonable doubt that no problems would ensue. To meet this circumstance, the Observational method was adopted and was proven to be entirely satisfactory. The observational method instigated at Tarbela was then used more and more for other dams in order to assure dam safety, particularly during first fillings of reservoirs. The method requires more extensive instrumentation and observation than customary, but provides an assurance that cannot be attained in any other way.
The work at Tarbela has been extensive, both in the site investigations and design as well as in monitoring the performance of the dam during construction and after completion. John Lowe III presented the fourth Nabor Carrillo Lecture on the
Foundation Design of Tarbela Dam, in 1978. This lecture has presented a case history of the design and behavior of the foundation of Tarbela Main Embankment Dam.
Causes of the development of sinkholes were studied and discussed with Dr.
Casagrande and other members of a Panel of Experts (POE). The following were discussed as the major causes of formation of sinkholes.
1. Differential settlement due to the load of reservoir on 200m thick alluvium in
the valley overlying uneven rock profile.
2. Erosion in vertical direction may had been supported by unfilled boreholes
and improperly filled open wells or pits in the valley. This phenomena was
observed mainly on the left bank due to open wells which existed in this area.
26
3. The most likely cause according to Dr. Casagrande was the migration of the
sand lying below the blanket into the open work gravels as most of the
sinkholes were found in the known position of the open work gravels.
4. Sinkholes may also have formed due to the settlement of fill around the
instrument cables. Some of the sinkhole were exactly located above the
vertical riser of the instrumentation cables.
Following evacuation of the reservoir, remedial measures that were taken and after that found to be effective were,
1. The sinkholes were filled and then overlain by a protective layer of well-
graded blanket material.
2. Blanket thickness was increased by 1.5m to 3m in the areas underlain by the
open work gravels. This additional material consisted of uncompacted filter
material grading from silt to cobble size.
3. Additional line of relief wells.
4. Side scanning sonar was used to monitor the reformation of sinkholes and the
sinkholes that reformed after some time were treated by dumping a mix of
material ranging from silt to cobbles prepared with some percentage of
bentonite and compacted at moisture content above optimum level.
5. Additional sinkholes developed during the following three years, but at
decreasing rates, and these were filled by barge dumping.
Knight (1990) identifies Tarbela Dam as the most extensively instrumented dam in the world. He has given a comprehensive account of the indispensable role of foundation instrumentation played in the dam’s safe establishment, through a chronological description of the potential profiles (1974 – 1986) from the deepest
27
piezometers. He identifies the cause of sinkhole development to be tied with the occasional cracking of the blanket, induced simultaneously with the increased differential stresses across the blanket due to reservoir sedimentation, indicated by the upstream movement of equipotentials.
Agha (1975), Izhar Ul Haq and Altaf Ur Rehman (1984), Agha and Ahmed
(1990) and Izhar Ul Haq (1996) have also shed light on critical aspects of the Tarbela impervious upstream blanket and have builtup the knowledge database regarding the experience from subsequent follow-ups of sinkhole development and treatments.
Peterson (1968) commented on the performance of the South Saskatchewan
River Dam in Canada, which was provided with an upstream blanket. This earth dam was founded on as much as 30 m of fine to medium sand. It was filled in July 1967 to a water depth of 52 m as compared to the design depth of 58 m. Observations revealed that the performance of the dam was satisfactory with no excess seepage pressure in the downstream area (downstream filter and relief wells were provided and a total measured seepage flow of 0.1 m3/s was observed).
Golder and Bazett (1967) discuss the Arrow Dam, where the upstream blanket was constructed by dumping glacial till through water. Ripley and Campbell (1964) refer to a Seymour Falls Dam which contains an upstream blanket to control under seepage through relatively deep deposits of alluvium. Lane and Wohlt (1961) discussed the programme of Ft. Randall and Gavins Point Dam, each of which was constructed in the 1950’s on the Missouri River. The dams were provided with an impervious upstream blanket and a system of downstream relief wells. Ft. Randall
Dam was underlain by alluvium, consisting primarily of fine to coarse sand with some
28
gravel. The maximum depth of the alluvium was in the order of 52 m. Periods of observation cover from 5 to 7 years. The observations revealed that the head loss under Gavins Point Dam, within five year after commencement of operations, had increased as much as 40 percent over the initial values; no net change in head loss occurred under the impervious blanket in Ft. Randall Dam. The persistent increase in head loss within a short period of time at Gavins Point Dam was attributed to the accumulation of sedimentation in the reservoir bottom.
Brown (1961) has reported additional performance data on the effectiveness of upstream impervious blankets adopted to control seepage through the abutments of two dams on the Columbia River-Chief Joseph and McNary Dams. In both cases, the abutments consisted of coarse alluvium with design coefficients of permeability in the order of 0.5 cm/sec. The blankets extended upstream from the axis from 305 m to 610 m. In both instances, seepage quantities, measured in downstream collector drains and tunnels, diminished by as much as 50 to 60 percent of their initial values, all within 5 years after commencement of operations. As before, the increase in effectiveness of the blankets with time was attributed to the tightening up of the blankets and additional sedimentation in the reservoir.
2.3.2 Cutoff Wall
Weston et al. (2003) described the design expectations in comparison with the measured performance of Wilsons Dam. The subject dam is located in a valley with
15m deep sequence of soft alluvial silts and clays interbedded with permeable sand and gravel lenses. These foundation soils were investigated comprehensively by trial pits, boreholes and CPTs with relevant laboratory and field testing during feasibility
29
studies and final design. During construction further geological information was obtained from drillholes for extensometers and piezometers. The final foundation model developed for design and monitoring purpose had significant uncertainities associated with foundation permeability and settlement and stability problems. Design feature specifically incorporated to deal with seepage through the near-surface high permeability layers included a cement-bentonite slurry cutoff wall. Two instrumented cross sections were used including 54 vibrating wire piezometers in foundation and 16 in the fill to monitor seepage through foundation during impoundment and operations.
The seepage model of Weston et al. (2003) included adequate assessments for the permeable layer thickness, allowing for the inter layering of permeable lenses within the ‘estuarine silts’. The model was also extended upstream and downstream of the dam to account for losses through the ‘lower gravels’. The permeability assumptions were taken equal in both horizontal and vertical directions. Weston et al.
(2003) used the model to evaluate and optimize penetration depths for the cutoff wall and the wick drains. It was determined that effectiveness of the foundation seepage control was dependant on the penetration depths of the cutoff wall when installed in connection with the main dam core through the permeable layer of the ‘upper gravels’. Similarly, for effectiveness of the wick drains, installed along the upstream and downstream footprints of the main dam with the exception of the core, required minimizing their interconnection with the ‘lower gravel’ layers.
Weston et al. (2003) have indicated evaluation of a local incidence of cutoff trench collapse, where gravel deposits were thicker and close to the surface, through piezometeric responses. A comparison of piezometer records, for the region where trench collapse occurred and where the trench stayed intact, was useful in eliminating
30
the initial concerns that these collapses would have bridged the trench. Design estimates of seepage were compared with the measured seepage in relation to reservoir level during commissioning. Their results showed that the measured seepage did not exceed the designed expectations rather stayed well below the target maximum from the design. It was also demonstrated that seepage losses up to a certain extent did not affect reservoir viability.
Bruce et al. (2006) have provided, in tabulated and summarized forms, details from a total of 22 North American dams with relevant technological citations, for which some form of “positive cut-off” have been used as effective seepage control measures. They have elaborated on the various techniques of construction such as paneled constructions / diaphragm walls, secant piles, soilcrete walls built by the
Deep Mixing Method, and soil-cement walls in relation to factors as depth capability, geotechnical suitability, constructability, quality assurance, control and verification, and cost. Six dams referenced in Bruce et al. (2006) are worth noting in similarity to the foundation characteristics of the current study, as listed below;
1. Jackson Lake Dam referred by Farrar et al. (1990) was founded over alluvium,
comprising mainly sands, but with interbedded coarse gravels and other
materials. An 865-mm-thick deep mixed wall was constructed to provide
seepage cutoff through variable, but generally permeable alluvials under
reconstructed dam.
2. Prospertown Dam, founded over relatively permeable sand and silt, and
alluvials over impermeable glauconitic clay. A “Conventional” cement
bentonite wall was constructed using a backhoe to prevent seepage through the
underlying alluvium. Khoury et al. (1989) shed some light in this regard.
31
3. Wells Dam, founded over miscellaneous alluvium and very dense till. A 760-
mm-thick concrete panel wall installed by clamshell and joint pipe ends was
used to prevent piping as referred by Kulesza et al. (1994) and Roberts and Ho
(1991)
4. Cushman Dam, founded over variable glacial deposits (outwash, lacustrine
and till), often dense with boulders. A 610-mm wide wall constructed using
the Deep Mixed Method (DMM), formed with 1 m triple augers was used as
the effective control measure in two sections adjacent to Spillway to arrest
seepage (Yang and Takeshima, 1994).
5. Meek’s Cabin Dam, founded over very variable glacial till and outwash
comprising sand, gravel, cobbles, and boulders. A 0.9 m thick plastic concrete
panel wall formed by Hydromill was used to prevent seepage through glacial
outwash deposits, as referred by Pagano and Pache (1995) and Gagliardi and
Routh (1993).
6. Cleveland Dam, founded on heterogeneous glacial sediments including silt,
sand, gravel and till with hard igneous boulders. To prevent seepage through
glacial and interglacial foundation sediments, especially a 6 m sand layer, an
810 mm plastic concrete panel wall constructed by cable suspended clamshell
was used. Singh et al. (2005) described this case study in detail.
2.3.3 Combination of Blanket and Cutoff Wall
Lafuente et al. (2006) have discussed La Loteta Dam in Spain, as an exemplary case history addressing the strategic application of seepage control measures in accordance with their engineering significance. The water reservoir to be constructed as an outcome of this dam held a vital economic and social value. The
32
dam designed was aimed at controlling foundation seepage and preventing initiation due to karstification processes in the presence of gypsum, i.e. dissolvable elements.
They described technical solutions adopted in maintaining a highly impermeable foundation, while reducing the hydraulic gradients and ensuring that the slat contents did not dissolve. The heterogeneous embankment was designed with a wide central clay core, which ran horizontally under the upstream shell in a thick impervious blanket. The far end of the impervious blanket was linked up with a plastic bentonite-cement curtain / diaphragm wall constructed with a trench cutter using drilling slurry. Apart from the blanket and diaphragm wall, which were incorporated into the dam typical section itself, watertightness of the foundation was improved using the classical grouting system. A double line grout curtain following an alignment spaced at 3.2 m along the toe upstream of the core enhanced the ground permeability in zones where evidence of karstification was detected.
The blanket and cutoff assembly managed to reduce hydraulic gradients to around 0.2%. The dam is 1472 m long and the top stands 34 m above its foundations.
The core width is 1.2 times its height; the impervious blanket extension of the core is
6 m thick that reaches 150 m long in the central zone from the dam axis. The diaphragm wall is 1580 m long, 0.80 m thick with an average depth of 23 m, embedded in clay layers where there is no gypsum. The blanket was designed thicker than necessary to ensure that the blanket behaves correctly, because any deficiency in it cannot be corrected by the wall. Furthermore, the diaphragm wall acts as an element preventing small streams of water, formed as natural seepage, from flowing through the marly formation and creating karstic formations.
33
2.3.4 Long Term Performance
Rice (2007) discussed the measured performance of existing dams and associated analytical studies to investigate mechanisms of erosion and piping, exclusively associated with seepage control measures. He compiled useful information pertaining to some 30 case studies of dams that have had seepage control measures installed and operative for over 10 years.
Most of the dams in his study were founded over native soils with variable depths of deposition. The depositional environments are inherently diversified, covering wide spread origins including alluvial, fluvial, glacial, and aeolian. Seepage analyses were performed on selected dams from this study to assess the changes that occur in the flow regime due to installation of the seepage barrier. The purpose of these analyses varied depending on the observed performance and thus was aimed at providing a better understanding of the foundation behavior of dams with seepage barriers. Finite element seepage analyses were also performed to investigate the effects of cracks in seepage barriers, the hydraulic conductivity of the surrounding soil, and the differential head across the barrier. Based on the findings from the case studies and analyses, potential failure modes specific to dams with seepage barriers were identified, and the sequences of events required for the propagation of these failure modes were developed. The observations and insights acquired in this study were distilled into conclusions regarding the long-term performance of dams with seepage barriers.
Presented hereunder are reviewed summaries of two specific case histories from Rice (2007). These were selected out of conditional similarity to the present case
34
study namely; the Virginia Smith Dam and the Manasquan Dam. Both are also included among those selected for detailed seepage analyses in the referred dissertation; thereby offer a comprehensive related assessment on the objectives.
Virginia Smith Dam (formerly known as Calamus Dam with reference to its location on the Calamus River about 9.65 km northwest of Burwell, Wyoming) is a
2212.5 m long zoned earth embankment and has a structural height of 29 m. The foundation conditions are a mixture of aeolian sand dune deposits and fluvial deposits up to 36 m thick. An 2.5 m thick seepage blanket of Zone 1 material (specified as a mixture of sand, silt, and clay with a minimum of 50 percent fines, compacted in 150 mm lifts) extends from the core 244 m upstream of the dam centerline. At a location
91.5 m from the dam centerline, a soil-bentonite slurry trench cutoff wall extends down from the seepage blanket as an accompanied seepage barrier.
Inclusion of the cutoff wall as an additional seepage barrier was anticipated to reduce the amount of seepage beneath the dam as well as to intercept any high permeability channels that may exist in the near-surface alluvium that could result in concentrated seepage pathways near the top of the foundation. It was embedded into the bedrock, ensuring a positive cut-off, along the right side of the foundation, whereas it was partially penetrated into the soft foundations toward the left side. The width of the trench varied from a maximum of 1.5 m for the fully penetrating to a minimum of 0.9 m for the partially penetrating portions. The width was designed to limit the hydraulic gradient across the wall, to prevent blowout of the barrier fill into the adjacent soil, and the results of seepage analyses.
35
A brief assessment resulting from the instrumented observations as presented in the case study by Rice (2007) indicated that the overall performance of the seepage control elements of the dam (specifically the seepage blanket and the seepage barrier) had not changed significantly over the life of the dam. The initial readings of the vibrating-wire piezometers indicated that the fully penetrating portion of the seepage barrier was effective in reducing piezometric heads across the barrier, while the partially penetrating seepage barrier was largely ineffective in this regard.
Unfortunately, these piezometers were all rendered inoperable within a few years after construction. Thus, long-term changes in the areas immediately around the seepage barrier could not be assessed.
According to Rice (2007), several piezometers in the embankment and foundation experienced some anomalous behavior over the life of the dam. Rises and drops of up to 4.5 m were recorded in piezometers before they returned to the previously recorded water levels. This behavior was observed in both embankment and foundation piezometers. The durations of these changes ranged from a few months to several years. In a few cases the abrupt changes had not returned to the original levels. Isolated perturbations of such incidence occurred in only a small percentage of the instruments. A further insight of the observations by Rice (2007) revealed that this may be the result of localized changes in seepage resistance either upstream or downstream of the measurement points. These changes may also be due to redistribution of fines within the soil matrix (a process often referred to as suffusion) that results in a higher or lower permeability along the critical seepage path. The locations where such changes take place are impossible to ascertain and,
36
thus, it is not possible to discern whether these perturbations are associated with the seepage barrier performance or another portion of the dam.
Manasquan Dam consists of a 1494 m long and 16 m high homogeneous earth embankment with a soil-bentonite seepage barrier as the primary means of seepage resistance. The dam is located in an area underlain by coastal plain sediments. The base of the embankment is in contact with a predominantly sandy upper stratum and a lower stratum that consists of interbedded layers of silt, sand, and clay. It was planned to tie-up the seepage barrier with the underlying clay layers throughout the embankment profile. During construction, it was discovered that the clay layer had been mined under the central portion of the embankment footprint and the clay blanket that the seepage barrier was to be keyed into was missing in this area. The dam foundation was modified where the clay had been excavated by constructing a 3 m thick imported clay layer beneath the dam crest and a 1.5 m thick clay blanket extending upstream for the extent of the mining.
The seepage barrier thus varies from three to five feet wide, has a top elevation of 33 m (1.5 m above the normal pool elevation) and keys a minimum of 1.5 m into the natural clay layer or the compacted clay repair in the mined area. By keying continuously into the clay layer a complete cutoff is achieved by the barrier and the efficiency of the barrier is not diminished by the effects of seepage around the ends of the barrier. The ends of the barrier are embedded into the flat-lying abutments so that seepage around the barrier is either driven by low hydraulic pressure (in the upper portions of the reservoir) or has a long seepage path (in the lower portions of the reservoir).
37
Rice (2007) has described that the seepage barrier in Manasquan Dam performed well from the first filling of the reservoir and no significant changes in the water pressure regime or seepage rates occurred over the life of the structure. The calculated head efficiencies (head drop over the barrier divided by the total head drop across the dam) range from 81 to 89 percent. The head efficiencies tend to be lower where the barrier is narrower and where the height of the embankment and the hydraulic gradient across the barrier are greatest.
Rice (2007) has attributed the good immediate and long-term performance of the seepage barrier to: (1) keying the seepage barrier into a continuous low- permeability layer, (2) a design that provides several layers of defence against the development of leaks in the barrier, and (3) quality control during construction.
Rice (2007) performed seepage analyses on these dams in order to assess effectiveness of the seepage control measures. The properties and issues specifically assessed through the seepage analyses included permeability ratios (between effective and design permeabilities) and exit gradient reduction.
Rice (2007) has reported that the range of calculated effective hydraulic conductivity values for Virginia Smith and Manasquan Dams are not much different from their design hydraulic conductivity values. The seepage analyses performed on both dams indicated that order of magnitude changes in the effective hydraulic conductivity of the seepage barrier had small effects on the calculated water pressure regimes. According to Rice (2007), the ability of the models for these dams to back calculate effective hydraulic conductivity presented a relatively low level of precision,
38
largely associated with the thickness of permeable layers and their embedment conditions at the base of the barrier (either clay or tight sandstone/claystone bedrock).
The hydraulic gradient values calculated by Rice (2007), ranged from 1.2 to
5.2, all of which could be considered erosive to soils if not protected by effective filtering. He also reported that the wide range of calculated hydraulic gradients in
Virginia Smith and Manasquan Dams are due to considerable differences between the configurations of the dam and seepage barrier in the cross sections analyzed. In both dams the high hydraulic gradients were calculated within low permeability soil or bedrock layers that are adjacent to soils that act as effective filters should erosion initiate due to the gradients. The higher hydraulic gradient was calculated in Virginia
Smith Dam where the seepage barrier completely penetrated the permeable layers and was embedded into low-permeability bedrock, and the lower gradients were calculated where the barrier partially penetrated the permeable layer. In Manasquan
Dam, the lower hydraulic gradient was calculated where the bottom of the barrier was embedded in a thick clay layer and the higher gradient calculated where the barrier was embedded in a thin clay layer underlain with permeable soil, thus adding a vertical component to the hydraulic gradient.
Based on the analyses, Rice (2007) concluded that the magnitude of the hydraulic gradients is a function of the following: (1) the differential hydraulic head across the barrier, (2) the depth of embedment into a low permeability layer, (3) the ratio of hydraulic conductivities of the permeable layer being cut off and the low permeability layer, (4) the configuration of the seepage barrier and other soil layers in the dam cross section. As stated by Rice (2007), the seepage barriers in Virginia
Smith and Manasquan Dams were built in the original construction and no
39
comparison could be made with a pre-barrier condition regarding the change in exit gradient due to the construction of the seepage barrier.
2.4 SUMMARY
Peck (2000) presented the application of risk/statistical analysis with appropriate site-specific engineering appraisals as the mainstream approach for foundation evaluation of dams in spite of the limited scope of investigations. Talbot et al. (2000) briefly describe the usual remediation of seepage problems. Elkateb et al.
(2003) highlight aspects of parametric quantification in difficult foundation scenarios.
These references highlight the intricate task of adequately defining the foundations as best representative of the actual problem domain to be undertaken prior to the actual modelling and sensitivity analyses. Some of the commercially available 3-D seepage analysis softwares have source codes equipped with similar parametric estimation techniques. However, the general approach in current practice (preferring a 2-D approach) is to adopt a conservative and safer but relatively limited approach, usually governed by time constraints. As such, 2-D seepage analysis models for heterogeneous foundations either use a homogeneous foundation with a precautious
K-value/K-function associated with it or a two-layered approach, i.e. a dominant foundation layer with a relatively pervious zone in-between. The randomness of depositional environments related to most heterogeneous foundations is not addressed as such for conventional 2-D seepage analyses.
Jafarzadeh and Soleimanbeigi (2005) conversed on the uncertainties in seepage analyses. An essential appraisal of the chronological ‘history of developments’ on a major foundation seepage hazard ‘piping’ is given by Richards
40
and Reddy (2007). These references have indicated the multivariate sensitivity of foundation seepage to different parameters of interest, the hydraulic gradient being of critical importance. Foster et al. (2000b, 2002), Fell et al. (2003) and McCook (2004) are listed among the recent works on piping. They establish different modes and mechanisms along various interfaces (structure-foundation or foundation-foundation being of significant relevance to the current study) mostly in relation of the critical exit gradient. It is still not clear as to where and to what limits the exit gradient or any other parameter has to be controlled. Richards and Reddy (2007) maintain a lack of methods for evaluating piping in cohesionless soils, which actually govern most of the embankment volumes and sometimes even the variant depositional constituents forming heterogeneous foundations.
Typical practical consideration regarding primary measures of seepage control through the foundations thereafter cites Mukhopadhyay (2008) in regards of a cutoff wall and Gohernejad et al. (2010) for an impervious upstream blanket. The remainder of this literature review presents the published literature identified that directly relate to experiences and performance evaluation in relevance to this study. There is limited published literature consecutively related to the investigation, analysis, design and performance evaluation for a combination approach of seepage control through deep pervious foundations. Usually this is due to the availability of bedrock or an impervious stratum at shallow depths for a positive cut-off. Occasionally, when neither of these is the reason, ample area for placing an impervious upstream blanket is used to advantage.
41
CHAPTER 3.
DESCRIPTION OF CASE STUDY
3.1 PROJECT DESCRIPTION
3.1.1 General
This research study is based on evaluation of Satpara Dam Project. Satpara
Dam is a multipurpose project, located at Longitude: 75º 46’ 12” E and Latitude: 35º
16’ 12” N on Satpara Nullah, about half a kilometer downstream and north of the existing Satpara Lake and about 6 km south of Skardu town along access road to
Deosai plain (226 Km and 760 Km from Gilgit and Islamabad, respectively). Figure
3.1 and Figure 3.2 illustrate the overall location and layout of the project site. A 40m high, zoned earth core rockfill dam was conceived here, for which the Satpara Lake would be developed and used as a reservoir with active and dead storages of 67.5 and
47.5 million cubic meters (MCM) respectively. This project is meant to provide 115
MCM of water for irrigation of 6300 hectares of land, 12000 m3/day water supply to local population for civic purposes and power generation of 17.16 Megawatts.
3.1.2 History of Development and Present Conception
Satpara Dam was initially constructed about 45 m downstream of Satpara
Lake as an earthen embankment, with a spillway of stone masonry in lime mortar. Ali
Sher, a local Raja of Skardu, constructed this dam during early seventeenth century for irrigation and water supply in the neighborhood. The embankment, after a long good service of about 100 years, was breached and the spillway damaged due to flood and lack of maintenance. Northern Area Works Organization (NAWO) re-constructed a 6 m high earth fill embankment over the remnants of the old earthen dam in 1976.
42
To impound water and utilize the same, a temporary arrangement of check gates through a channel along the upstream of the embankment was devised and put to use.
On authorization from the Northern Area Public Works Department
(NAPWD), Planning and Investigation (P&I) Division of Water And Power
Development Authority (WAPDA), carried out initial field investigations and planning studies of Satpara Multipurpose Project in 1988. Later on in 2001, through a cabinet decision, Government of Pakistan included the project in Vision 2025
Programme. Accordingly, Hydro Electric Power Organization (HEPO) of Water And
Power Development Authority (WAPDA) prepared detailed design and tender documents in 2002 for a 40m high ECRD (Earth Core Rock fill Dam, i.e. a zoned embankment dam) downstream of the old dam and using the old embankment as cofferdam. Figure 3.2 shows the project layout plan and the historical chronology.
Subsequently, a joint venture of consultant firms (Satpara Dam Consultants,
SDC) was appointed towards the end of 2002 for review of the design and construction supervision of the project. Design review was completed in 2004 and construction drawings were prepared. Dam construction continued from 2004 to 2011.
Partial reservoir impounding was initiated in 2007 during the progress of construction works due to political restraints. The writer’s association with the joint venture has enabled access to the project data, the dam site and necessary experimentation for the present study. The main project features are shown in Figure 3.3.
43
Figure 3.1 Project location map
44
Figure 3.2 Project layout plan
45
Figure 3.3 Main project features
46
3.2 PHYSICAL INVESTIGATIONS
3.2.1 Exploratory Efforts
A comprehensive assessment of the foundation strata was carried out over three phases of exploratory efforts, spanning from the initial conception of the project in 1988, covering the design phase in 2002, and extending over to the construction phase in 2004. Figure 3.4 presents the different features related to these investigation phases.
During the first phase of feasibility study in 1988, Planning and Investigation
Division of WAPDA (P&I) and Central Design Office (CDO), WAPDA were involved in the initial exploratory efforts for conception of the project. The investigation works associated with this period included percussion drilling of six boreholes and 13 test pits on the existing embankment and upstream and downstream of the embankment.
The second phase commenced prior to the tendering stage for design, wherein another program of subsoil investigations was sponsored by HEPO, WAPDA, conducted and reported by Central Material Testing Laboratory (CMTL), WAPDA in
2003. Surface geological mapping of the project area was covered during this phase of investigations. The proposed investigations in this phase also included vertical drilling of 7 boreholes of varying depths at different locations using rotary rigs, 8 test pits and
5 test trenches. Drilling of exploratory bore holes was performed by Dams
Investigation Division of HEPO whereas the assignment pertaining to pitting and trenching were carried out by CMTL, WAPDA.
47
Figure 3.4 Subsurface investigation plan – Phase wise distribution
48
To strengthen already available geologic information, some additional investigations were framed by SDC, the Project Consultant’s, during 2004 through the design review and initial construction stage. The proposed additional investigations included vertical drilling of 5 more exploratory holes of varying depths using rotary rigs and excavation of another series of 9 test pits. A detailed description from the above mentioned exploratory efforts is presented in the following sections.
3.2.2 Surface Geological Mapping
Surface geological mapping of the project area was carried out in Phase II of the exploratory efforts. The topographic maps used were based on topo-survey carried out by Topo Division of Planning and Investigation Directorate in 1988. Surface geologic details were picked up during the mapping using telescopic alidade and plane table. Major findings from the surface mapping are discussed hereunder:
Satpara Dam Project has been proposed in an area with exposures of barren rocks together with glacial deposits, topographically characterized by a deep cut U- shaped valley with steep slopes on both sides, amid rugged mountains. The main dam is proposed over the Satpara nullah, which flows from south to north through the valley and is a left tributary of Indus River.
Satpara Nullah originates from an elevation of 4440 m amsl and its confluence with river Indus is at an elevation of 2167 m amsl near Skardu having a total mainstream length of 34.5 km. The whole catchment of nullah comes under snow cover during winter. Satpara Nullah has a natural lake, about 76 m. deep, located 6
49
km south of Skardu town along access road to Deosai plain. The dam axis is proposed over the Satpara nullah slightly downstream of the Satpara lake.
The Satpara Lake in the valley is an outcome of glacial activities from the past. It might have originated after a descending glacier from the high altitude Deosai plains was dammed up when encroaching upon the flowing stream of the Satpara
Nullah. Satpara village is situated upstream of Satpara Lake along the left bank of
Satpara nullah.
Cirques and lateral moraines are present throughout the valley. Apart from the glacial characteristics, the valley shows talus/scree accumulations and colluvial/alluvial terraces, pointing out a possibility that the valley is a modified glacial valley, where glacio-fluvial and colluvial activities have occurred. The rocks exposed in the project vicinity are mainly slates with intrusive igneous rocks (Biotite
Granodiorite). An island in the lake however, indicates the presence of a more resistant and hard rock “Hornfels”.
The area is generally disturbed but no evidence of any fault or major discontinuity was observed in the study area. The main stratigraphic units exposed within the project area are described below:
Overburden: Soil units present in the project area as overburden are differentiated on the basis of their composition. Different types of overburden are described as under:
The Satpara Nullah has a steep gradient downstream of the old dam/cofferdam
and its bed contains large boulders of various sizes, sub-angular to sub-
rounded with occasional gravels, cobbles and little alluvial sand.
50
The material presents on the hill slopes and at the toe of the hills are slope
deposits / scree, consist of angular boulder, gravels with appreciable amount
of fines throughout the project area on both sides of the nullah. These are most
conspicuous products of the physical weathering of cliffs, scarps and hillsides
in the valley.
Thick moronic material is lying at different levels throughout the project area.
Its main concentration is distributed across both sides of the nullah, below the
Satpara lake on left bank and from mid of the lake to downstream end of the
right bank. It includes mixture of silts, clay with gravels/boulders and sand in
varied proportions. These deposits are loose the semi-consolidated having
steep slopes. Gulies, slides and slips are common in this area.
A 0.3 to 0.6 m thick layer of clay has been found downstream of the lake. This
layer is also seen in the cuts within the existing embankment/cofferdam area.
The clay layer has either been deposited by the lake which might had quite a
large extent in the past and receded with the passage of time or has been
deposited by the outwash with emerges along the ice front of glaciers.
The clay layer upstream of the existing embankment is not met with; probably
it has been washed away.
Rock Units: Main rock units exposed in the project area are described as under:
Slates: The rocks exposed on the left hills of the valley near the lake are
slates. The Slates are brownish grey to grey and greenish grey in colour, fine
grained and moderately hard. The rock exposures are more than 305 m. away
from the left bank of the embankment. The slates are imperfectly inter-locked
and break easily along cleavage planes which are poor to moderately
developed.
51
Granodiorite: This is an intrusive igneous rock and is exposed on right
abutment above the dam crest level, in the downstream area along the right
bank of nullah and at higher elevation away from the nullah in rest of the area.
It is coarse grained, crystalline and medium hard to hard, light grey to whitish
grey and greenish grey in colour with brownish weathered surfaces. Joints are
present in the rock, which are open to tight, well developed and few are filled
with weathered material.
Hornfels: The country rock, slates near the contact with granite has been
converted into a more resistant rock “Hornfels” which is hard to very hard,
due to contact metamorphism. The hornfels rock is brownish grey to dark
grey, fine grained and compact. The exposed Hornfels rocks in the lake island
suggest that granite and hornfels rock contact may be present midway in the
valley section of the embankment under the glacial drift deposits. However,
since the glaciers generally transport distant material, the possibility of
volcanic rocks transported from other areas can also not be ruled out.
3.2.3 In-situ Testing
The scope of in-situ testing when related to underseepage potential for the foundation has been restricted in this study to permeability tests only. In-situ testing for permeability was performed, in the drill holes and selected test pits, at an interval of 3m or at change of strata in overburden for assessment of the foundation strata.
These tests were carried out utilizing standard field test methods as falling head tests in accordance with Lambe and Whitman (1979) and constant head injection test
(ASTM D4630-96).
52
3.3 DETAILED SUBSURFACE EVALUATION
The program of investigations was conducted through phases of evolution of the project as described above. The detailed log reports have been annexed at the end, for ready reference; these logs were reproduced in a tabular format from actual log- reports of each investigation phase, available with SDC.
3.3.1 Phase-I (Feasibility Stage - 1988)
Phase-I investigations included six boreholes and 13 test pits located as shown in Figure 3.4. All boreholes were drilled with percussion rigs. Boreholes BH-1, BH-2 and BH-3 were drilled in the Nullah bed, to a depth of 18 m each, to get subsurface information downstream of the existing embankment. The subsoil thus evaluated comprised of predominantly sandy strata with little silt, occasional gravels and some boulders.
Boreholes BH-4, BH-5 and BH-6 were drilled upstream of the existing embankment. BH-4 was drilled to a depth of 9 m in the nullah bed. BH-5 and BH-6 were drilled on the right and left sides of the nullah respectively, to depths of 6 m each. The subsoil evaluated similarly consisted of sand with little silt, occasional gravels and some boulders. Silt and boulders generally appeared to reduce towards the downstream; however, bedrock was not reached in any of these probings.
Thirteen (13) test pits covered the valley section. The pits P1, P2, P5, P6, P10,
P11, P12 and P13 were dug in the existing embankment. P8 and P9 were excavated to explore upstream area of the embankment while P4 and P7 were located on the downstream side. Groundwater level was only recorded in P9, as 1.8 m (EL. 2636.215
53
m). The predominant stratum evaluated from these pits was sandy silt. P2 and P7 revealed clay lenses between 4.5 – 9 m in P2 and between 2.4-3.35 m in P7, respectively. P6 and P7 also show appreciable sand contents, whereas P5 and P6 indicate openwork gravels between 1.8 - 2.7 m in P5 and 4.5 - 6 m in P6.
Field permeability tests were conducted in boreholes BH-1, BH-3 and BH-6.
The analysis revealed that permeability values as 1.3 x 10-1, 1 x 10-3 and 9.5 x 10-2 cm/sec, respectively. The field permeability tests were also conducted in the test pits at various depths. Permeability values are of the order of 1.68 x 10-1 to 2.0 x 10-1 cm/sec. However, in pit P4, it is 3.027 cm/sec. this may be due to the presence of large boulders in the vicinity. The results of these tests indicate that the material is fairly pervious and a representative value maybe of the order of 10-1 cm/sec near the surface, decreasing at lower levels to an order of 10-3 cm/sec.
3.3.2 Phase-II (Design/Tender Stage - 2003)
A total of 7 boreholes were drilled during this phase of investigations. All the boreholes were drilled vertically, using Rotary Rigs. Location of these boreholes is also shown in Figure 3.4.
Out of these, three boreholes namely SDR-1, SDR-2 & SDL-3 were drilled at dam axis. Two bore holes SDR-1 (El. 2629.619 m) and SDR-2 (El. 2636.825 m) have been drilled along the dam axis on the right abutment and bed rock was encountered at 49.4 m and 19.7 m depths (2580.114 and 2617.683 m amsl) respectively. The strata mainly comprised of boulders with traces of fines and occasional gravels. Average
54
groundwater recorded in SDR-1 was found at 2.06 m depth (EL. 2627.376 m) whereas in SDR-2 it was at 10 m depth (EL. 2626.766 m).
Thirteen (13) permeability test attempts were associated with SDR-1, out of which 2 were water pressure tests performed in bedrock. Permeability is of the order of 1.5 x 10-1 to 1.3 x 10-2 cm/sec in SDR-1, whereas water pressure tests performed in borehole SDR-1 show Lugeon1 values of 9.2 and 6.6 (equivalent K-values being
1.196 x 10-4 and 8.58 x 10-5 cm/sec).
Nine (9) tests were performed in SDR-2, out of which one test resulted in 100
% water loss. Out of the rest, 6 tests were water pressure tests performed in bedrock with equivalent K-values of less than .001 cm/sec resulting in four. Maximum and minimum Lugeon values in bore hole SDR-2 are 51 and 28 (equivalent K-values being 6.63 x 10-4 and 3.64 x 10-4 cm/sec), with an average Lugeon value of 42.25
(equivalent K-value = 5.493 x 10-4 cm/sec). The remaining tests indicate that foundation permeability is of the order of 3.1 x 10-1 to 2.8 x 10-1 cm/sec, showing a relatively permeable stratum.
One bore hole SDL-3 (EL. 2665.781 m) has been drilled on the left abutment and no bedrock is met with down to explored depth of 79.86 m. The strata comprised boulders, cobbles and gravels. Upper levels of the probing also indicated the presence of some silt contents. Groundwater has been encountered at 28 m (EL. 2638.044 m).
1 Under ideal conditions (i.e., homogeneous and isotropic) one Lugeon is equivalent -5 to 1.3 x 10 cm/sec (Fell et al., 2005).
55
Permeability value ranges from 8.87 x 10-5 cm/sec (at 5.5 m depth) to 3.65 x 10-3 cm/sec (at 13.8 m) after which no permeability tests could be performed, as the head could not be maintained in the borehole. The last 2 out of a total of 6 permeability test attempts resulted in 75 % water loss.
Two other boreholes, SDL-5 for spillway (proposed to be located on left bank of Satpara Nullah adjacent to the dam axis, EL. 2672.971 m) and SDSB-1 (EL.
2630.515 m) for the spillway stilling basin, were drilled towards left bank of nullah.
These boreholes were drilled down to 29.8 m (EL. 2642.921m) and 25 m (EL.
2605.522 m) depths, respectively. Groundwater level was encountered in both boreholes at 26.5 m depth (EL. 2646.578 m) in SDL-5 and at 3.7 m depth (EL.
2626.766 m) in SDSB-1.
In SDL-5, a distinct lense of clay was encountered at depths 3.35 - 5.5 m whereas the remaining strata mainly comprised of silty / gravelly boulders. In SDSB-
1, the subsoil consists of angular to sub angular gravelly boulders, mainly of granodiorite and slate with varying percentage of clayey silty sand.
The permeability ranges from 4.9 x 10-2 cm/sec (5.18 m depth) to 6.5 x 10-3 cm/sec (at 14 m) in borehole SDL-5 from 4 successful test attempts. 2 test attempts failed subsequently at depths of 16.15 m and 26.5 m with 100% water loss. A total of
6 test attempts were associated with borehole SDSB-1, which resulted in variable permeability values from a maximum of 3.2 x 10-1 cm/sec to 9.3 x 10-3 cm/sec at different depths. This indicated no depth dependence for the in-situ down-hole permeability of the strata.
56
One borehole, SDR-6 (BHTGO-1) for Irrigation outlet was drilled on the right bank of nullah, at EL. 2629.793 m. In this bore hole, rock was not encountered up to
40.7 m depth. The drilling results indicated presence of thick heterogeneous material containing boulders, with varying amount of cobbles, gravels, and silt/sand. Ground water was encountered at 5.19 m (EL. 2624.633 m). Permeability values in general ranges from a maximum of 3.8 x 10-1 cm/sec to a minimum of 1.5 x 10-3 cm/sec, with no depth dependence for 8 out of 10 test attempts. The remaining two tests failed with
100% water loss, again without any depth dependence, suggesting unpredictable permeabilities in the glacial deposition of foundation moraines.
An additional borehole, SD-1 was drilled for subsurface evaluation at the cofferdam (proposed at the existing embankment) and slurry trench wall. The maximum-drilled depth was 18.3 m. Average groundwater level was at 1.03 m depth
(El. 2632.558 m). The sub-soil consists of sandy gravels with little silt and some boulders. Three successful permeability tests carried out reveal that permeability is of the order of 1.96 x 10-1 cm/sec (4.57 m depth) to 1.9 x 10-2 cm/sec (at 14 m), which shows that the material is fairly pervious and that depth barely affects the down-hole order of magnitude of the permeability.
Five (5) test pits were proposed to investigate the nature of overburden material at various locations along dam axis and spillway. Out of these, three pits were dug at the dam site. Two pits (TP-2 and TP-2A) were located near dam axis, towards the right and left abutments respectively, while another pit TP-1 was 22.86 m d/s of SDR-1. The material encountered was silty gravel and boulders. Groundwater level was recorded in TP-1 at 3.65 m depth (El. 2654.808 m).
57
Two pits (BAP-1 & BAP-2) were excavated to probe the formation of the spillway area. One pit was located upstream and other was located downstream of spillway axis. The material encountered was angular to sub angular gravels with boulders and varying percentage of clayey silt.
In order to establish the foundation condition of the future reservoir, pits and trenches were excavated during the resent phase of investigation. Three pits (TP-3,
TP-4 & TP-5) were excavated on the left of the Satpara Nullah. The material evaluated through these pits was predominantly clayey silt, with gravels and boulders.
No pit was excavated on the right bank area; however five (5) test trenches covered the right bank and the approach area for the proposed spillway. Two trenches along the right bank / wraparound (TRB-1 & TRB-2) and three around the proposed spillway approach, on the left bank (TLB-1, TLB-2 & TLB-3) were excavated. The material encountered in trenches mainly composed of gravels and boulders of varying sizes with clayey silt / silty clay.
3.3.3 Phase-III (Construction Stage – 2004)
Five (5) additional boreholes of varying depths were drilled at different structures in this phase of investigations (Figure 3.4). Foundation insight from these explorations, although no different than the previous, was deemed necessary during this phase of study for relative evaluation.
Borehole SBH-1 was drilled at the dam axis, the strata encountered comprised of gravels in silty matrix with some boulders. SPD5A was located along the spillway axis; it showed dominant boulders with gravels and silty sand. Groundwater was
58
encountered in SPD-5A at an average depth of 44.8 m (El. 2626.766 m). Permeability tests were performed in this borehole below 39.624 m depth. The resulting permeability values range within 1.23 x 10-1 cm/sec (at 54.8 m depth) to 9.49 x 10-4 cm/sec (at 78 m depth).
SBH-2 was drilled downstream of the cofferdam for the cutoff wall. It indicated mainly boulders and gravels in silty matrix. However, a dominant layer of clayey silt with boulders and gravels was also depicted in the upper 3 m. Groundwater was encountered at a depth of 9.10 m (El. 2628.9 m). Successful permeability tests resulted in a range within 3.41 x 10-2 cm/sec (7 m depth) to 1.36 x 10-3 cm/sec (at 3 m depth), with no depth dependence.
SBH-3 was drilled at the left abutment of the cofferdam, and SBH-4 was located at the intake structure. SBH-3 revealed boulders in silty matrix whereas SBH-
4 indicated mainly gravels in silty matrix, occasionally with boulders. However, a dominant layer of clayey silt with gravels and boulders was also depicted in the upper
3 m. Groundwater was encountered at an average depth of 4.64 m (El. 2628.9 m). In both of these boreholes, 24 and 19 permeability test attempts were undertaken respectively. The permeability values lie from 6.92 x 10-2 cm/sec to 2.35 x 10-3 cm/sec for SBH-3. For SBH-4, permeability values range between 1.35 x 10-1 cm/sec to 2.05 x 10-3 cm/sec. 6 out of 24 and 8 out of 19 tests failed with 100% water loss in both these boreholes with no specific depth dependence.
Test Pits SB-1, SB-2, SB-3, SB-4, and SB-6 were located towards the left edge of the impervious blanket. SB-5 was shifted slightly downstream, to get an insight towards left abutment of the spillway. SB-7 was dug in the valley whereas SB-
59
8 and SB-9 were oriented towards the left edge of the nullah bed. The strata generally varied from gravelly silt to silty gravels. SB-3 specifically encountered boulders below 0.91 m depth whereas the presence of boulders in vicinity of all other pits, except for SB-1 and SB-6, was supplemented from 100 % water losses encountered in the permeability test attempts. The permeability values generally lie within a maximum of 7.38 x 10-2 cm/sec (SB-2; at 1.83 m depth) to a minimum of 3.17 x 10-3 cm/sec (SB-9; at 1.83 m depth).
3.4 DESIGN AND MONITORING PRIOR TO STUDY
The project design review from 2003 to 2004 was mainly focused on adequacy checks in regards to the foundation seepage control, specifically governed by site conditions and serious time constraints. A conservative and safer but relatively limited approach was used for conducting the 2-D seepage analyses. As such, some of the critical modelled parameters like the upstream impervious blanket, partial cut-off wall, contributing foundation depths were considered invariable.
The heterogeneous foundations were modelled used a two-layered approach.
The selected K-values for the dominant foundation zone was an average of the test results and a pervious zone was modeled with a higher K-value selected from literature corresponding to open-work zones/gravels in general. The acceptability of the designed features was evaluated through comparison of actual (single value) resultant ‘exit gradients’ from the analyses to a standard conventional and established value of unity, i.e. 1.0, usually reported in literature for initiation of internal erosion in gravels.
60
The instrumentation record was available with the project site staff, since pre-impounding in July 2006. Monitoring and performance evaluation at Satpara Dam was independently done by the field staff, through spot checks on the observed instrumented data. Different teams of experienced representative personnel from a
PoE, DSO (WAPDA), and M&S (WAPDA) were also frequently involved in this process. A final conclusive and comprehensive review of the design, construction and operational performance to date and an on-site inspection of Satpara Dam was conducted and completed by a team of independent Dam Safety Specialists at the request of USAID in January 2011.
In general, field spot checks for performance evaluation were probably oriented at identification of satutrated zones or seepage faces along the downstream of the main dam structure. Conclusive remarks from the independent experts were, perhaps, based on comparison of some instrumented values from selected piezometers with their most likely values, in light of an experienced engineering judgement.
However, the exact performance evaluation details are not available.
61
CHAPTER 4.
METHODOLOGY
4.1 INTRODUCTION
This chapter is sub-divided into several sections. The first section describes general data collection, followed by the next section briefly describing methods of estimating valid / realistic foundation permeability (K) values and including an eloquent description of K-modelling using a geological software package
“RockWorks”. The third section explains seepage analyses, highlighting salient inputs, the sensitivity criteria, variables and pertinent scenarios. The fourth section subsequently explains special case considerations for the multivariate seepage analyses conducted as part of this research. Presented next is the methodology of performance evaluation through data analysis of instrumented observations, followed by the last section encapsulating uncertainty propagation analyses to contemplate the post-construction foundation response.
4.2 DATA COLLECTION
In view of the prevailing heterogeneous foundation characteristics, quantitative hydraulic conductivity data, for the foundations at Satpara Dam Project, was collected from the results of field permeability tests, performed in-situ during the exploratory efforts. The previous chapter has described in detail the different phases of exploratory efforts and the respective inferences in this context.
Pertinent data collected from the project investigation reports in regards to the foundation characterization included the specific locations, co-ordinates, test depths, material logs and resulting permeability values corresponding to each reported
62
permeability test. Construction material permeabilities were generally investigated through in-situ field tests in test pits. Some permeability values, like those of the filter zones, were assessed from grain size or soil descriptions. Material zones for critical features, like the impervious material for core and blanket, were also tested for permeability values on undisturbed and disturbed samples in the laboratory.
Data collected included the representative permeability values as given in the test reports for the project, i.e. HEPO (2002) and CMTL (2003). Other important data collected includes typical project drawings (plans and sections), post construction instrumentation records, and relevant operational history of the project.
4.3 ASSESSMENT OF FOUNDATION PERMEABILITY
Hydraulic Conductivity is the most common material property associated with seepage analysis and control. It is often used interchangeably with the term
‘Permeability’, which is taken from the more precise coefficient of permeability (K).
4.3.1 Foundation Permeability Profiles
By virtue of similar deposits, the foundation permeability conditions noted throughout the field-testing are taken to represent the entire area. To facilitate understanding of the foundation and subsequent design of seepage control measures, attempts were initiated to develop an idealized permeability profile representative of the foundations. Variation of permeabilities through the three phases of investigations was prerequisite for a probable conception of the sort.
63
Logical permeability trends were subsequently sought for possible layers or zones in the foundation in lateral and longitudinal directions. These efforts involved
K-El. (Permeability-Elevation) plots along the main dam as well as along the cofferdam axes and K-D (Permeability-Depth) plots. The K-D plots included phase wise distribution of the gathered permeability data and a variant to evaluate the distribution with respect to the nullah proximity, through the river valley. Best estimates of the recorded values and their uncertainties, based on both statistics and judgment were also explored. These included comprehensive frequency analysis for the K-Test results and percentile distribution of permeability ranges corresponding to the tested depths. A typical K-D plot is shown in Figure 4.1.
Depth (m) 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 100
10
1
0.1
0.01
0.001 Permeabilities (cm/sec) Permeabilities 0.0001
0.00001
Figure 4.1 Typical K-D (permeability vs. depth) plot output
4.3.2 Foundation K-Modelling
The available K-data work sheet from results of the in-situ field testing was exported to RockWorks2006, a version of RockWare’s integrated software package
64
for geological data management, analysis, and visualization. This software was utilized to interpolate a 3-dimensional solid model representing the downhole point- sampled data as control points for its component voxels (volumetric pixels, or more correctly, volumetric picture elements).
The Closest Point Algorithm (CPA) was utilized for generating the foundation model for analysis, whereby the solid model voxel nodes would specifically honor the control points. The Closest Point option sets the value of a voxel node to be equal to the value of the nearest data point, regardless of its distance from the point or the value of its other neighbors.
In view of the apparent scatter in the available permeability data, it was preferred to use all of the available data points when computing a voxel node’s value for creating a site specific 3-dimensional solid model. Using the CPA also ensured exact correspondence between the maximum and minimum model node and data point values along with equivalent ranges.
The Rockworks sub-menu P-Data / Section helped display the model as a 2- dimensional (flat) vertical profile composed of multiple panels, sliced along the desired cross section to be modeled in the study area. The same sub-menu was also useful in filtering the data over a number of user defined options, suited to match the desired outputs, through a P-Data Track tab. An expanded view of the “P-Data
Section Options” dialog box is shown in Figure 4.2 and typical plan and sectional outputs from Rockworks are presented in Figure 4.3.
65
Figure 4.2 Dialog Box “P-Data Section Options”
66
(a)
Note that both extremes are not shown in this section
Minimum permeability Maximum permeability (low extreme) was not (high extreme) has been encountered in this neglected in this section section
(b)
Figure 4.3 Typical outputs from Rockworks in RockPlot2D: (a) Plan view specifying orientation of the modeled section, and (b) The modeled section.
67
The 2-dimensional crosssection through the 3-dimensional solid model created utilizing Rockworks could be exported in various external file formats. A DXF format was selected as compatible to serve as a background reference for seepage modeling in SEEP/W as explained in the next section.
4.4 MULTIVARIATE SENSITIVITY ANALYSES
4.4.1 2D Seepage Modelling
Steady state ‘as designed’ 2-dimensional seepage analyses were performed in context of this study, using the finite element computer program SEEP/W. SEEP/W is a part of the GeoStudio suite of programs for geotechnical modeling, and specifically deals with seepage analysis. To find out an optimum solution for the seepage control along the moranic foundation deposits of Satpara, a scaled model of the dam section and its foundations was developed in SEEP/W.
The basic outline of the problem was initially sketched independently in Auto
CAD and imported in a DXF format for a background in SEEP/W. Different material regions were individually traced over the boundary outlines of the sectional imprint from the DXF file in the background. Finite elements and nodes were auto-generated at the end of each region, and the generated mesh was again automatically adjusted to correspond with the meshes for adjacent regions.
Sometimes, when the elements of a region were disturbed out of shape, the mesh pattern of such a region was visually adjusted in the Material Tab of the
Regions Property dialog box under Mesh Pattern as “Structured” or “Unstructured” to suite the condition. The boundary conditions assigned to the seepage models included
68
the total head (H) upstream of the dam and a default unit flux (q) with potential seepage face review along the outer boundaries of the chimney filter and the drainage blanket.
Figure 4.4 illustrates a typical model domain for the subsequent sectional analysis, individually highlighting the regions developed, the finite element mesh generated, the applied boundary conditions, and the material boundaries along with the material fill colors indicating assigned material properties.
In SEEP/W the hydraulic material properties must be defined as functions, not single point values. For each material type identified with a unique material number, a
Vol. W. C. (volumetric water content) function, a hydraulic conductivity
(Conductivity) function and the ratio of the Ky to Kx hydraulic conductivities were individually specified.
A Vol. W. C. Function corresponds to a volumetric water content function. In order to specify a function it had to be defined first. Volumetric water content functions corresponding to different materials were imported from the GeoStudio functions library. Suitable functions were selected based on the nomenclature assumed to represent the actual physical materials for the project as found on site.
Selected functions from the materials database file are shown in Table 4.1. The permeability / hydraulic conductivity inputs also require a function, describing its variation with changes in suction (negative pore-water pressure = suction). Actually measuring the hydraulic conductivity function is a time-consuming and expensive procedure. Relevant data for the appropriate input parameters for each material function as required in SEEP/W were not available.
69
(a)
Upstream BC Downstream BC
(b)
Upstream BC Downstream BC
(c)
Figure 4.4 Typical model domain setup for SEEP/W sectional, i.e. 2-D seepage analysis: (a) specified material regions (b) finite element mesh with upstream and downstream boundary conditions, and (c) assigned material properties with foundation representation from modeled section in ROCKWORKS.
70
Table 4.1 Selected functions from material database file
Allocated Fn No. WC Fn from material database of GeoStudio functions library Fn No. Description 1 8 Well Graded #2 2 10 Glacial Till (Uncompacted) 3 11 Glacial Till (Compacted) 4 16 Clay / Silt 5 19 Sand 6 20 Fine Sand
The hydraulic conductivity function was therefore specified for all materials
using the “Fredlund and Xing (1994)” predictive method built into the software. The
procedure involved estimation (just by clicking “Estimate”) using a measured
volumetric water content function and the saturated hydraulic conductivity.
The corresponding WC functions and the saturated hydraulic conductivity,
Ksat required as “K at saturation” in the Estimate Hydraulic Conductivity Function
dialog box, for each required material is listed in Table 4.2. The Ksat values were the
representative permeabilities selected after an assessment for the most reliable
estimates of hydraulic conductivity, based on the methods by which the hydraulic
conductivities were evaluated (refer to section 4.3). All SEEP/W input functions are
separately annexed at the end.
Once the volumetric water content function and a Ksat value were specified,
the hydraulic conductivity function was estimated. Some hydraulic conductivity
functions were adjusted / modified slightly from the estimated data points in order to
create a smooth function. This approximation was necessary to increase applicability
of the derived functions and suit the required input situation being modeled.
71
Table 4.2 Individual material WC functions and Ksat values for each K-Fn
K-Fn Material Description WC Ksat 1 Core / Blanket / Partial Cutoff Wall (Impervious Zones): IZ 4 10-6 2 Shell: S 3 10-5 3 Drain: D 1 10-3 4 Coarse Filter: CF 5 10-3 5 Fine Filter: FF 6 10-4 6 Homogeneous Foundation: H 2 10-4 7 Pervious Zone in Homogeneous Foundation: PZ 2 10-3 8 K-Zone 1 in Heterogeneous Foundation modeled from 2 10-2 9 K-Zone 2 in Heterogeneous Foundation modeled from 2 10-3 10 K-Zone 3 in Heterogeneous Foundation modeled from 2 10-4 11 K-Zone 4 in Heterogeneous Foundation modeled from 2 10-5 12 K-Zone 5 in Heterogeneous Foundation modeled from 2 10-6 13 K-Zone 6 in Heterogeneous Foundation modeled from 2 10-7 * values in ft/sec
The K-Ratio (Ky/Kx) was assigned as 0.111 for the blanket and core and
0.333 for the shell and homogeneous foundation. The remaining materials were
assigned the default i.e. 1.0 (one).
Typical resultant outputs from SEEP/W analysis include total head
distribution contours and flow path orientations, which can be viewed in the
“CONTOUR” interface correspondingly for sectional models (Figure 4.5). Data
interpretation in this study uses nodal values of the resultant hydraulic gradients and
hydraulic head. The ‘Node Information’ dialog box of the SEEP/W Contour window
(Figure 4.6), allows tabulation of resultant node values for different critical
parameters, individually through a click on a specific node in the finite element mesh.
72
(a)
(b)
Figure 4.5 Typical SEEP/W resultant outputs showing equipotential total head contours and flow paths for (a) No foundation seepage control treatment (b) 15 m deep partial cutoff wall at upstream end of a 3m thick full length upstream impervious blanket.
Figure 4.6 Typical ‘Node Information’ dialog box in SEEP/W Contour window
73
4.4.2 Foundation Seepage Control Scenarios
With the absence of bed rock and presence of unknown extents of openwork
zones within a highly pervious foundation stratum, the possibility of a positive
vertical impervious cut-off was ruled out. The area between the old dam / cofferdam
axis and the new dam axis also imposes space restrictions for horizontal extent of
seepage reduction measures such as the impervious upstream blanket.
A detailed multivariate sensitivity analysis with treatment trials was framed in
regard of all acceptable foundation under-seepage control alternatives as extensions
for the ‘as designed’ considerations for seepage analysis. These treatment trials were
proposed to investigate feasibility of individual seepage reduction elements out of a
partial cutoff or an impervious upstream blanket to be adopted, and their usage in
combination also. The different case scenarios considered are described in Table 4.3,
with the detailed description of each case following thereafter.
Table 4.3 Tabulated list of modeled case scenarios Blanket Only Cases ( B ) Impervious upstream blanket B1 B2 B3 B4 B5 1.5 m thick, 3 m thick, 1.5 m thick, 3 m thick, None half length half length full length full length Cutoff Wall Only Cases ( C ) A partial cutoff wall under core categorized on the basis of penetration depths C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 3 m 6 m 9 m 12 m 15 m 18 m 21 m 24 m 27 m 30 m Combination Cases ( UC and DC ) Partial cutoff wall with the blanket case B5 categorized on the basis of penetration depths UC Cases: A partial cutoff wall at upstream end of the blanket case B5 UC1 UC2 UC3 UC4 UC5 UC6 UC7 UC8 UC9 UC10 3 m 6 m 9 m 12 m 15 m 18 m 21 m 24 m 27 m 30 m DC Cases: A partial cutoff wall at downstream end of the blanket case B5, repositioned under core DC1 DC2 DC3 DC4 DC5 DC6 DC7 DC8 DC9 DC10 3 m 6 m 9 m 12 m 15 m 18 m 21 m 24 m 27 m 30 m
74
a. Initial cases “B” and “C”.
The initial two case sets were aimed at evaluating the requirement of a primary seepage control measure. The selected alternates included a horizontal impervious upstream blanket for the first case set and a partial cutoff wall positioned under the core, for the other case set. The cutoff wall was partial in nature due to the absence of bed rock.
The “Blanket Only” cases, designated by “B” were framed for investigating different blanket thicknesses and the longitudinal extents of the upstream impervious blanket, i.e. the placement length between the old dam / cofferdam axis and the new dam axis. These involved five (5) distinct cases, namely; B1: No blanket case, B2:
Impervious blanket case 1.5m (5ft) thick and placed for half the length, B3:
Impervious blanket case 3m (10ft) thick and placed for half the length, B4:
Impervious blanket case 1.5m (5ft) thick and placed for the full length, B5:
Impervious blanket case 3m (10ft) thick and placed for the full length.
The blanket thickness specified above refers to the minimum thickness throughout its length. The thickness actually varies from the specified minimum thickness at the upstream end of the blanket to a maximum of 15 to 25%
(correspondingly for each minimum blanket thickness specified i.e. 1.5 m and 3 m) of the active maximum head (30 m) at the downstream end, where the blanket joins with the embankment core, forming a knuckle. The main reason for this tapering shape of the blanket is to avoid high blanket gradients developing near the embankment, from the experiences at Tarbela Dam.
75
The “Cutoff Only” cases, designated by “C” were framed for a partial downstream cutoff wall under the main dam core without any upstream impervious blanket. These cases were numbered C1, C2, upto C10, corresponding to the respective cutoff penetration depths at subsequent and alternate increments of 3m
(10ft), i.e. C1 for a 3m (10ft) cutoff penetration depth, C2 for a 6m (20ft) cutoff penetration depth, and so on till C10, for a 30m (100ft) cutoff penetration depth. The maximum depth of penetration considered was relative in proportion to the maximum conservation reservoir level of 30m (100ft) and was correspondingly investigated for a foundation twice as deep.
b. Combination cases “UC” and “DC”.
Cases defining combination of the upstream impervious blanket with a partial cutoff, designated by UC and DC for the Upstream Cutoff wall and the Downstream
Cutoff wall under the main dam core, respectively. These cases were framed under the realization of an Upstream Impervious Blanket as the prime seepage control measure and investigating the alternatives for positioning the cutoff along the longitudinal section. These cases were similarly numbered corresponding to the respective cutoff penetration depths at subsequent and alternate increments of 3m
(10ft), i.e. UC1 or DC1 for a 3 m (10 ft) cutoff penetration depth, UC2 or DC2 for a 6 m (20 ft) cutoff penetration depth, and so on till UC10 or DC10, for a 30 m (100 ft) cutoff penetration depth.
4.4.3 Foundation Representations
Relative comparisons of each treatment trial / ‘as designed’ seepage analysis were proposed over realistic foundation representations, revised in light of the
76
detailed statistical insight of the random variability. These included an ‘as designed’ two-layered approach of a predominant homogeneous foundation condition, with internal pervious zone considerations of variable thicknesses and at variable depths, followed by a ‘multi-zoned’ heterogeneous foundation with different K-zones traced out from the Rockworks foundation database and the subsequently developed p-data section. The corresponding assessment of input permeability values for the seepage analyses software was explained earlier. In accordance with the conditions assumed to represent the prevailing foundation stratum, a two-phased set of analyses was framed for each case scenario described earlier.
In the first phase for the ‘as designed’ conditions, a continuous pervious zone having higher order permeability interspersed within a homogeneous foundation, known and designated by PZ was modeled for detailed analysis. The second phase of analysis trials, for optimization of foundation seepage control measures, involved the
RockWorks modeled multi-zoned heterogeneous approximation of the pervious strata
(with different K-zones), known and designated by RW.
The pervious zone (PZ) conditional assumptions were considered for variable zone thicknesses of 3, 6 and 9m respectively. These assumptions were further diversified to cover variable considerations of the associated depth of the pervious zone from surface alternately spaced at increments of 3 m each for an overall foundation depth of 30 m. Each case thereby involved 10 trials for a 3 m thick pervious zone (PZ1: Pervious Zone Condition 1) considered at 3 m depth intervals from the surface, 9 trials for a 6 m thick pervious zone (PZ2: Pervious Zone Condition
2) considered at 3 m depth intervals from the surface and 8 trials for a 9 m thick pervious zone (PZ3: Pervious Zone Condition 3) subsequently considered at 3 m
77
depth intervals from the surface. Quantitatively, 27 independent seepage analyses were performed corresponding to each case for the PZ foundation approximation and one seepage analysis corresponding to each case for the RW foundation approximation, summing up to a total analysis count of 1064 treatment trials. Typical
SEEP/W Define window shown in Figure 4.7, as developed for the first phase model comprised 87 Regions with 206 Points and 12562 Elements with 11271 Nodes, and in
Figure 4.8, as developed for the second phase model comprised 55 Regions with 312
Points and 29946 Elements with 15133 Nodes.
4.4.4 Sensitivity Criteria
High water losses by seepage would not be objectionable from the standpoint of reservoir operation since minimum releases of 566 m3/s (20,000 cfs) for irrigation, water supply and hydro-power generation are required during the dry season and excess water is spilled in the wet season. It is necessary, however, to control what leakage does occur so that it does not pose adverse effects in terms of initiating internal erosion or piping, through the relatively porous and pervious strata.
As an outcome of the critical appraisal of piping by Richards and Reddy
(2007), it is established that while the condition required for initiating internal erosion or piping differs considerably from site to site, assessment for its development has to consider functioning gradients with suggested engineering guidance of limiting the threshold values of safe hydraulic gradients to less than 1.0. Sensitivity criteria thereby involved a credible magnitude of hydraulic gradients carried along in the foundations for each analysis trial.
78
Figure 4.7 Typical “Define” window for PZ foundation approximation showing the layered hydraulic conductivity K assignment
Figure 4.8 Typical “Define” window for RW foundation approximation showing the multi-zoned hydraulic conductivity K assignment
79
In order to compare predictions to the piping criteria, the selected parameter
‘functioning gradient distributions’ for this study follows the traditional criteria, derived on gradient only, and not particle size or tractive shear stress. The comparative relevance of total head variation along the foundation profile is also significant (Figure 4.5) and therefore additionally included for relative comparison.
Different locations along the foundation profile were investigated for sensitivity of comparative distributions under the proposed scenarios with respect to the piping potential under adequate head and gradient efficiency. The selected points of consideration are shown in Figure 4.9 and are described as; Point 1 at the foundation interface, after initial 15 m; Point 2 approximately at the middle of the upstream impervious blanket; Point 3 at the toe of the core and Point 4 at the drain ditch. Nodal values were determined from the ‘Node Information’ dialog box of the
SEEP/W Contour window (Figure 4.6), for the critical parameters selected and tabulated individually for the four points of consideration, against each analysis trial for comparison and optimization of the adopted seepage control measures.
4.5 INSTRUMENTED OBSERVATIONS
4.5.1 Project Instrumentation
The instrumentation system established for Satpara Dam was aimed at exclusive monitoring of the performance of adopted seepage control measures in terms of integrity, effectiveness, pore water pressures and seepage patterns. The data and information used for the subsequent performance evaluation in accordance with the objectives of this research was obtained from the WAPDA staff deputed for the purpose.
80
Figure 4.9 Pertinent case scenarios and sensitivity criteria for the multivariate sensitivity analysis
81
The instrumented network of piezometers has been exclusively considered in this research. The distribution generally covers instrumentation requirements through installations across Cutoff Wall / along Cofferdam, across the wrap around, along the outlet structure (Profile of the dam), across the main embankment dam (MED), along the spillway and the left abutment as shown in Figure 4.10.
The number and distribution of these instruments was preplanned so as to improve the general knowledge by correlating measurements and tracing hypothetical seepage paths. A total of 63 vibrating wire piezometers and 20 stand pipe piezometers were installed at different locations throughout the project. Specific details of the project instrumentation are given in Table 4.4 and Table 4.5.
With respect to the nature of required monitoring aspects, giving due regards to the importance of each aspect, 32 piezometers were installed for foundation measurements, 21 specifically for the performance of the blanket, 5 piezometers were placed in the core, 3 piezometers were installed in the shell / shoulders (2 within the upstream and 1 for the downstream shoulder) and 2 were installed with their tips placed in the filter material of the horizontal drainage blanket. Out of the 53 piezometers for foundation and blanket, 16 were located at interfaces, i.e. 5 at foundation interfaces (2 at foundation-blanket interface, 1 at foundation-core interface and 2 at foundation-filter interface, respectively), whereby the remaining 11 were located at blanket/core interfaces. The latter interface is defined in terms of compaction criteria since the area traverses footprints of the core but is considered in the upstream impervious blanket extents.
82
Figure 4.10 Instrumentation layout plan
83
Table 4.4 Vibrating Wire Piezometer Details
Tip Factory Co-ordinates (m) Installation Initial P. Id. Elevation Constant Position Date Reading N E (m) (m) VWP1 1243767.78 3438388.70 2646.34 05-Aug-05 6309.10 391.8391 Foundation VWP2 1243799.79 3438413.56 2649.39 30-Sep-05 7036.80 457.6856 Blanket VWP3 1243799.79 3438413.56 2640.24 18-Sep-05 6553.60 445.5796 Foundation VWP4 1243816.71 3438429.23 2649.39 10-Aug-05 4975.50 Foundation VWP5 1243816.71 3438429.23 2640.24 08-Aug-05 5719.70 379.8158 Foundation VWP6 1243855.86 3438461.79 2649.39 29-Jul-05 5452.69 399.5004 Foundation VWP7 1243855.86 3438461.79 2640.24 27-Jul-05 5034.64 384.4954 Foundation VWP8 1243759.66 3438468.20 2652.44 08-Jul-10 5782.45 461.3947 Foundation VWP9 1243811.50 3438513.13 2661.59 10-Jul-10 5830.10 451.1 Shell VWP10 1243817.24 3438518.14 2655.49 31-Aug-10 6716.86 Blanket / Core VWP11 1243829.48 3438528.70 2655.49 18-Sep-10 6532.42 403.1777 Fdn / Filter VWP12 1243705.59 3438495.85 2634.15 20-Jun-08 7225.10 434.8275 Blanket VWP13 1243751.75 3438535.68 2631.10 28-Nov-07 5992.70 442.0273 Blanket VWP14 1243763.30 3438545.65 2646.34 09-Apr-09 6679.50 424.2181 Shell VWP15 1243769.07 3438550.62 2637.20 04-Dec-07 6719.70 459.619 Blanket VWP16 1243774.85 3438555.61 2646.34 08-Apr-09 5987.35 448.3331 Shell VWP17 1243774.85 3438555.61 2628.05 17-Jun-08 6016.20 442.8095 Fdn / Core VWP18 1243780.62 3438560.59 2640.24 03-Aug-08 6555.31 418.2111 Core VWP19 1243801.24 3438578.45 2640.24 14-Apr-09 6735.69 419.246 Shell VWP20 1243800.05 3438577.28 2628.05 28-Jun-08 5628.00 466.2533 Fdn / Filter VWP21 1243837.01 3438609.24 2628.05 27-Jun-08 5779.60 436.8276 Foundation VWP22 1243528.54 3438465.98 2637.20 22-May-07 6704.46 457.2807 Blanket VWP23 1243528.54 3438465.98 2625.00 02-Dec-05 5607.40 425.2367 Foundation VWP24 1243541.97 3438473.20 2634.15 12-Apr-07 5712.50 420.6478 Fdn / Blanket VWP25 1243541.97 3438473.20 2628.05 05-Dec-05 7402.20 468.9596 Foundation VWP26 1243749.85 3438585.10 2649.39 11-Jul-08 7197.30 448.2016 Core VWP27 1243756.54 3438588.83 2640.24 03-Aug-08 6579.07 436.226 Core VWP28 1243764.67 3438593.08 2634.15 22-Jun-08 6045.12 438.8035 Filter VWP29 1243778.09 3438600.30 2634.15 24-Jun-08 5063.80 395.2082 Filter VWP30 1243732.30 3438570.73 2640.24 26-Apr-08 6818.13 421.3073 Core VWP31 1243732.30 3438570.73 2634.15 29-Sep-07 6433.80 423.3381 Blanket / Core VWP32 1243723.83 3438587.18 2634.15 28-Sep-07 6647.60 419.5187 Blanket / Core VWP33 1243723.83 3438587.18 2625.00 17-Dec-05 6033.40 435.9779 Foundation VWP34 1243677.28 3438557.78 2637.20 13-Jul-07 6457.60 426.7298 Blanket / Core VWP35 1243677.28 3438557.78 2628.05 14-Dec-05 5722.60 437.7365 Foundation VWP36 1243673.92 3438564.62 2646.34 23-Oct-08 6546.82 394.8534 Core VWP37 1243673.92 3438564.62 2634.15 20-Nov-07 5938.90 426.7819 Foundation VWP38 1243659.01 3438487.12 2634.15 18-Aug-07 6578.00 409.6649 Blanket VWP39 1243638.95 3438542.07 2637.20 13-Jul-07 6610.00 429.9524 Blanket / Core VWP40 1243638.95 3438528.21 2625.00 10-Dec-05 5844.70 429.2576 Foundation VWP41 1243632.13 3438542.07 2646.34 23-Oct-08 6334.26 431.0736 Core VWP42 1243632.13 3438542.07 2637.20 13-Jul-07 7200.00 444.39 Blanket / Core
84
Table 4.4 (continued)
Tip Factory Co-ordinates (m) Installation Initial P. Id. Elevation Constant Position Date Reading N E (m) (m) VWP43 1243605.34 3438487.37 2634.15 31-Mar-07 6470.90 407.1854 Blanket VWP44 1243593.35 3438512.35 2628.05 28-Nov-05 5924.30 440.0743 Foundation VWP45 1243591.39 3438516.29 2634.15 01-Apr-07 6749.60 429.9691 Blanket / Core VWP46 1243576.26 3438543.35 2649.39 18-May-09 6420.69 428.1497 Core VWP47 1243578.18 3438543.77 2643.29 01-May-09 5351.90 387.4774 Foundation VWP48 1243660.38 3438413.07 2640.24 06-Jul-08 6833.50 411.8635 Blanket VWP49 1243662.22 3438413.30 2631.10 20-May-08 6045.60 408.5166 Foundation VWP50 1243574.59 3438386.74 2637.20 13-May-05 5665.20 385.7009 Foundation VWP51 1243574.59 3438386.74 2628.05 11-May-05 6563.30 411.7764 Fdn / Blanket VWP52 1243559.98 3438382.32 2637.20 13-May-08 6106.80 385.3662 Blanket VWP53 1243559.98 3438382.32 2628.05 11-May-08 6209.00 397.8021 Foundation VWP54 1243679.01 3438320.94 2640.24 24-Nov-05 5938.40 392.0347 Foundation VWP55 1243679.01 3438320.94 2631.10 22-Nov-05 6101.80 405.2308 Foundation VWP56 1243607.46 3438294.65 2640.24 27-Apr-08 6719.70 392.7114 Blanket VWP57 1243607.46 3438294.65 2631.10 17-May-05 5986.80 400.6199 Foundation VWP58 1243593.10 3438289.54 2637.20 16-May-08 6104.30 386.7736 Foundation VWP59 1243593.10 3438289.54 2628.05 14-May-08 5870.40 394.1658 Foundation VWP60 VWP61 1243714.56 3438202.74 2640.24 13-Aug-10 5786.00 436.0269 Foundation VWP62 VWP63 1243757.48 3438292.46 2649.39 22-Aug-10 5400.05 393.922 Foundation
85
Table 4.5 Standpipe Piezometer Details
Tip Top Co-ordinates (m) Installation Sounding P. Id. Elevation Level Type Date Depth (m) N E (m) (m) SP1 1243847.63 3438635.15 2625.00 20-Jul-09 2633.88 8.88 Slotted Pipe SP2 1243795.07 3438641.51 2626.52 31-Jul-09 2651.16 24.64 Slotted Pipe SP3 1243774.67 3438633.13 2631.10 30-Aug-09 2659.00 27.91 Slotted Pipe SP4 1243818.87 3438465.49 2628.05 27-Aug-08 2668.02 39.97 Casagrande Tip SP5 1243856.33 3438517.91 2626.52 14-Oct-08 2669.53 43.01 Casagrande Tip SP6 1243884.97 3438546.60 2626.52 18-Sep-08 2661.34 34.82 Casagrande Tip SP7 1243902.91 3438569.28 2625.00 5-Nov-08 2649.44 24.44 Casagrande Tip DTSP8-1 1243912.34 3438623.51 2614.33 13-Nov-08 2636.76 22.43 Casagrande Tip DTSP8-2 1243912.34 3438623.51 2621.95 18-Nov-08 2636.76 14.80 Casagrande Tip DTSP9-1 1243891.56 3438648.77 2614.33 3-Sep-08 2630.66 16.33 Casagrande Tip DTSP9-2 1243891.56 3438648.77 2621.95 5-Sep-08 2630.66 8.70 Casagrande Tip DTSP10-1 1243935.92 3438651.38 2614.33 11-Jul-09 2630.69 16.36 Casagrande Tip DTSP10-2 1243935.92 3438651.38 2621.95 21-May-09 2630.69 8.73 Casagrande Tip SP11 1243887.92 3438605.61 2625.00 28-Sep-07 2633.86 8.86 Casagrande Tip SP12 1243875.10 3438637.53 2625.00 15-Sep-07 2630.73 5.73 Casagrande Tip SP13 1243867.36 3438655.95 2625.00 24-Sep-07 2631.58 6.58 Casagrande Tip SP14 1243838.52 3438420.88 2628.05 6-Oct-09 2667.91 39.87 Slotted Pipe SP15 1243902.94 3438483.49 2626.52 15-Nov-09 2656.79 30.27 Slotted Pipe SP16 1243919.59 3438502.70 2625.00 3-Jun-10 2653.36 28.36 Slotted Pipe SP17 1243935.46 3438525.88 2621.95 30-Jun-10 2652.41 30.46 Slotted Pipe
86
Initially only 13 standpipes were distributed downstream of the dam, at different positions of critical importance. Three standpipes were proposed with dual tips, hence a total of 16 standpipe tips were proposed. Four standpipe piezometers were additionally included at a much later stage to account for valley flows from the left abutment.
A line of drainage relief wells spaced about 3m (10ft) on centers, completes the instrumentation system, providing a ready source for monitoring any seepage discharges downstream of the dam. The relief wells discharge into the toe drain, connecting to the nullah for final disposal of collected seepage. A total of 30 relief wells were proposed for the project. Initially only 12 relief wells were installed at the downstream toe of the MED, at a later stage the remaining 18 relief wells were proposed along the toe drain circumscribing the natural weighted berm formed by the access road left of spillway.
In addition to the above proposed network of installations, the instrumentation system also included seepage monitoring stations (SMSs). A blow-up of the main dam downstream vicinity highlighting these SMSs is shown in Figure 4.11 and their tabulated description follows in Table 4.6. These SMSs were installed at different periods through progress of the impounding stages under realization of seepage emergence towards a downstream vicinity of the dam. A total of four SMSs were installed, all operating in accordance with open channel flow conditions. A pipe- culvert structure was placed with the start of the third impounding stage immediately downstream of the stilling basin stone apron in the Satpara Nullah and just upstream of the channels from the toe drain and regulating outlets.
87
Figure 4.11 Seepage Monitoring Stations (SMS)
Table 4.6 Seepage Measurement Station Details
SMS Co-ordinates (m) Structure Location Measurement Installation Date Id. N E Parshall Turning of toe- Main Dam SMS1 1243578.716 3437758.232 20-Jul-09 Flume drain underseepage
V-Notch Left of road Spillway by-pass SMS2 1243622.452 3437804.873 31-Jul-09 Weir culvert seepage Spillway / Left V-Notch Right of road SMS3 Abutment 1243648.668 3437817.257 30-Aug-09 Weir culvert Underseepage Pipe - Downstream of SMS4 Total seepage 1243688.231 3437832.35 27-Aug-08 Culvert spillway
88
With the completion of the toe drain towards the start of the fourth impounding stage, three more SMSs were installed along its alignment. These included a Parshall flume and two v-notch weirs installed to measure collected seepage / leakage at different locations in the downstream toe drain provided for collecting and conveyance of the drainage provisions.
4.5.2 Impounding Status and Data Acquisition
The project has undergone four complete phases of partial impounding, while the fifth phase is currently being monitored. The maximum reservoir level recorded to date was 2656.88 m (8716.8ft) on 25-08-2010 and the minimum drawdown level was
2629.964m (8628.49ft) on 15-05-2009. The designed maximum reservoir level being
2663.952m (8740ft), the foundations and project features have endured approximately
80 % of the design head to date. Despite the ensued impounding stages, the designed maximum conservation level was not yet achieved. This was mainly due to regulation of inflows in accordance with the downstream requirements and construction progress. This aspect has limited the present study to include the available observation data to-date for the purpose of research.
Regular record of all the installed instruments was available with the site staff, since pre-impounding in July 2006. Reservoir levels and Piezometeric levels were manually recorded, based on a specified predefined frequency for each installed piezometer. An electric probe was used to measure the water levels in the reservoir and the standpipes, and a hand held readout unit was employed for reading the vibrating wire instruments. Initially the frequency of readings was weekly, while during impounding it was changed to a regular frequency of daily recordings. The
89
complete period of impounding comprised a total of 89,372 physical piezometeric observations through 1563 days. These days were distributed as 46, 221, 362, 418,
365 and 151 days covering the pre-impounding and subsequent five partial impounding stages, respectively.
Visual observations of seepage emergence were recorded starting from 26 Jun
2009 in the third impounding stage, when the pipe culvert at a location downstream of the spillway was operable. The pipe-culvert was considered to cumulate the overall seepage flows and hence utilized as the collective seepage measuring station. The monitoring frequency at the pipe-culvert varied initially as random weekly or bi- weekly measurements, it was revised to consistent daily records from June 2010 to date, with an interim interruption to weekly readings during the low flows corresponding to the receding lake levels of the fourth impounding stage. With the installation of the parshall flume and the two v-notch weirs, additional seepage measurements at their locations were made with the same frequency of observations.
The first additional seepage measurement was taken on August 12, 2010, nearly in the mid of the fourth impounding stage.
4.5.3 Data Processing and Graphical Plots
Instruments were installed by the civil works contractor in the presence of representatives of the project consultants (SDC). Installation sheets & borehole logs for each instrument were prepared for record and further process. The installation sheets recorded the installed tip elevation of each instrument and its respective zero reading, which was recorded just prior to installation and was used for further data processing. The zero reading was the reference or NSL elevation for the standpipe
90
piezometers and a stable linear reading of each vibrating wire instrument established in water of negligible hydrostatic head. Standpipe piezometeric elevations were computed by subtracting the measured head of water in the piezometer from the recorded zero / initial reading. Data obtained from vibrating wire piezometers was in digital or “raw” form. Each vibrating wire instrument was supplied with a Generic
Calibration Constant value, to convert raw data into required engineering units (head of water in linear units). The measured raw data was then processed and transformed into the required engineering “linear units” by applying the following formula.
-4 H = C (Ri - R) x 10 (4.1)
Where H = Head of water in linear units expressing ground water level
C = Generic Calibration Constant value for the required engineering
units from the supplied Factory calibration sheet.
Ri = Initial Linear Reading of the Instrument.
R = Current Linear Reading of the Instrument.
Numerically tabulated data was not readily conducive in evaluating behavior or detecting trends. The processed data obtained from the individual piezometeric digital or “raw” data was therefore converted into graphical plots, using linked worksheets for this research.
The graphical plots of the processed data provided visual means to detect data acquisition errors, to determine trends or cyclic effects, to compare and predict the foundation behavior and to determine the effectiveness of adopted SCMs. The data obtained from the individual piezometers installed was plotted in several different
91
ways to evaluate coherence and functionality of the instruments. The most straightforward method was to plot both the piezometer level and the reservoir level against the same calendar time scale. The general rise and fall of piezometric levels corresponding to each individual piezometer with the reservoir level was drawn as a time history plot (Figure 4.12). The effect of reservoir level on piezometric levels was even better illustrated by graphs of piezometric level vs. reservoir level (Figure 4.13).
The average percent potential plots (Figure 4.14) were even more conducive in establishing the effect of foundation behavior corresponding to the different impounding stages. The potential has been defined with the reservoir level as the
100% potential and the zero potential as the piezometric level observed in the standpipe SP12. The potential at a particular piezometer is computed by the formula:
Percent Potential = (PL – PL0) / (PL100 – PL0) x 100 (4.2)
Where PL = Piezometeric level at the piezometer
PL0 = Piezometeric level corresponding to 0% potential (at the
standpipe piezometer: SP12)
PL100 = Reservoir level corresponding to 100% potential.
The average observed percent potential contour plot plans as shown in Figure
4.14 were produced using the inbuilt ‘Surface Modeling’ routine from the Eaglepoint software. The contours were based on the percent potentials computed from actual water pressure measurements made in each piezometer at its specific installed depth zone within the study area, portrayed corresponding to its proper tip elevation. The deepest level piezometers were selected for a representative foundation plan.
92
Figure 4.12 Typical Time Plot
Figure 4.13 Typical Head Plot
93
Figure 4.14 Percent Potential Contour Plot Plan
The ‘Surface Modeling’ routine of Eaglepoint software uses an advanced interpolation algorithm to develop the respective plans. Inputs were provided for the model domain of the present study in the form of x and y coordinates for piezometer locations and a z coordinate for the calculated percent potential. The plan was then developed by connecting the given piezometer points and forming triangles to create a surface model. Triangulation (creation of triangles) can be controlled by defining
“breaklines” or external and internal boundaries for areas where significant grade breaks / water divides occur, fortunately for the research model domain, no such boundary existed. The contours were generated by interpolation along the triangle lines to find the given place where the specified contour interval would lie.
94
4.6 POST-CONSTRUCTION SEEPAGE MODELLING
4.6.1 Comparative 3-D Seepage Modelling
Critical assessment follow-ups also involved steady state comparative analyses of the 2D simulation results along different sectional models with 3D simulations.
Steady state 3-dimensional seepage analyses in context of this study were performed using the finite element computer program FEFLOW (Finite Element subsurface
FLOW), an interactive and comprehensive groundwater modeling system.
a. Model Domain, Spatial Discretization and 3-D Configuration.
The FEFLOW modeling environment inherently utilizes external map files to facilitate definition of model boundaries and assist in assigning specific input requirements. Two types of external map files were used in the mesh generation process of the 3-D model for this study, namely CAD oriented files for layout with
DXF extensions and parameter table files with DAT extensions.
The framework for a finite-element mesh generation in FEFLOW was then formed through a Supermesh, providing all the basic geometrical information needed by the mesh generation algorithm. The model area outline was defined using a single polygon and the location of the upstream cutoff wall using a line, as active Supermesh components for the model (Figure 4.15). A flexible triangulation algorithm called the
“Gridbuilder” was used for obtaining a suitable spatial discretization (generation of a finite element mesh) of the model domain in this study (Figure 4.16).
95
Supermesh Polygon: Project Boundary
Supermesh Line: Cutoff Wall
Figure 4.15 Model domain described as ‘Supermesh’ components in FEFLOW.
Figure 4.16 Finite element mesh in FEFLOW.
96
A layer-based approach (into a 5 layer, 6 slice geometry) facilitated by the 3D
Layer Configuration dialog (Figure 4.17) was then applied to extrude the triangular or quadrangular 2D mesh in the third dimension, resulting in prismatic 3D elements distributed between exactly horizontal slices defined by the top and bottom of each layer.
Figure 4.17 3D Layer Configuration dialog in FEFLOW.
b. Model Inputs and Assignment Procedures.
Specific model inputs and their assignment procedures were involved as the next step of the 3-D model setup for the study. The basic input requirements for the 3-
D model setup for the study included permeability or hydraulic conductivity, actual elevations and boundary conditions. Typical assignment procedures were based on the
97
individual activation of one of these parameters to be assigned in the Data panel and selection of the target elements or nodes. The material properties i.e. the K-values were assigned on an elemental basis, while the process variables (actual elevations) and boundary conditions were defined on the mesh nodes.
FEFLOW allows import of attributed data files with global grid co-ordinates in the .DAT format extension table files (Figure 4.18). As such, the attribute files were developed for the K-values and the actual elevations. The K-data attribute files included co-ordinates, the layer number and the respective K-value for each grid cell on the layer, whereas a similar attribute file for the elevations included co-ordinates, the slice numbers with the actual elevations for each slice.
Figure 4.18 Typical “.DAT” format extension table file for importing K-values in FEFLOW.
98
The target parameter was activated in the Data panel by a double click or via the context menu. The activation was indicated in bold letters. The elevations and K- values were imported as external attributed data files with global grid co-ordinates for each parameter in the .DAT format extension table files. These attributes were used as the data source for each parameter assignment as a time constant automatic data input.
Each desired map attribute was linked to a target FEFLOW parameter in the
Parameter Association dialog accessed via the context menu of the respective map in the Maps panel. Detailed settings for the data transfer were defined in the properties of the link in the Parameter Association dialog (Figure 4.19). The source data unit was specified as meters (m) for elevation and m/s for the K-values. The regionalization method “Akima” was selected for a classic two dimensional data interpolation.
Furthermore, the elevations were related to slices and the K-values to layers by selecting the slice or layer column descriptors contained in the attribute data of the table map. For the K-data inputs, different cross-sections were extracted from the project domain along specified grid rows from the actual site data modeled as a 3- dimensional solid model in Rockworks (as explained earlier under section 4.3.2).
These cross-sections were used to manually develop layered maps on grid in the
AutoCAD interface, tracing K-values / zones at regular depth intervals (15 m) from each consecutive section. These foundation depth plans were subsequently used to extract K-values corresponding to the global grid co-ordinates, as an external excel file for the 3-D K-value inputs. Typical cross-sectional profiles are shown in Figure
4.20 and the developed plan on project grid for layer 2 is shown in Figure 4.21. The dam surface was idealized using the top slice, the blanket thickness above the NSL was specified with Slice 2, thus the fill over surface was presented as layer 1. Slice 3 defined bottom of the first foundation layer, i.e. layer 2.
99
Figure 4.19 Typical Parameter Association dialog linking K-values in FEFLOW.
The cutoff permeability was reassigned specifically in layer 2, due to its adjunctive location below the blanket in the first foundation layer bordering the surface. An interactive data input was used with the map geometry (cutoff wall - line) serving as the target geometry for data input. All elements within the line-based target geometry selected were manually assigned the specified value i.e. 5.5 x 10-6 m/s (10-6 ft/s). Figure 4.22 shows the K-value plan from FEFLOW for layer 2, wherein the cutoff wall is present. The figure also shows a blow-up of the cutoff wall area, indicating the manual assignment of K-values for the cutoff wall for the selected elements (along the super mesh line) in the specified layer (layer 2). The 3-D elevation model obtained in a similar fashion through individual data assignment on six slice plan plots of all the five layers for the mesh nodes is shown in Figure 4.23.
100
Figure 4.20 Typical layered cross-sectional profiles extracted through RockWorks
Figure 4.21 Typical manually developed layer map on project grid for Layer 2.
101
Figure 4.22 K-data assignment for the cutoff wall (blowup) within the FEFLOW K-map for Layer 2.
102
Supermesh Polygon: Project Boundary
Supermesh Line: Cutoff Wall
Supermesh Elements: 1 Polygon, 1 Line
Mesh Geometry: 5 Layers, 6 Slices 171,600 Elements and 113,640 Nodes
Figure 4.23 FEFLOW Snapshot View for 3D Mesh and Digital Elevations Model
The assignment of boundary conditions was made through constant values for the parameter to the target nodes. It was executed by a map-based node selection through import and activation of an upstream / downstream boundary domain using an external DXF map, and by entering the value (upstream constant head or selecting the seepage face as a toggle option) in the input box in the Editor Toolbar. Figure 4.24 shows the typical assigned boundary conditions for the 3D model in FEFLOW.
c. Basic Settings and Simulation.
The basic settings defining the simulated process as a saturated, steady state, constant time step, 3D fixed mesh phreatic aquifer were done via the Problem Settings dialog (Figure 4.25). By default, a newly created FEFLOW model reflected a confined aquifer. Saturated simulations of unconfined conditions were induced by specifying a “phreatic” mode to the top slice.
103
Downstream Boundary Condition: Seepage Face
Upstream Boundary Condition: 2656 m (8715 ft)
Figure 4.24 Typical assigned boundary conditions in FEFLOW.
Figure 4.25 Problem Settings dialog in FEFLOW for “phreatic” mode assignment to the top slice, i.e. Slice 1.
104
The Play button in the Simulator toolbar starts the simulation. The resulting hydraulic-head distribution was contained as a process variable after simulation, exportable as individual slice plans. A typical slice plan for the resultant hydraulic- head distribution is shown in Figure 4.26. DXF plan plots of the visualization features such as isolines or fringes (by using their context menu in the View Components panel) were used to extract selected sectional values of the hydraulic head distribution for comparative interpretation of the results with a 2D simulation.
Figure 4.26 Typical resultant head distribution for the top slice (FEFLOW Snapshot View from 3D model).
4.6.2 Post-Construction Loading Conditions
The available impounding data from five years of consecutive operation
(2007-2011) did not achieve the designed conservation level. It also could not depict a
105
steady state reservoir level from the different stages of reservoir impounding. As such, pseudo-steady state reservoir levels were identified from the last three impounding stages. The selected reservoir levels for the percent potential contour plots included two different pond elevations with consecutive recurrence levels of +/- 0.15 m (0.5 ft) persistent for approximately more than two weeks to ensure a pseudo-steady state.
These reservoir levels were 2647 m (8685 ft) a.m.s.l. corresponding to the maximum level of the third partial impounding stage and 2656 m (8715 ft) a.m.s.l. corresponding to the maximum levels achieved to-date in the fourth partial impounding stage. A third pseudo-steady reservoir level of around 2648 m (8687 ft) a.m.s.l. from the fifth partial impounding stage was neglected in view of the minor difference of only around 1 m for the stage three and stage five pseudo-steady reservoir levels.
These pseudo-steady state reservoir levels defined new loading conditions, which were used to extend the 2-D ‘as designed’ SEEP/W model and additionally simulate the 3-D seepage model, under realization of a 3-D flow domain in the constricted valley topography, using FEFLOW. The typical details of model development have been described earlier. Both post-construction seepage analyses models were based on the pre-construction multi-zoned heterogeneous foundation representation from the Rockworks foundation database.
The aim of post-construction modelling was to simulate and match the observed ‘pseudo-steady’ foundation response. Comparative observed values for the simulation analyses were established through interpolated values from average observed piezometeric elevation contour plans. These plans were developed corresponding to the selected pseudo-steady reservoir levels and selected lower level
106
foundation piezometers at projected locations of each piezometer. Total head results from the sectional SEEP/W seepage model were directly obtained through the nodal information. Comparative values from the 3-D model had to be extracted manually from the resultant head contour plans. For this, a relevant comparative section line was superimposed and the corresponding point values were readout for the desired comparative reservoir level within close vicinity of the section / piezometer orientations.
4.7 UNCERTAINITY PROPAGATION ANALYSES
4.7.1 General Analyses Procedure
An essential step subsequent to the multivariate sensitivity analyses and the instrumented observations, required evaluation of the modelled foundation profiles with respect to the selected seepage control measures in a post construction scenario.
This contribution of the research required analysing a profile using the best estimates of a representative hydraulic conductivity with a comparative analysis of the results from post-impounding measured piezometric data. The initial foundation permeability profile approximations included the ‘RockWorks’ heterogeneous foundation model from the pre-construction investigated down-hole point sampled data. The pre- construction ‘RockWorks’ foundation model comprised of a heterogeneous foundation comprising six (6) different K-zones ranging from a maximum K-zone permeability of 1 cm/sec to a minimum of 10-5 cm/sec.
The post construction scenario also included a rationale for probable foundation redistributions corresponding to the reservoir level fluctuations in a 3-D flow domain. As such, likely fines migration from the silty matrices in the high K-
107
zones (comprising cobbles and boulders) to low K-zones (comprising sandy gravels) could induce probable K-zone changes. Quantification of the extent of these uncertain
K-zone changes and the associated K-values to use formed the basic idea for an investigative approach in the ‘uncertainty propagation analyses’.
The uncertainty propagation analysis was essentially a 2-D seepage back- analysis (i.e. an inverse 2-D modelling approach) using SEEP/W, comprising of trials for reducing the differences between simulation results and measured data (total heads) by adjusting the model input parameters (permeability or K-zones). The 2-D seepage modelling procedure by SEEP/W has been described in an earlier section.
This approach specifically accentuated redefining of the hydraulic conductivity estimates for the heterogeneous foundation profile. The six (6) different K-zones from the original pre-construction RockWorks heterogeneous foundation profile were selectively refined / interchanged, within the vicinity of an identified modified seepage regime for post construction K-zone adjustments. Each analysis was rerun with post construction adjusted K-zones. This process was repeated until reasonable agreement was reached between the computed results and the measured values, consequently qualifying the design stage / pre-construction K-zones associated with changes and quantifying the extents.
Two typical instrumented cross-sections of the project were selected for this procedure. Initial trials were done through comparison of the observed and simulated results at a section ‘X-X’, whereas a section ‘K-K’ was used for substantiating the initial trial assumptions. Figure 4.27 shows the orientation of the selected sections in the layout plan with percent potential contour plots of the selected pseudo-steady reservoir levels. Total head results of the finite element models with multi-zoned
108
foundation representations from the pre-construction and the re-adjusted back- analyses for K-value approximations regarding the dam foundations were compared with observed piezometeric elevations for selected lower level foundation piezometers, installed along the selected instrumented cross-sections.
Figure 4.27 Layout plan for orientation of section X-X and section K-K
4.7.2 Statistical Evaluation Parameters
A coherent statistical evaluation of the modelled results was supplemented utilizing three different statistical measures, namely the RSR, i.e. Root mean square error (RMSE) – observations Standard deviation Ratio, the NSE, i.e. Nash-Sutcliffe
109
Efficiency and R2 i.e. the coefficient of determination. Each model evaluation statistic is briefly described hereunder, with the corresponding formulae presented using X as the observed value of the parameter being evaluated, X’ as the mean observed value,
Y as the simulated value, n as the total number of observations and i representing the numbers 1, 2, 3, …….n.
Singh et al. (2004) recommend RSR, it standardizes model RMSE using the observations standard deviation, and combines the benefits of error index statistics and the additional information of a scaling/normalization factor. RSR value less than
0.5 is an acceptable indicator of low RMSE or residual variation and therefore a good model simulation.
n 0.5 2 (X Y ) RMSE i1 RSR 0.5 (4.3) STDEV n obs 2 (X X ) i1
NSE is a normalized statistic that determines relative magnitude of the residual variance “noise” compared to the observed data variance “information”
(Nash and Sutcliffe, 1970). Moriasi et al. (2007) advise the use of NSE based on recommendations of ASCE (1993), Legates and McCabe (1999) and its widespread application. Values less than zero (0) indicate unacceptable model performance, whereas values between zero (0) and one (1) are generally viewed as acceptable levels of performance.
n 2 (X Y ) i1 NSE 1 n (4.4) (X X )2 i1
110
R2 describes the proportion of variance in model results from the observed data over a value range of zero (0) to one (1), with higher values indicating less error variance. In general, values greater than 0.5 are acceptable indices for the degree of collinearity between simulated and observed data. (Santhi et al., 2001, Van Liew et al., 2003).
2 n XY X Y R 2 (4.5) 2 2 2 2 n X X n Y Y
111
CHAPTER 5.
RESULTS AND DISCUSSIONS
5.1 INFERENCES ON SUBSURFACE CONDITIONS
5.1.1 Summary Findings from Exploratory Efforts
The segment of the valley that has been studied in connection with the underseepage problem extends from just upstream of the cofferdam shown on Figure
3.4, to just downstream of the spillway chute, a distance of about 609.6 m, covering the entire expanse of the reservoir spanning from the left to the right abutments.
Various types of subsurface investigation works were carried out to determine site specific features like depth, nature, condition and composition of the fundamental subsurface constituents at the dam site. The program of investigations was conducted through phases of evolution of the project as described above. The phase wise summaries for the logs are presented in Table 5.1.
The overall evaluation of subsurface composition has indicated the dominance of fines (approximately 84%) with occasional Boulders (about 10%) and minor lenses of Open works (about 6 %) in the first phase of investigations. Whereas the subsequent phases of investigation have reversed the composition to dominance of boulders with assured presence of openworks (about 1/4th) and varied percentages of fines (10-26%). Bedrock was not encountered at any of the boreholes except two bore holes SDR-1 and SDR-2 drilled along the dam axis on the right abutment during the second phase of investigations. Granodiorite was encountered at 49.4 m and 19.7 m depths respectively for these boreholes, confirming the continuity of the exposed outcrop at higher levels, from the surface geological mapping.
112
Table 5.1 Phase wise summary - Logs
In-Situ K- Log Counts Distribution (%age) Test Counts
Easting Northing Borehole / Test Pit / / Pit Test / Borehole Trench Test (m) Elevation (m) Depth Total AvgGWL (m) (m) (m) Rock Boulders Openworks Fines Rock Boulders Openworks Fines K-Test Attempts Successful Tests Phase I (Feasibility Stage - 1988) BH-1 3437556 1243213 2637 18 1 3 25 75 1 1 BH-2 3437528 1243228 2633 18 4 100 BH-3 3437554 1243242 2633 18 2 100 1 1 BH-4 3437526 1243195 2633 9 1 4 20 80 BH-5 3437605 1243159 2637 6 1 2 33 67 BH-6 3437454 1243191 2635 6 2 3 40 60 1 1 P-1 3437663 1243154 2656 6 2 100 1 1 P-2 3437591 1243204 2633 9 3 100 1 1 P-3 3437634 1243260 2633 0 P-4 3437601 1243244 2633 4 2 100 1 1 P-5 3437406 1243272 2638 6 2 2 50 50 1 1 P-6 3437352 1243286 2638 6 1 3 25 75 1 1 P-7 3437430 1243291 2637 6 5 100 P-8 3437388 1243241 2638 3 1 100 P-9 3437433 1243246 2638 2 1 100 2 P-10 3437431 1243262 2637 6 1 100 P-11 3437477 1243243 2637 4 1 100 P-12 3437640 1243192 2656 3 1 100 P-13 3437628 1243226 2637 3 1 100 Phase II (Design/Tender Stage - 2003) SD-1 3437536 1243217 2633 19 2 3 1 33 50 17 1.04 3 3 SDL-3 3437635 1243521 2666 78 11 6 65 35 27.99 6 4 SDL-5 3437578 1243529 2673 30 14 1 2 82 6 12 26.49 6 4 SDR-1 3437703 1243462 2630 61.5 2 18 2 9 82 9 2.06 13 13 SDR-2 3437721 1243446 2637 35 9 5 1 60 33 7 10.00 9 8 SDR-6 3437825 1243602 2630 40.7 13 5 8 50 19 31 5.20 10 8 SDSB-1 3437793 1243653 2631 25 12 100 3.67 6 6 TP-1 3437722 1243505 2658 4 1 100 4 TP-2 3437710 1243430 2660 6 1 100 TP-2A 3437645 1243499 2666 3 1 100 TP-3 3437412 1243368 2646 3 1 100 TP-4 3437331 1243328 2671 3 1 100 TP-5 3437484 1243304 2643 3 1 100 TRB-1 1.4 2 100 TRB-2 1.4 2 100 TLB-1 1.4 2 100 TLB-2 1.4 2 100 TLB-3 1.4 2 100 BAP-1 3437580 1243458 2671 3 1 100 BAP-2 3437668 1243595 2673 6 1 100 Phase III (Construction Stage - 2004) SBH-1 3437672 1243488 2638 20 5 100 SBH-2 3437477 1243271 2638 20 11 5 3 58 26 16 9.10 19 19 SBH-3 3437252 1243314 2658 25 24 100 24 18 SBH-4 3437597 1243185 2633 20 6 13 2 29 62 10 4.64 19 11 SPD-5A 3437587 1243539 2672 80 31 3 5 79 8 13 44.78 29 29 SB-1 3437278 1243445 2667 3 4 100 3 3 SB-2 3437278 1243366 2652 3 4 100 3 2 SB-3 3437275 1243287 2650 3 3 1 75 25 3 1 SB-4 3437358 1243445 2662 3 1 3 25 75 3 2 SB-5 3437436 1243525 2661 3 4 100 3 2 SB-6 3437357 1243366 2644 3 5 100 3 3 SB-7 3437516 1243366 2652 3 4 1 80 20 3 SB-8 3437596 1243366 2637 3 5 1 83 17 3 SB-9 3437652 1243432 2633 3 5 100 3 1 Summary:Overall Investigations 80 11 155 73 92 3.32 46.8 22.1 27.8 11.71 179 145 Phase I (1988) 18 5 3 41 10.2 6.12 83.7 2 8 8 Phase II (2003) 78 11 75 34 13 8.27 56.4 25.6 9.77 10 53 46 Phase III (2004) 80 75 36 38 50.3 24.2 25.5 20 118 91
113
The exploratory efforts have generally indicated that the alluvium is made up largely of boulders, skip graded gravels, cobbles, and variable fines from sand to silt with traces of clay and a few isolated lenses. It is also considered that the alluvium in the river bed is more than 91.44 m deep at the dam site under consideration. With the exception of an outcrop of granodiorite, exposed at a higher elevation towards the right abutment of the proposed dam axis, bed rock is situated at much greater depths
(more than 100 m).
Permeabilities recorded from the overall field tests conducted throughout the subsurface investigations exhibit a generous range of the order of 10-1 to 10-5 cm/sec, over variable depths. At some levels, excessive water losses were encountered and tests were abandoned. A total of 145 permeability tests were successfully completed at different locations/elevations, out of 179 test attempts. The phase wise distribution includes 8, 53 and 118 test attempts through the three phases respectively. Nine (9) tests resulted in K-values lower than 0.001 cm/sec out of which 6 values pertain to pressure test results on rock (equivalent permeabilities from lugeon values recorded).
The remaining three values are aberrant, one pertains to K-value from test performed on boulders 8.87 x 10-5 cm/sec (SDL-3; 5.5 m depth) and the two others pertain to tests in gravels 6.1 x 10-4 cm/sec (SDL-3; 7.62 m depth) and 9.49 x 10-4 cm/sec (SPD-
5A; 78 m depth).
Permeability-Elevation plots are shown in Figure 5.1 corresponding to the investigative field testing done in boreholes along the main dam axis and along the cofferdam axis upstream of the reservoir. It is seen that the permeability trend increases at upper layers towards the nullah bed. In general no specific depth dependence is observed to be associated with the recorded permeabilities. However, it
114
Elevation (m) Elevation (m) 2560 2580 2600 2620 2640 2660 2680 2600 2610 2620 2630 2640 2650 2660 2670 100 100
10 10
1 1
0.1 0.1
0.01 0.01
0.001 0.001
0.0001 0.0001 Permeabilities (cm/secPermeabilities - log scale) Permeabilities (cm/sec - log scale) 0.00001 0.00001 SDL-5 SPD-5A SDL-3 SDR-1 SDR-2 SBH-3 SBH-2 BH-6 BH-3 SD-1 BH-1 SBH-4
a. b.
Figure 5.1 Permeability distribution and the corresponding investigative drilling
(a) along main dam axis and (b) along existing old dam / cofferdam axis
115
is seen in some boreholes, like SD-1, SDR-1, SDR-2 and SDL-3 that the higher order permeabilities, i.e. K > 10-1 cm/sec, are present at upper levels towards the nullah bed.
A similar inference is drawn for the shallower depths from the water losses in test pits. Nevertheless, the subsurface composition evaluated through the exploratory efforts has ascertained absence of large scale stratification in the foundation stratum.
5.1.2 Geological Evolution of the Depositional Environment
The major findings from investigatory efforts are correlated hereunder with a literature search for a relative description of the geological evolution of the subsurface conditions and knowledge of the prevalent depositional environment. Subsequent surface and subsurface geological studies have revealed that the overburden deposits of the strata actually comprise of moraines of dual nature, resulting from more than one phenomenon. This postulation has been a point of interest as evident from the following referenced literature.
Norin (1925) and Shroder et al. (1993) have viewed the Satpara Lake as being dammed-up by a terminal moraine. Burgisser et al. (1982), Owen (1988) and Cronin
(1989) interpreted the massively disturbed gravel fan and lacrustine sediments as
“glaciotectonic or disturbed till”. This contradicts Hewitt (1999) who not only queried the interpretation as “glaciotectonic”, but also considered the accumulation damming- up the Satpara Lake as being a “rock slide or rock avalanche”. Hewitt (2002) projected the plan along with long and cross profiles of the Satpara Lake avalanche, categorizing it under extreme impact slope effects and has also tabulated the pertinent location, dimension and lithology. Hewitt (2002) has given pictorial evidence (Fig. 15
116
in the referred paper) of the main accumulation of around 950 m depth above the
Satpara stream, and the picture somewhat also resembles the present dam axis.
Kuhle (2001, 2004), however, states that besides the correctly described rock avalanche deposits and river terraces etc., accumulations of lodgement till do also exist in different altitudinal positions and states of preservation above those rock avalanche deposits on the left side, 2 to 3 km up-valley of the Satpara Lake.
It is thereby established from the references cited above, that two rare but main interactive phases identified relevant to the project site are the rock avalanche(s) triggered in the area and the response of extensive glaciations. Subsequent evaluations are made herein through personal communications of the author with Mr. Arshad
Hashmi, Principal Geologist of SDC on site2.
A receding glacier triggered an avalanche at the dam site. The material of the slopes was broken up by frost-shattering action and the valley was blocked. The subsurface investigations validate this supposition because the material is angular gravels and boulders in silty matrix. This silty matrix was produced by the grinding action of boulders and gravels during the avalanche slide. The most prominent covering layer of the deposit at the dam site comprises only angular gravels and boulders, classified as terminal moraine deposits. The presence of lateral moraines along the lake edges substantiates that after the avalanche; glaciations reemerged and by the end of the Ice Age period, resulted in additional depositions over the avalanche
2 Personal communication, December 2004.
117
layer. The glacier remained to flow for a longer period of time along the sides and over the avalanche deposits, eroding away the fines. Permeabilities also correspond to the standard materials identified, e.g. higher permeabilities along the sides and in the middle of the valley. A transitional shift in the glacial flow pattern at later stages is indicted from smaller depths of high permeability zones in the middle. It also suggests that the Satpara Lake-Skardu rock avalanche must have taken place very close to the period of deglaciation. Primarily all references have implied absence of stratification in the foundation stratum under a predominant glacial depositional environment with its geological evolution relating two mutually supporting phenomena, i.e. avalanches and recession of glaciers.
5.2 ASSESSMENT OF FOUNDATION PERMEABILITY
5.2.1 Initial Inferences
Permeability of the substrata is the greatest concern, when related to underseepage potential for the foundation. The deposits forming the foundation strata at site are an unsorted mixture of ground-up rock debris, constituting a complex material composition of moraines, i.e. from gravels and boulders to little binding material like silt and clay as repeatedly identified in the field logs. The determination of a representative permeability value for design was based on the results of field test attempts made in boreholes and test pits. The permeability field tests cover only a part of the Satpara valley within selective probing and excavated pits.
Permeabilities and related log descriptions did not represent foundation stratification or chronological layers of any kind. Relating the information from tests conducted through the different investigation phases (Figure 5.2) only sufficed in
118
overlapping the K-data obtained over extended test depths. A modified plot of the K- data, to investigate dependence of permeability distributions on the nullah proximity
(Figure 5.3) was also inconclusive. Although a decrease in permeability was found at certain depths, it was not sufficient to be considered as rationale for depth dependence of the permeability data. This uncertainty was explicable since most of the permeability tests were made in the upper 50 meters but the permeability values were distributed over variable depths. It was evident that no specific depth dependent trend could be noticed in the permeability for the foundations.
5.2.2 Statistical Inferences
The wide variation in permeability from the K-D plots in Figure 5.2 and
Figure 5.3 also indicated that the order of magnitude of permeability was more important than establishing a relationship for anisotropic variations between vertical and horizontal permeability. This aspect implied using parametric / descriptive statistics for numerical estimates of "true" representative values. This endeavor was considered to reveal underlying patterns in the K-data not readily observable and help separate the probable from the possible. The available K-data was quantitative or continuous, described best as a scale variable. Summary measures for statistical inferences included measures of central tendency and measures of dispersion. The most common measures of central tendency were the mean (arithmetic average) and median (value at which half the cases fall above and below). Consequently, statistics that measured the amount of variation or spread in the data included the minimum, maximum, inter-quartile range, standard deviation, and variance.
119
Depth (m) 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 100
10
1
0.1
0.01
0.001 Permeabilities (cm/sec)
0.0001
0.00001 Phase 1 (Feasibility Stage 1988) Phase 2 (Design/Tender Stage 2003) Phase 3 (Construction Stage 2004)
Figure 5.2 Phasewise distribution of permeability test attempts
Depth (m) 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 100
10
1
0.1
0.01
0.001 Permeabilities (cm/sec)
0.0001
0.00001 Near Nullah Surroundings
Figure 5.3 Distribution of permeability with reference to Nullah Proximity
120
For the available data from all test attempts, there was a large difference
between the mean and the median (Figure 5.4). The mean was greater than the median
for about three orders of magnitude, indicating that the values were not normally
distributed. Table 5.2 reports descriptive statistics, providing information about the
central tendency and dispersion variability for the permeability data set. The majority
of K-values from all test attempts were clustered at the lower end of the scale, with
most falling below 0.01 cm/sec (refer Figure 5.2 and Figure 5.3). There were,
however, 35 cases with values above one (1) cm/sec, out of which only one with a
value of 3.027 cm/sec corresponded to an actual test attempt. All other values related
to the extreme water losses causing the test attempts to fail (the K-value reported
therein was not calculated but represented the percent water loss being more than 50%
to failure). These high values had a significant effect on the mean but little or no
effect on the median, making the median a better indicator of central tendency.
Table 5.2 Descriptive Statistics for the permeability data set
All Test Successful Tests neglecting values >1 or Descriptive Statistics Attempts Tests <0.001
(179 Tests) (145 Tests) (135 Tests) Median (50th Percentile) 0.015 0.00953 0.0105 Central Tendency Mean 18.763 0.059 0.041
Mode 100 0.168 0.168
Minimum 0.0000858 0.0000858 0.001
Maximum 100 3.027 0.38
25th Percentile 0.00393 0.00317 0.00373
Dispersion 75th Percentile 0.16 0.0314 0.0416
Interquartile Range 0.156 0.028 0.038
Standard Deviation 38.820 0.257 0.069
Variance 1506.997 0.066 0.005
121
Figure 5.4 Measures of central tendency for the permeability dataset
The data was rearranged by disregarding the tests failed to have a more realistic range. Summary statistics from the successful tests represented a closer compliance between the mean and median with still one order of magnitude in- between. Only one value of 3.027 cm/sec exceeded 1 cm/sec. The main reason for this difference in the orders of magnitude was explained through the few values (a total of only eight values) falling below 0.001 cm/sec.
In the third trial for statistical inferences of the K-data, an assumption was tailored to exclude all values greater than 1 cm/sec and all values less than 0.001 cm/sec as outliers. This final attempt gave an equal correlation between the orders of magnitude for the mean and median values through the available K-data range. Figure
5.5 below shows the representative data set between 1 cm/sec and 0.001 cm/sec, highlighting the extreme values as outliers over a categorical scale of depth ranges for a better insight of the permeability distribution over depths.
122
Frequency histogram for the K-data corresponding to all the test attempts along with a breakup of this histogram with respect to the outliers is shown in Figure
5.6. The frequency plot corresponding to the representative dataset approximates a log normal probability distribution. Table 5.3 summarizes the overall quantitative permeability data. It describes a comprehensive data count, its frequency distribution and the related descriptive statistics excluding outliers, i.e. neglecting the extreme K- values greater than 1 cm/sec and less than 0.001 cm/sec. The data description was categorically distributed further over the investigation phases, presumed depth ranges as depicted from Figure 5.5 (based on data concentration) and material composition
(based on the logs summary from Table 5.1).
m
m
m
85 30 ‐ m ‐
15 8 ‐ 30 ‐ 15
8
0
I: II: III: IV:
Range Range Range Range
Depth Depth Depth Depth
Figure 5.5 Representative permeability data set
123
100.00% 100 100% Frequency 90 90% 80.45% 81.01% Cumulative % 80 80% 70.95% 70 70% 60 60%
50 42.46% 50% 40 40% Frequency
66 % Cumulative 30 30% 51 20 20% 34 5.59% 10 1.12% 17 10% 2 8 1 0 0%
Impermeable Outliers Representative Dataset Permeable Outliers 0.0001 0.001 0.01 0.1 1 10 Tests Failed (100% Water Permeability (cm/sec) Loss)
Figure 5.6 Frequency Histogram for the permeability test results
Cumulative box plots in Figure 5.7 summarize the statistical data of the representative permeability dataset excluding outliers by showing the median, centre half of the data (the inter-quartile range), and extremes. The red line passing through the blue boxes is the median (middle value). The ends of the lines on either sides of each blue box (the ‘whiskers’) are the maximum and minimum values for the specific permeability range. The blue box ends are the first and third quartiles (25th and 75th
Percentile values), so that the blue box’s length is equal to the inter-quartile range (the difference between the two quartiles) and includes approximately half the data. The sole values depicted in the quantitative description from Table 5.3, conducted for cobbles and clay were correspondingly indicated as a single line in Figure 5.7. The best representative data range was reflected from the phase II (2003) investigations, was between depth ranges less than 30 m, and covered a coarser material range of boulders, gravels and sand.
124
Table 5.3 Quantitative description of permeability data
Investigation Phases Nullah Proximity Depth Range (m) Material Composition Overall Description Phase I Phase II Phase III Near Distribution Surroundings 0 - 8 8 - 15 15 - 30 30 - 85 Rock Boulders Cobbles Gravels Sand Silt Clay (1988) (2003) (2004) Nullah All Test Attempts 179 8 53 118 46 133 61 36 42 40 8 105 1 28 12 24 1 Failures Excluded 145 8 46 91 33 112 46 27 34 38 8 90 1 14 11 20 1 Neglecting Extremes (>1 and <0.001) 135 7 38 90 30 105 43 27 32 33 2 89 1 12 11 19 1 Pervious (>1) 35 1 7 27 14 21 16 9 8 2 0 15 0 14 1 5 0 Impervious (<0.001) 9 08127 20256102000
K-Data Count K-Data Extremely Pervious (>10) 34 0 7 27 13 21 15 9 8 2 0 15 0 14 1 4 0 Extremely Impervious (<0.0001) 2 02011 10011100000 0.0001 2 02011 10011100000 0.001 8 16126 10345002100 0.01 6609573 631410152714805480 0.1 511193118332313105 13215471 1 17510289 64610902240 K-Frequency 10 1 10010 10000000010 Tests Failed (Water Loss) 34 0 7 27 13 21 15 9 8 2 0 15 0 14 1 4 0 25th Percentile 0.00373 0.1125 0.01173 0.00313 0.016025 0.003320 0.009315 0.00712 0.00748 0.00268 0.02113 0.00332 0.07 0.00839 0.00633 0.00395 0.049 Max 0.38 0.2 0.38 0.135 0.196000 0.380000 0.2 0.28 0.38 0.123 0.068 0.38 0.07 0.168 0.196 0.2 0.049 Min 0.001 0.001 0.0015 0.00136 0.001000 0.001360 0.00136 0.0015 0.001 0.00147 0.0055 0.00147 0.07 0.00161 0.001 0.00136 0.049 75th Percentile 0.04155 0.168 0.10675 0.01488 0.106750 0.021000 0.05805 0.02565 0.08875 0.00548 0.05238 0.023 0.07 0.03405 0.1 0.04175 0.049 Median (50th Percentile) 0.0105 0.168 0.0255 0.00799 0.043300 0.008570 0.0158 0.0115 0.012 0.00327 0.03675 0.00933 0.07 0.01165 0.0738 0.0112 0.049 Mode 0.168 0.168 0.1 0.0105 0.150000 0.168000 0.168 0.015 0.15 0.017 #N/A 0.015 #N/A #N/A 0.1 0.168 #N/A Mean 0.0405984 0.13286 0.07975 0.01689 0.065142 0.033586 0.0438991 0.0376489 0.0684572 0.0116962 0.03675 0.03585 0.07 0.04178 0.06556 0.04607 0.049
(Excluding Outliers) (Excluding Interquartile Range 0.03782 0.0555 0.09503 0.01175 0.090725 0.017680 0.048735 0.01853 0.08127 0.0028 0.03125 0.01968 0 0.02567 0.09367 0.0378 0 Descriptive Statistics Descriptive StDev 0.06858071 0.06713 0.10074 0.02581 0.060124 0.069488 0.0575263 0.0637668 0.1000228 0.0263904 0.04419 0.07185 #DIV/0! 0.06089 0.06535 0.0667 #DIV/0! Var 0.00470331 0.00451 0.01015 0.00067 0.003615 0.004829 0.0033093 0.0040662 0.0100046 0.0006965 0.00195 0.00516 #DIV/0! 0.00371 0.00427 0.00445 #DIV/0!
125
Maximum
75th Percentile Median (50th 25th Percentile) Percentile Minimum
Figure 5.7 Box plots for representative permeability data set
Generally, the scatter in the K-values from the initial K-D plots of the in-situ test data reflected uncertainties as if derived from some random process (aleatoric).
However, statistical insight showed that the variation was actually spatial, not random and the corresponding statistical errors were derived from the limited numbers of observations categorized as epistemic (lack of knowledge). The cumulative inferences from the permeability data set showed that a representative model for the heterogeneous foundation permeability would be associated with absence of depth dependence or any other trend.
5.2.3 K-Value Estimates for Modeling Dam Foundations
The solution of seepage problems is made easier when k, the coefficient of permeability, is constant. The flow lines (stream lines) and the lines of equal head
(equipotential lines) in the medium intersect at right angles. The median of the overall dataset for Satpara indicated 50 percent of the data having permeability around 0.01
126
cm/sec, with the standard deviation also in agreement. This corresponded to a representative permeability value of 0.01 cm/sec for a homogeneous foundation approximation scenario. The assumption of a constant k is rarely justified in practical problems of flow in natural soils, however this is often assumed for an initial estimate.
The foundation stratum at Satpara Dam is not homogeneous, as described in
Chapter 3 and appended in the log descriptions. The basic approximations and initial estimates for numerical modeling would thereby readjust the foundation assumptions to comprise a homogeneous stratum with at least one intermediate pervious zone. The corresponding foundation permeabilities would have to be separated by appropriate differences in orders of magnitude for effective modeling. The adopted construction design of Satpara Dam considered a two-layered approach to model the heterogeneous foundations. The selected K-value for the dominant foundation zone was an average of the successful test results (0.059 cm/sec) and the pervious zone was modeled with a higher K-value selected from literature corresponding to open-work zones/gravels in general (1 cm/sec).
From a statistical insight as discussed above, the most frequently occurring value from the data set, represented as the mode for the overall dataset was around 0.1 cm/sec, which is relatively pervious. If the mode was considered to correspond for pervious zone permeability, a minimal difference of two orders of magnitude towards the lower side can be adopted for the overall foundation in relative homogeneity. As such, using a homogeneous foundation with an average permeability of 10-3 cm/sec and a continuous pervious zone having higher order permeability of 10-1 cm/sec was used for the initial estimates of ‘as designed’ conditions for seepage modeling.
127
Utilization of Rockworks (described in Chapter 4) on the complete permeability dataset with Closest Point Algorithm actually coupled geostatistics and optimization, to characterize the spatial variability of the in-situ test results from pre- construction exploratory investigations into different K-zones. Modeling the dam foundations for seepage analyses aimed at evaluating the relative efficiency of seepage control measures. Over parameterization in such analyses could mask serious systematic errors. The RW foundation model was thereby utilized to filter the extreme zones (permeable i.e. >1 cm/sec and impermeable i.e. <0.0001 cm/sec) for four (4) K- zones ranging from a maximum K-zone permeability of 10-1 cm/sec to a minimum of
10-4 cm/sec. Figure 5.8 shows the sectional dam foundation approximations used for the initial model estimates of optimizing and selecting the most effective seepage control measure, through the multivariate sensitivity analysis.
Figure 5.8 Foundation profiles for SCM sensitivity analyses
It is worth noting that the permeability frequencies depicted in Figure 5.6 refer to the dataset distributed over the complete investigated area. The K-zone frequencies differ in the Rockworks heterogeneous foundation model from Figure 5.8, because of
128
the sectional orientation, correlating with the actual permeability data from the boreholes along the specified section, in agreement with the described representative dataset.
5.3 MULTIVARIATE SENSITIVITY ANALYSES
5.3.1 General
The relative distribution of selected comparative parameters i.e. hydraulic gradients and total heads corresponding to the selected points (points 1 through 4, described in the previous chapter) were determined for all the analyzed case scenarios. The 50th percentile values for the homogeneous foundation PZ approximation were compared with values from the heterogeneous RW modeled foundation approximation. The results presented hereunder have been distributed in a two phased relative sensitivity comparison, i.e. the adoption of a primary seepage control measure and secondly the selection of an adequate combination measure.
5.3.2 Upstream Impervious Blanket versus Partial Cutoff Wall
Computed results for the blanket (B) cases, B1 to B5, are given in Table 5.4 alternately for the two different foundation profile approximations, homogeneous with pervious zone (PZ) and heterogeneous modeled with RockWorks (RW). Figure 5.9 shows the comparative results, blanket cases designated numerically from 1 to 5, along the points of consideration P1 to P4. Similar tabulated results for the cases representing only a partial cutoff wall under the core, are given in Table 5.5. This case scenario was considered as an alternative to the upstream impervious blanket. Figure
5.10 shows comparative results of the partial cutoff wall (C) cases.
129
Table 5.4 Computed Gradients and Total Heads at Selected Points for Case Scenarios of Impervious Upstream Blanket
Gradients at: Total Heads at: Case Scenarios P1 P2 P3 P4 P1 P2 P3 P4 B1 PZ 0.0274 0.1363 2.1193 0.7762 2664.0 2664.0 2638.5 2633.5 (None) RW 8.6E-05 0.0255 0.3149 0.1019 2664.0 2664.0 2637.6 2629.2 B2 PZ 0.0778 1.0117 1.1712 0.0226 2661.7 2656.2 2638.0 2633.2 (1.5 m x ½ L) RW 0.0004 0.0537 0.1418 0.0605 2664.0 2664.0 2631.8 2629.2 B3 PZ 0.0837 1.0017 1.1439 0.0229 2661.8 2657.5 2638.0 2633.2 (3 m x ½ L) RW 0.0004 0.0579 0.1478 0.0688 2664.0 2664.0 2631.6 2629.2 B4 PZ 0.2099 1.2770 1.0577 0.0203 2661.9 2657.3 2638.0 2633.2 (1.5 m x L) RW 0.0131 1.8652 0.0815 0.0623 2663.8 2652.3 2631.4 2629.2 B5 PZ 0.2527 0.6978 0.9318 0.0183 2661.6 2657.3 2638.0 2633.2 (3 m x L) RW 0.0129 0.9887 0.0832 0.0595 2663.8 2653.1 2631.6 2629.2 Note: See Table 4-3 for case description and Figure 4-7 for location of analysis points P1, P2, P3 and P4.
Figure 5.9 Computed Gradients and Total Heads at Selected Points for Case Scenarios of Upstream Impervious Blanket
130
The gradients increased consecutively from point 1 to point 2, with the least differences corresponding to the blanket case scenario 5. Point 3 notably differed in the trends for the computed gradients, representing the onset of downstream drainage blanket. Point 4 marked the seepage exit / collection at the toe drain, where the sensitivity parameters (gradients and total heads) were minimized for all case scenarios. The gradients decreased for each consecutive case scenario at the last two points P3 and P4. The total heads on the other hand consistently decreased in the direction of flow, with the maximum difference corresponding to the application of a full length blanket for case scenarios B3 and B4. These results primarily indicated the dominant effect of media permeability through the initial points of consideration (P1 and P2) where the gradients were observed to increase with a decrease in total head.
This effect was evident through steepness of the gradients for consecutive case scenarios, indicating the corresponding sensitivity of gradients to assumed permeability values in foundations. The refinement in K-assumptions from layered to the multi-zoned approach modified the comparative scale of magnitudes for improved relative inferences. The pronounced visual differences from comparison of results from both foundation approximations also implied in this regard. The head dependency of computed gradients revived later at points P3 and P4 where the gradients decreased with the total heads.
In order to further simplify the representation and to arrive at conclusive recommendations, results of only five out of the ten analyses selected based on the cutoff penetration depths at 6 m (20 ft) intervals, i.e. designated consecutive case scenarios C2 to C10, are described for only the partial cutoff wall under the core (C) cases as reflected in Table 5.5 and Figure 5.10.
131
Table 5.5 Computed Gradients and Total Heads at Selected Points for Case Scenarios of Partial Cutoff Wall under Core
Gradients at: Total Heads at: Case Scenarios P1 P2 P3 P4 P1 P2 P3 P4 C2 PZ 0.0248 0.1229 1.3072 0.7877 2664.0 2664.0 2637.8 2633.5 (6 m) RW 8.4E-05 0.0247 0.4107 0.1016 2664.0 2664.0 2637.2 2629.2 C4 PZ 0.0236 0.1123 1.3749 0.7509 2664.0 2664.0 2637.7 2633.4 (12 m) RW 6.7E-05 0.0192 1.6015 0.1031 2664.0 2664.0 2636.0 2629.2 C6 PZ 0.0156 0.0704 1.4635 0.6966 2664.0 2664.0 2637.7 2633.4 (18 m) RW 5.4E-05 0.0153 2.4373 0.0970 2664.0 2664.0 2635.0 2629.2 C8 PZ 0.0084 0.0362 1.5196 0.6377 2664.0 2664.0 2637.7 2633.4 (24 m) RW 3.7E-05 0.0185 2.8448 0.0763 2664.0 2664.0 2632.7 2629.2 C10 PZ 0.0057 0.0266 1.5321 0.5939 2664.0 2664.0 2637.7 2633.4 (30 m) RW 3.7E-05 0.0185 2.8452 0.0763 2664.0 2664.0 2632.7 2629.2 Note: See Table 4-3 for case description and Figure 4-7 for location of analysis points P1, P2, P3 and P4.
Figure 5.10 Computed Gradients and Total Heads at Selected Points for Case Scenarios of Partial Cutoff Wall under Core
132
These cases were designated even numbers from C2 to C10, corresponding to typical consecutive scenario of incremental penetration depths of 6 m (20 ft), i.e. 6
(20), 12 (40), 18 (60), 24 (80) and 30 (100) m (ft) deep cutoff wall penetration depths respectively, along the different points of consideration (P1 to P4). The gradients increased relatively from point 1 to point 2, with a consistent decrease prominent for the consecutive cases at the individual points. A steady rise in the computed gradients was notable at point 3. Point 4 followed up with the characteristic reduction in gradients due to the exit point orientation. The total heads decreased in the direction of flow, with relative uniformity observed through each consecutive case scenario at each point. The comparative sensitivity trends through each point of consideration favored the deepest penetration depth for all points excluding point 3. The sensitivity of gradient distributions to the different permeability assumptions through a modified comparative scale of magnitudes was also pronounced under these results. The distinct variation of comparative results from both foundation approximations, especially at P1 and P4, implied at the dominant effect of media permeability on the sensitivity of gradients. A deeper penetration depth appeared to overcome this anomaly as depicted from the converging gradients at P2. Whereas, gradients higher than theoretically acceptable values of unity, i.e. 1.0 at point 3 for consecutive case scenarios indicated the hazards of probable piping initiation at the drainage blanket.
The difference in comparative trends of hydraulic gradient at point 3 corresponding to distinctive permeability assumptions was indicative in this regard.
The comparative results indicated relative reliability of placing a horizontal impervious blanket over the foundations as the primary selection for effective seepage control in absence of bedrock, rather than constructing a partial cutoff wall under the
133
core penetrating into the foundations. The results indicated indispensable considerations of a greater length and excessive thickness for the upstream impervious blanket. They also implied that a deeper cutoff wall when partial in nature may not be advisable without considering the trend of computed hydraulic gradients at critical points for indicative piping initiation.
5.3.3 Upstream Cutoff Wall versus Downstream Cutoff Wall
Taking into account the primary seepage control provision comprising a full length of placement for a 3 m thick upstream blanket, accentuated by the initial trials, the next phase of analysis trials was aimed at a second / additional line of defense.
The subsequent trials were intended to select a cutoff wall (partial in nature due to absence of bed rock) combination alternate, with respect to its position in addition to the impervious blanket, which can reduce the gradient distribution range (preferably under 1) for all points of consideration and control comparative fluctuations in a more consistent manner as compared to the other position alternate. Table 5.6 presents the computed results whereas Figure 5.11 and Figure 5.12 show comparative results of the case scenarios representing these combination cases of a partial cutoff wall upstream of the impervious blanket (UC) and downstream of the impervious blanket repositioned under the core (DC).
A reversal in the comparative trends was apparent for both position alternates along points 1 and 2. Computed gradients for the partial cutoff wall upstream of the blanket were increasing, while conversely a persistent decrease was shown in the gradients for the partial cutoff wall downstream of the blanket. The total head values on the other hand decreased for the upstream cutoff wall and increased for the
134
downstream cutoff wall. Both the foundation approximations generally supported the same trends but the influence of different permeability approximations was prominent through the considerable variations, specifically for the gradients computed at P1.
Point 3 however was again a critical situation, for both the position alternates.
The values of computed gradients negotiated a mildly decreasing trend for the upstream cutoff wall, while the trend for the downstream cutoff wall was reversed to a mildly increasing one. A distinct variation was obvious between comparative results for the different foundation approximations for each position alternate. The upstream alternate showed parallel correspondence, indicating stability of the gradients at different magnitudes, whereas the downstream alternate indicated converging gradients with increased penetration depths, even exceeding unity.
Values for the total head were relatively linear and consistent. The trends decreased from P1 to P2 for the upstream cutoff wall, whereas they increased for the other position alternate. The median results of the homogeneous foundation approximation for the upstream position alternate slightly pronounced the trends after a penetration depth of 15 m (50 ft). Total heads at P3 were similar for both alternates.
The gradients and total heads both at Point 4 followed the same trends for both cutoff wall alternates. Apparently, it seemed that gradients at point 3 were the decisive element in optimizing the combination of effective seepage control combination measure, favoring the upstream cutoff alternate. Gradients gradually reduced; stayed consistent and controlled with the least comparative fluctuations in the upstream position alternate for the cutoff wall as compared to the downstream substitute for an additional seepage control measure.
135
Table 5.6 Computed Gradients and Total Heads at Selected Points for Combination Case Scenarios of a Partial Cutoff Wall with 3 m Thick and Full Length Impervious Upstream Blanket Gradients at: Total Heads at: Case Scenarios P1 P2 P3 P4 P1 P2 P3 P4 UC1 PZ 0.346 0.601 1.029 0.021 2661.9 2657.3 2638.0 2633.2 (3 m) RW 0.015 0.989 0.083 0.059 2663.8 2653.1 2631.6 2629.2 UC2 PZ 0.426 0.579 1.047 0.021 2661.8 2657.5 2638.0 2633.2 (6 m) RW 0.018 0.989 0.083 0.059 2663.8 2653.1 2631.6 2629.2 UC3 PZ 0.512 0.594 1.029 0.021 2661.6 2657.3 2638.0 2633.2 (9 m) RW 0.022 0.989 0.083 0.060 2663.8 2653.1 2631.6 2629.2 UC4 PZ 0.560 0.604 1.026 0.020 2661.5 2657.2 2638.0 2633.2 (12 m) RW 0.025 0.990 0.100 0.060 2663.8 2653.1 2631.6 2629.2 UC5 PZ 0.590 0.633 0.932 0.019 2661.4 2656.9 2638.0 2633.2 (15 m) RW 0.030 0.991 0.083 0.060 2663.8 2653.1 2631.6 2629.2 UC6 PZ 1.812 0.939 0.831 0.018 2656.9 2653.5 2637.9 2633.2 (18 m) RW 0.036 0.992 0.083 0.059 2663.7 2653.1 2631.6 2629.2 UC7 PZ 2.437 1.194 0.681 0.018 2654.5 2650.6 2637.9 2633.2
Upstream Cutoff Wall Upstream Cutoff Wall (21 m) RW 0.043 0.993 0.090 0.059 2663.7 2653.1 2631.6 2629.2 UC8 PZ 2.672 1.208 0.657 0.017 2653.8 2650.5 2637.8 2633.2 (24 m) RW 0.055 0.994 0.068 0.062 2663.7 2653.1 2631.6 2629.2 UC9 PZ 2.834 1.240 0.642 0.017 2653.4 2650.1 2637.8 2633.2 (27 m) RW 0.096 1.002 0.068 0.061 2663.5 2652.9 2631.5 2629.2 UC10 PZ 2.862 1.260 0.633 0.016 2653.3 2649.9 2637.8 2633.2 (30 m) RW 0.200 1.025 0.084 0.059 2663.0 2652.7 2631.6 2629.2 DC1 PZ 0.203 0.609 0.656 0.022 2662.3 2657.2 2637.7 2633.2 (3 m) RW 0.013 0.990 0.089 0.063 2663.8 2653.1 2631.6 2629.2 DC2 PZ 0.183 0.556 0.719 0.021 2662.4 2657.8 2637.7 2633.2 (6 m) RW 0.013 0.984 0.089 0.057 2663.8 2653.1 2631.6 2629.2 DC3 PZ 0.173 0.525 0.795 0.020 2662.5 2658.1 2637.7 2633.2 (9 m) RW 0.013 0.986 0.093 0.060 2663.8 2653.2 2631.5 2629.2 DC4 PZ 0.141 0.445 0.816 0.021 2662.7 2659.0 2637.6 2633.2 (12 m) RW 0.013 0.969 0.217 0.055 2663.8 2653.4 2631.5 2629.2 DC5 PZ 0.135 0.397 0.945 0.020 2662.8 2659.5 2637.6 2633.2 (15 m) RW 0.012 0.950 0.264 0.059 2663.9 2653.6 2631.5 2629.2 DC6 PZ 0.127 0.389 0.955 0.020 2662.8 2659.6 2637.6 2633.2 (18 m) RW 0.012 0.922 0.408 0.057 2663.8 2654.0 2631.5 2629.2 DC7 PZ 0.107 0.341 1.072 0.019 2663.0 2660.2 2637.6 2633.2 (21 m) RW 0.011 0.815 0.893 0.058 2663.9 2655.2 2631.5 2629.2 Downstream Cutoff Wall DC8 PZ 0.092 0.304 1.095 0.018 2663.1 2660.6 2637.7 2633.2 (24 m) RW 0.011 0.815 0.894 0.058 2663.9 2655.2 2631.5 2629.2 DC9 PZ 0.089 0.287 1.215 0.018 2663.2 2660.8 2637.6 2633.2 (27 m) RW 0.011 0.815 0.894 0.058 2663.9 2655.2 2631.5 2629.2 DC10 PZ 0.066 0.213 1.279 0.016 2663.4 2661.6 2637.7 2633.2 (30 m) RW 0.011 0.815 0.895 0.058 2663.9 2655.2 2631.5 2629.2 Note: See Table 4-3 for case description and Figure 4-7 for location of analysis points P1, P2, P3 and P4.
136
Figure 5.11 Computed Gradients and Total Heads at Selected Points for Combination Case Scenarios of a Partial Cutoff Wall at the upstream end of a 3 m Thick and Full Length Impervious Upstream Blanket
Figure 5.12 Computed Gradients and Total Heads at Selected Points for Combination Case Scenarios of a Partial Cutoff Wall at the downstream end (Repositioned under Core) of a 3 m Thick and Full Length Impervious Upstream Blanket
137
5.3.4 Selected Seepage Control Combination
The multivariate sensitivity of computed gradients and their distribution was shown to vary with the different foundation approximations and conversely with the case scenarios. The representation of total heads in the graphs highlighted a relative influence on the head distribution as well. The computed gradients were shown to vary inversely with assumptions of foundation permeability, showing no appreciable corresponding effects on the head distribution. The results however revealed a consistent trend in head reduction, in the direction of flow. Generally, at each point of consideration, the rate of change in total heads was represented by the comparative steepness for consecutive case scenarios. On the other hand, sensitivity of functioning gradient distributions was described by the scale of order.
Earlier references e.g. Cedergren (1972) had established that for a cutoff wall, partial in nature, to adequately function in controlling seepage and reducing gradients deeper penetration depths are preferential. However under consideration of the results from sensitivity analyses for partial cutoff walls as described above, the optimal depth of a cutoff wall and its relative position in combination to an upstream impervious blanket has to consider functioning distributions of both gradients and hydraulic heads within the foundations.
For the given set of foundation approximations and the analyzed depths of the cutoff walls, a depth of 15 m (50 ft) showed acceptable inflections, under a controlled threshold of unity (1), for gradient distributions through the variable range at all points of consideration. Both for the PZ and the RW foundation representations, a relative consistency in the overall decreasing head distribution established an
138
upstream position alternate of the cutoff wall as the best case scenario combination
seepage control measure with the impervious blanket.
This outcome of the present study is in line with the conclusions of Malik et
al. (2008), who differed in their approach to foundation heterogeneity along a
sectional model. They modeled K-zone frequencies in a similar sectional model
correlated from all the recorded K-dataset, while alternately this study maintains the
actual orientation of the modeled section in the overall three dimensional
(Rockworks’) foundation model.
The selected foundation seepage control combination at Satpara was an
upstream impervious blanket, minimum 3 m thick and placed for the maximum
possible length, additionally with a partial cutoff wall at the upstream end of the
blanket of an optimal penetration depth of 15 m (50 ft). The relative effectiveness of
the selected measures of foundation seepage control can be quantified with reference
to the absence of any seepage control measure i.e case scenario B1. The computed
percent differences of the selected scheme from the no control measure scenario are
tabulated in Table 5.7.
Table 5.7 Computed Percent Differences for Gradients and Head Potentials Corresponding to Rockworks Modelled (Multi-Zoned) Foundation Representation for Relative Effectiveness of the Selected Seepage Control Measures % B1 UC5 Difference Case Scenarios (selected relative (no SCM) SCM) reduction Gradients at P3 0.3149 0.0835 73.5 % Head Potentials at P3 24.1 6.7 72.0 (w.r.t. Total Head at P4)
139
Since it has been established that the multi-zoned foundation, modeled through RockWorks is a better representation of the foundation conditions, the corresponding values have been considered for relative comparison of effectiveness.
The hydraulic potential is computed in accordance with the equation 4.2, modified for the total head differences in place of the piezometeric levels, at the point of consideration at the toe of the main dam core (P3) with zero potential at the drain ditch (P4) and 100 percent potential corresponding to the maximum conservation reservoir level. The point P3 also marks the transition from the impervious core to pervious filter zones, identified as critical for determining the influences of foundation permeability and trend distributions for optimizing and selecting the seepage control measures. The computed results verify that both control measures complement each other for an overall reduction of the total hydraulic potential to 72% and hydraulic gradients to 73.5% at toe of the main dam core.
The partial cutoff wall at upstream end of the impervious blanket developed a unique seepage regime in response to its inclusion, analogous to the insertion of an external element into a packed container intensifying the pressures within. In the same way a cutoff wall within the foundations modified the foundation response to the reservoir head in terms of the gradient distributions and the resulting head differences across and around the cutoff wall. The buildup of this excess pressure was dissipated for the upstream cutoff wall, through the creep length and travel time it needs to reach the drainage blanket. For the downstream cutoff wall alternate the drainage blanket was in close vicinity of the cutoff wall, which induced high gradients with probable chances of particle migration or ‘initiation’. The upstream cutoff wall had consequently imparted favorable circumstances downstream and proved helpful in
140
preventing subsequent stages of piping i.e., blocking the path of continuation, progression and formation of pipes, right at the upstream end. These aspects are critically scrutinized in the next section.
5.4 PERFORMANCE EVALUATION
5.4.1 Impounding
The project to-date has completed four cycles of partial impounding stages and is currently experiencing the fifth partial impounding stage. The maximum conservation level of the reservoir i.e. El. 2664 m (8740 ft) a.m.s.l. has not yet been achieved. The different stages of impounding along with salient levels and dates are graphically presented in Figure 5.13.
The maximum and minimum recorded pre-impounding reservoir levels were
2634.86m (8644.54ft) and 2633.33m (8639.55ft), respectively. The first impounding phase started on 17-08-2007, from a reservoir level of 2634.20m (8642.40ft) on 17-
08-2007. It attained a maximum RL of 2636.43m (8649.69ft) on 01-10-2007. The second impounding phase commenced on 15-05-2008, from a minimum reservoir level of 2631.03m (8632.00ft) and achieved a maximum RL of 2640.87m (8664.28ft) on 25-09-2008. The third impounding phase began from 15-05-2009 and a minimum reservoir level of 2629.96m (8628.49ft). It reached a maximum RL of 2647.37m
(8685.60ft) on 10-08-2009, which was maintained till 20-08-2009 with minor reductions of about 0.1m in-between.
141
Figure 5.13 Impounding Status of Satpara Dam
The fourth impounding stage commenced with the arrival of extreme floods on 13-05-2010, from a minimum level of 2632.19m (8635.78ft) and achieved the maximum reservoir level recorded to date of 2656.88 m (8716.8ft) on 22-08-2010.
The fifth impounding stage started from a minimum reservoir level of 2631.23m
(8632.66ft) on 15-05-2011, reached a peak level of 2648.04m (8687.80ft) on 18-08-
2011 and as of 13-10-2011, is at 2647.11m (8684.75ft). The designed maximum reservoir level being 2663.952m (8740ft), the foundations and project features have endured approximately 80 percent of the designed head to-date.
A comparative appraisal of the instrumented observations was made for selected reservoir levels with consecutive recurrence levels of +/- 0.15 m (0.5 ft) persistent for approximately more than a week to ensure a pseudo-steady state. Three
142
such pseudo-steady windows are identified in Figure 5.13. These reservoir levels correspond to an elevation around 2648 m (8687 ft) a.m.s.l. from the fifth partial impounding stage, El. 2656 m (8715 ft) a.m.s.l. corresponding to the maximum levels achieved to-date in the fourth partial impounding stage and El. 2647 m (8685 ft) a.m.s.l. corresponding to the maximum level of the third partial impounding stage.
The performance of foundations through the impounding history was done with the basic aim of distinguishing the foundation response to the anthropogenic changes in the geo-hydraulic aspects through the adopted seepage control features.
5.4.2 Seepage Observations
Under consideration of the steep valley slopes, specifically along the left abutment of the main dam and adjacent to the spillway structure (where the valley constricts to a narrow gorge, refer Figure 4.10) the hydraulic grade line seems to concentrate within the constricted downstream area of the main dam. Visual observations throughout the partial impounding stages indicated emergence of seepage from either sides of the toe of the mound, retained as a weighted berm from the access road, between the spillway and the main dam axis. The seepage trends were continuously emerging, diminishing and shifting within the downstream vicinity of the main dam axis. Subsequently, during the third impounding stage, a pipe-culvert was placed downstream of the spillway to accumulate the overall seepage for tentative measurements starting from June 26, 2009. These tentative measurements were compared later on with the limited measurements during the fourth impounding stage, at the parshall flume just downstream of the dam and the two v-notch weirs. The limited data collected from the SMSs is presented in Figure 5.14.
143
Figure 5.14 Measured Seepage Record
The flows at the gauged seepage measurement stations (Parshall flume and two v-notch weirs) dried up with the receding reservoir levels and did not re-emerge in the fifth impounding stage. Local vicinities to the seepage collection and measurement features were treated to some effect but some seepage continued to by- pass at these points. The observed and measured seepage discharges downstream of the spillway, at the piped cumulative measurement station SMS-4 (Figure 5.14), were evidence of the fact that some seepage continued to by-pass the seepage measuring stations downstream of the main dam.
This phenomenon confirmed the existence of underground springs, which originated and traversed through the prevalent boulders and gravels strata, associated with unknown patterns of seepage. These explicit natural pathways increased the improbability of effective under seepage control resulting from water level fluctuations in the reservoir.
144
The installed relief wells are present along the right side of the spillway and parallel to the downstream toe of the dam. No flow was observed from these wells, which extended to a depth of about 10 m below the ground surface. Some arbitrary water level measurements were also conducted in these relief wells which supported that no flow had occurred in the relief wells. This suggested that the upstream impermeable blanket, cutoff wall, and grout curtain are effective at reducing the head under the dam to a level where the highly permeable materials downstream of the dam, combined with the steep topographic gradient, allow the seepage water to flow down gradient at low enough potential that water does not rise to the relief well discharge invert elevation. Verbal communications with the representative site staff of
SDC during site visits at different stages of impounding, specifically after the third impounding stage, made aware that flows could be heard in several of the relief wells adjacent to the drainage ditch downstream of the parshall flume used to measure flows from the toe drain. These wells were between the toe drain and a portion of the visible seepage that exits into the area downstream of the spillway stilling basin
(measured downstream at SMS 4). This knowledge indicated chances of a probable link between the two features. This link if it does exist would actually help decrease potential pore pressures at the toe of the embankment.
The uncertainty related to the effectiveness of the upstream impervious blanket with the cutoff wall combination inherently lied within the complex material composition of the foundation moraines. The glacial deposits forming these moraines constituted marked open works of gravels, cobbles, boulders of variable sizes and little binding material like silt and clay as repeatedly identified in the field logs. In
145
consequence of this diversified subsurface structure, these deposits also comprised of highly permeable zones of unknown extents.
5.4.3 Piezometeric Observations
In view of logical representations of the actual foundation response and in order to separately evaluate the individual and combined effectiveness of seepage control measures, the results of monitoring piezometeric data and relevant discussions hereunder are divided into sub-sections separately relating to the cutoff wall and the impervious blanket. Evaluation of the adopted combination is discussed in an additional section for piezometeric potential plots and a closure section sums up the inferences.
5.4.4 Cutoff Wall and Corresponding Piezometeric Data
The cutoff wall at Satpara Dam was installed upstream of the impervious blanket. The cutoff wall was 370 m long, 1 m wide and 15 m deep, constructed underground to impede under-seepage. The cutoff wall starting from right abutment ran almost parallel to the coffer dam and ended at the left abutment. A slurry trench type was selected out of the various different types and installation methods (Bruce et al., 2006). The excavations were carried out using a Kelly grab, by the "alternate panels method" with continuous overlapping of 6 m primary and secondary panels using a ‘key-in’ concept and backfilled with impervious material. The open trench excavations were kept filled with bentonite slurry at all times, hence ensued slurry losses through permeation into the unusual composition, depth and extent of the foundations.
146
The slurry trench cutoff wall was also termed as a hanging cutoff wall, attributable to its inconclusive 15 m penetration within the overburden deposits characterized as glacial moraines. The primary objectives in regards to the cutoff wall were to evaluate its integrity within surrounding moraines, its effectiveness in seepage control and the performance of porewater pressures through different phases of impounding.
Permeability of the inplace cutoff wall duly contemplates the scope of interest but insitu measurement was multifaceted and complex. A critical appraisal of the observed behaviour using recorded piezometeric time-history patterns and head responses was thereby improvised. This approach was preferred over in-situ testing, subject to time restraints and additional benefits of establishing baseline data, determining initial conditions and monitoring flexibility through different phases of actual reservoir fluctuation.
A total of twelve (12) vibrating wire piezometers were arranged across three instrumented sections along the length of the cutoff wall. Figure 5.15 shows the location of these piezometers in a part plan and indicates their relative positions and depth zones from to the different installed tip elevations, in the corresponding section blow-ups.
Each instrumented section of the cutoff wall included 4 vibrating wire piezometers, 2 on each side, arranged to obtain observations u/s and d/s immediate of the cutoff, at two different depth zones. Instruments with installed tip elevations greater than 2630 m were categorized in depth zone 1 whereas instruments with tip elevations at or lower than 2630 m were grouped in the depth zone 2.
147
Table 5.8 tabulates average percent potential, corresponding to selected pseudo-steady reservoir level windows from the last three stages of partial impounding, for each installed piezometer across the instrumented cutoff wall sections. It also shows the average peizometeric head along with the minimum and maximum range of piezometeric heads between the pseudo-steady reservoir level windows. The table also presents piezometeric heads corresponding to the maximum reservoir levels for each stage. It is pertinent to note here that the piezometeric heads recorded against the maximum reservoir levels slightly differs from the maximum of the computed range through the pseudo-steady window. A maximum difference of
1.42 m can be seen for VWP-22, corresponding to the fifth impounding stage. This indicates variation in the foundation response at the same reservoir level through different impounding stages. The potential drops across the cutoff wall, similarly varied through the impounding stages at both the depth zones (Table 5.9). All these instruments were installed as per the execution phases adopted for the cutoff wall excavation. Readings from piezometers of section K-K covered the pre-impounding and the five subsequent stages; whereas piezometers of section I-I and H-H covered only readings from stage 2 onwards. It should be noted that out of the 3 sections, section K-K was the first one to be instrumented; thereby it covered the complete impounding cycle to-date. The other two sections were instrumented at the start of
Stage 2. The time and response data plots for upstream and downstream piezometers of each section at the specified depth zone are separately presented from Figure 5.16 to Figure 5.27. Time plots along with the head plots for all piezometers in depth zone
1 are combined and presented for a summary comparison in Figure 5.28 and Figure
5.29, whereas time and response head plots for all piezometers in depth zone 2 are separately combined in Figure 5.30 and Figure 5.31.
148
Figure 5.15 Location of cutoff wall piezometers
149
Table 5.8 Piezometeric data corresponding to cutoff wall instrumented sections
Section K‐KI‐IH‐H Position Upstream Downstream Upstream Downstream Upstream Downstream Piezometers VWP 22 VWP 23 VWP 24 VWP 25 VWP 52 VWP 53 VWP 50 VWP 51 VWP 58 VWP 59 VWP 56* VWP 57 Tip Elevations (m) 2637.00 2624.00 2633.00 2627.00 2637.00 2627.00 2637.00 2627.00 2637.00 2627.00 2640.00 2630.00
Average Piezometeric Potential 76.70 73.86 69.18 65.41 92.40 95.91 54.87 50.57 88.21 84.67 108.15* 57.23 Stage ‐ 3 (2009‐2010) Reservoir Levels (m) Piezometeric Head (m) Selected Pseudo‐Steady Average 2647.30 Average 2642.05 2641.41 2640.36 2639.51 2645.59 2646.38 2637.14 2636.17 2644.64 2643.85 2649.13 2637.67 Reservoir Level: 2647.23 Minimum 2647.12 Minimum 2640.67 2641.27 2640.18 2639.35 2644.88 2646.09 2636.84 2635.90 2644.41 2643.64 2649.12 2637.30 ± 0.15 m Maximum 2647.38 Maximum 2643.01 2641.51 2640.44 2639.63 2645.72 2646.50 2637.31 2636.31 2644.75 2643.95 2649.14 2637.91 Piezometeric Head at Maximum Reservoir Level (m) 2642.56 2641.49 2640.43 2639.60 2645.68 2646.49 2637.23 2636.26 2644.73 2643.92 #VALUE! 2637.78
Average Piezometeric Potential 94.90 81.68 69.43 68.05 94.48 94.76 52.00 47.67 87.43 84.28 100.01* 57.39 Stage ‐ 4 (2010‐2011) Reservoir Levels (m) Piezometeric Head (m) Selected Pseudo‐Steady Average 2647.30 Average 2654.78 2650.76 2647.05 2646.63 2654.65 2654.73 2641.76 2640.45 2652.51 2651.55 2656.32 2643.40 Reservoir Level: 2656.34 Minimum 2647.12 Minimum 2654.69 2650.62 2646.77 2646.51 2654.49 2654.57 2641.68 2640.36 2652.38 2651.43 2656.12 2643.24 ± 0.15 m Maximum 2647.38 Maximum 2654.89 2650.87 2647.28 2646.72 2655.09 2654.84 2641.82 2640.50 2652.62 2651.65 2656.51 2643.51 Piezometeric Head at Maximum Reservoir Level (m) 2654.95 2651.17 2647.55 2646.81 2655.20 2655.15 2642.09 2640.63 2652.93 2651.92 2655.71 2643.66
Average Piezometeric Potential 88.91 77.55 66.87 66.33 93.11 90.70 49.92 42.29 86.06 81.90 112.22* 48.34 Stage ‐ 5 (2011‐2012) Reservoir Levels (m) Piezometeric Head (m) Selected Pseudo‐Steady Average 2647.30 Average 2645.43 2642.81 2640.35 2640.23 2646.40 2645.84 2636.45 2634.69 2644.78 2643.81 2650.81 2636.08 Reservoir Level: 2647.90 Minimum 2647.12 Minimum 2643.71 2642.51 2639.98 2639.88 2646.14 2645.41 2636.21 2634.50 2644.60 2643.70 2650.64 2635.81 ± 0.15 m Maximum 2647.38 Maximum 2646.51 2642.97 2640.59 2640.47 2646.75 2646.05 2636.78 2634.83 2644.93 2644.01 2650.95 2636.29 Piezometeric Head at Maximum Reservoir Level (m) 2645.09 2642.89 2640.43 2640.26 2646.48 2646.05 2636.52 2634.78 2644.93 2644.00 2650.95 2636.20 Table 5.9 Average percent piezometeric potential drops across cutoff wall instrumented sections
Stage 3 Stage 4 Stage 5 Section Piezometer Levels Upstream Downstream (2009‐2010) (2010‐2011) (2011‐2012) Depth Zone 1 at Upper Levels (Tip El. > 2630 m) VWP 22 VWP 24 7.52 25.47 22.05 K‐K Depth Zone 2 at Lower Levels (Tip El. < 2630 m) VWP 23 VWP 25 8.45 13.63 11.22 % Potential difference between depth zones ‐0.93 11.85 10.82 Depth Zone 1 at Upper Levels (Tip El. > 2630 m) VWP 52 VWP 50 37.53 42.48 43.18 I‐I Depth Zone 2 at Lower Levels (Tip El. < 2630 m) VWP 53 VWP 51 45.35 47.09 48.41 % Potential difference between depth zones ‐7.82 ‐4.61 ‐5.23 Depth Zone 1 at Upper Levels (Tip El. > 2630 m) VWP 58 VWP 56* ‐19.94 ‐12.58 ‐26.16 H‐H Depth Zone 2 at Lower Levels (Tip El. < 2630 m) VWP 59 VWP 57 27.44 26.89 33.55 % Potential difference between depth zones ‐47.39 ‐39.47 ‐59.72 * Abnormally higher average piezometeric potentials of VWP 56 and the corresponding piezometeric levels higher than the reservoir, resulting in aberrant percent potential difference between depth zones, indicates malfunctioning of the instrument 150
Figure 5.16 Time plots for upper level piezometers installed in depth zone 1 (>2630 m) in cut-off wall instrumented section K-K
Figure 5.17 Response plots for upper level piezometers installed in depth zone 1 (>2630 m) in cut-off wall instrumented section K-K
151
Figure 5.18 Time plots for lower level piezometers installed in depth zone 2 (<2630 m) in cut-off wall instrumented section K-K
Figure 5.19 Response plots for lower level piezometers installed in depth zone 2 (<2630 m) in cut-off wall instrumented section K-K
152
Figure 5.20 Time plots for upper level piezometers installed in depth zone 1 (>2630 m) in cut-off wall instrumented section I-I
Figure 5.21 Response plots for upper level piezometers installed in depth zone 1 (>2630 m) in cut-off wall instrumented section I-I
153
Figure 5.22 Time plots for lower level piezometers installed in depth zone 2 (<2630 m) in cut-off wall instrumented section I-I
Figure 5.23 Response plots for lower level piezometers installed in depth zone 1 (<2630 m) in cut-off wall instrumented section I-I
154
Figure 5.24 Time plots for upper level piezometers installed in depth zone 1 (>2630 m) in cut-off wall instrumented section H-H
Figure 5.25 Response plots for upper level piezometers installed in depth zone 1 (>2630 m) in cut-off wall instrumented section H-H
155
Figure 5.26 Time plots for lower level piezometers installed in depth zone 2 (<2630 m) in cut-off wall instrumented section H-H
Figure 5.27 Response plots for lower level piezometers installed in depth zone 2 (<2630 m) in cut-off wall instrumented section H-H
156
Figure 5.28 Time plots for all upper level piezometers installed in depth zone 1 (>2630 m) in cut-off wall instrumented sections K-K, I-I and H-H
The lines beyond the 1:1 line actually indicate non-responsive behaviour, with the VWP tip elevation being higher than the water level.
Figure 5.29 Response plots for all upper level piezometers installed in depth zone 1 (>2630 m) in cut-off wall instrumented sections K-K, I-I and H-H
157
Figure 5.30 Time plots for all lower level piezometers installed in depth zone 2 (<2630 m) in cut-off wall instrumented sections K-K, I-I and H-H
Figure 5.31 Response plots for all lower level piezometers installed in depth zone 2 (<2630 m) in cut-off wall instrumented sections K-K, I-I and H-H
158
Initial readings from the pre-impounding stage were taken at weekly intervals; the frequency was changed to daily readings after initiating stage 1 impounding. All the installed piezometers were functional and had reacted immediately or within a few days to rises in the pond level. The comparative graphs (Figure 5.28 to Figure 5.31) show a delayed response with the reservoir for piezometers from the depth zone 1
(with installed tip elevations above 2630 m). On the other hand, piezometers in depth zone 2 (with tip elevations at or below 2630 m), responded immediately. The applied hydraulic loading activated from the lake-wards side and with a rise of the impounded water levels, additionally through the surface. The direct path through the natural foundations from the lakeside favoured a preferential flow path compared to permeation of seepage through the impervious upstream blanket on the surface.
Figure 5.15 illustrates section blow-ups indicating tip elevations for the piezometers in depth zone 1were installed at or near the upstream impervious blanket contact with the cutoff wall, whereas tip elevations for piezometers in depth zone 2 were in the natural foundations. Neither of the depth zones was in direct contact with the impounded water, rather the water had to seep through the vicinity stratum to reach the tip elevations and induce the effects in measured pore pressure readings. The piezometric response thereby was dependant not only on the water level during the impounding but also on the permeability of the medium. As such, the time acquired in the impounding stages for the water to rise above the blanket and seep through its low permeability has been depicted as the delay in the both the time and head plots for the piezometers with tip elevations in depth zone 1 (Figure 5.28 and Figure 5.29). On the other hand, the comparatively higher permeability of natural foundations and a direct contact with the impounded water from lakeside is a zero-delay phenomenon in the piezometers with tip elevations in depth zone 2 (Figure 5.30 and Figure 5.31).
159
The average drop for a pseudo-steady reservoir of 2647.23 m a.m.s.l. across the depth zone 1 corresponding to the upper levels i.e. greater than El. 2630 m of the piezometer tip elevations, during the stage 3 impounding was 7.52 % across section
K-K for VWP 22 and VWP 24, 37.53 % across section I-I for VWP 52 and VWP 50, whereas result was not determined across section H-H for VWP 58 and VWP 56, due to malfunction of downstream piezometer VWP 56. For the lower depth zone, corresponding to installed piezometer tip elevations lesser than El. 2630 m, the average potential drops observed were 8.45, 45.35 and 27.44 % respectively across section K-K (VWP 23 - VWP 25), section I-I (VWP 53 - VWP 51) and section H-H
(VWP 59 – VWP 57), for the three instrumented sections. The potential drops increased across each instrumented section, for both depth zones during the fourth impounding stage against a 9.11 m higher pseudo–steady reservoir at 2656.34 m a.m.s.l. Average observed values for instrumented sections K-K and I-I are 25.47 and
42.48 % corresponding to the upper depth zone. Average observed percent potential drop values corresponding to the lower depth zone are 13.63, 47.09 and 26.89 %, respectively for each instrumented section.
The relative increase over the consecutive impounding stages 3 and 4 in the potential drops across the cutoff wall was variable for both the depth zones in each instrumented section. The observed average was more prominent for the upper level piezometers, a maximum of 17.95% across section K-K and a minimum of 4.95 % observed across the section I-I. Conversely, for the lower level piezometers, a maximum increase in the potential drop of 5.18 % was observed across section K-K again, however a decrease of 0.55% was observed in the potentials across the lower depth zone for section H-H. This decrease in potentials across section H-H can be
160
associated with left side valley flow contributions at these levels. Neglecting the erroneous upper level potentials from section H-H, an overall increase in the associated potential drops with depth is reflected from instrumented sections K-K and
I-I showing a 12.77 and 3.21 % difference, corresponding to the pseudo-steady reservoir levels from stage 3 to stage 4. The varied potential drop at different sections is attributed to profile heterogeneity.
The highest potential drops and maximum increase with reservoir loading across the instrumented section K-K was primarily due to its location in the valley (at the toe of the steep right abutment slope), and the fact that these instruments were the first ones to be installed (having endured an extra complete cycle of impounding stage
1 prior to observations from the other instrumented sections). The associated pore water pressures in the piezometer tip vicinity at both depth zones recorded in the duration and cyclic hydraulic loading from the earlier impounding phase were significant to these observations as well.
In reference to cutoff walls it is generally known that these subsurface structures impede seepage flows by constricting the natural stream lines and redirect prospective flow paths to traverse around it, circumscribing the cutoff wall geometry
(also shown in Figure 4.5 for the blanket and upstream cutoff wall). It may be noted in the same context that the corresponding tip elevations of the upstream and downstream piezometers for both the depth zones from instrumented sections of the cutoff wall are asymmetrical. The downstream piezometer tip level for the upper depth zone in section K-K is lower not only in comparison to the upstream piezometer at the same depth zone but is also lower than similar tip elevations from the other two sections. On the contrary, the downstream piezometer tip is higher than its
161
corresponding upstream piezometer tip elevation, for the lower depth zone. The lower downstream tip allowed an early water contact with the installed instrument tip at the upper depth zone, resulting in the maximum increase in potential drops and the higher difference in potentials across the upper level piezometers from the consecutive stages. Conversely, the higher downstream tip elevation reduced both the observed potential drops and the difference across the lower depth zone from the consecutive stages. The presence of low permeable silty matrix around the installed instrument tips also significantly contributed in retarding approach and enhancing retraction of the water contact, pore pressures and the associated drops across the cutoff wall. The same tip levels across installations in section I-I and the higher downstream tip levels in section H-H, for both depth zones, correspondingly reflect comparable phenomena in the observations of their peculiar response differentials.
It was noteworthy that VWP22 and VWP56 were the only high air entry instruments, by virtue of their placement in the impervious blanket covering the cutoff wall within the depth zone 1. The upstream piezometer VWP22 was mainly delayed in response to the reservoir levels, due to its location within the blanket material. The delayed and divergent head plot for VWP22 in drawdown and its peculiar response in descent, showed pore water pressure retention within the upstream impervious stratum and restriction of dissipation due to the concrete intake walls downstream.
The downstream piezometer VWP56 responded abnormally, with reported piezometeric levels even higher than the achieved reservoir levels. This is reflective of some malfunctioning of the instrument, maybe due to its placement within the impervious blanket covering around the cutoff wall. Although observations across the
162
upper depth zone for section H-H correspond to a higher downstream tip elevation, but negative values from Table 5.9 also reflect the error in computed potentials.
The downstream piezometers associated with section K-K specially the one located within the depth zone 2, i.e. VWP25, showed a rise in response to impounding. All other downstream piezometers showed a marked decline in piezometeric heads as the construction of cutoff wall progressed except this piezometer. The low resulting differential across VWP23 and VWP25 initially gave an impression that the instruments might be choked being installed prior to the cutoff wall construction. However subsequent continuous trend with the impounding confirmed its proper functioning. Further, the peculiar behavior of VWP25 also indicated the possibility of concentrated leakage through a crack or puncture at this location, raising the downstream water levels. A steady trend in the associated readings over the impounding stages removed this possibility.
Table 5.9 also shows the observed values from stage 5, corresponding to 0.67 m of reservoir increase from stage 3. It indicates a 14.53, 5.65 and -6.22 % difference in the potentials across upper level piezometers respectively across each instrumented section. The negative value corresponds to the error considered due to malfunctioning of the VWP 56 downstream of the section H-H, which shows higher piezometeric levels than the reservoir. For the lower depth zone, a 2.77, 3.06, 6.11 % difference is shown corresponding to potentials across the instrumented sections respectively.
These observations suggest that the reason for different differential head values in some locations is more likely to be the result of variation in the permeability of the subsurface vicinity of the cutoff wall. Such permeability variations could be
163
due to the natural silty matrix or the migrated fine contents from the cutoff wall slurry during installation as a result of the water level fluctuations through the ensued impounding stages. Stage 5 observations reflect a stage of saturation being achieved, after enduring 80 % of the designed head at these locations during stage 4. The silty matrix still being in the relative effect has however redistributed itself, under the cyclic hydraulic loadings through the impounding stages, to allow development of permeation paths for water contacts to the installed piezometer tips.
The time graphs when reviewed in this context indicate a slight lag in the upstream and downstream piezometer levels at the same reservoir elevation. The piezometeric head response plots show a prominent decrease in the piezometeric heads correspondingly. The performance graphs on the whole present an improvement in the drop across the cutoff wall for both depth zones, with the successive impounding stages. The piezometeric observations hence corresponded with the presence of different permeability zones and also the valley topography across each instrumented section. The relative gaps / differentials associated with both depth zones across the cutoff were notable through time graphs for the three instrumented sections. The drop observed across the cutoff was a representation of its effectiveness; however it was variable for each section in place. The delayed response of the head plots of the downstream piezometers was also comparable in nature. It showed the relative impediment of pore pressures across the cutoff wall, hence substantiating evidence of effective seepage control. The cut-off wall, partial in nature, also did not extend laterally for a positive tie-up due to the absence of rock at the extremity of both its abutments. The surrounding strata comprised of gravel cobbles and boulders with little silty sand, and marked inferences of open work zones.
164
Moreover, the cutoff wall was tapered at its ends to reduce the differential influx through the permeable strata, causing high potential drops at the edges (Figure 5.32).
The depth of the wall at its ends was reduced ranging from 4.5 to 9 meter, leaving behind a wide gap between the abutments and the cutoff wall. Specifically towards the right abutment, the upstream water could be connected through one or more preferential flow paths provided by the available open works circumscribing below the wraparound and along the steep right abutment slopes. These preferential flow paths might also be contributing to the higher ground water potentials immediately downstream of the cutoff wall near its right side (section K-K).
Figure 5.32 Cutoff wall longitudinal profile and typical cross-section
5.4.5 Upstream Impervious Blanket and Corresponding Piezometeric Data
The piezometers considered for evaluation of the foundation response, relative to the blanket placement, subsequent to impounding stages were distributed at different locations and variable tip-elevations with respect to the valley topography, within the foundation layers underlying the upstream impervious blanket. It would not
165
be out-of-context to refer back to Figure 3.3, showing the main project features. The abutment slopes at the dam axis especially at the right abutment were very steep and prone to instability when submerged, therefore a wraparound of the embankment was extended to control seepage and provide stability. The wraparound comprised of components from upstream section of the main dam, i.e. the impervious core as the main water barrier and shell or shoulder covering for stability. On the right abutment only a small knob of rock was present (Figure 3.4), which was used to tie-up both the cores of the main dam and the wrap-around. On the left abutment where the slopes were flatter, the upstream impervious blanket was provided up to the reservoir conservation level. It can be presumed that some seepage, under influence of the upstream impounded water would traverse behind the wrap-around along natural preferential flow paths, by-passing the project domain. The valley constriction would impart typical response scenarios under these conditions.
As such, relative to probable creep lengths in the direction of flow, six piezometers are selected and discussed hereunder with reference to their position in the instrumented layout of the project (Figure 4.10). Three piezometers i.e. VWP-49,
VWP-40 and VWP-17 pertain to the foundations underlying the blanket area upstream of the main dam axis, whereas the remaining three piezometers VWP-21,
SP-11 and SP-13 are located downstream of the main dam axis.
VWP-49 presents a typical representation of the response of foundation layers directly underlying the impervious blanket in the widened valley downstream of the natural lake. VWP-40 by virtue of its orientation towards the right abutment indicates the typical foundation response under the wrap-around core slopes. VWP-17 is representative of the foundation response under the main dam core. VWP-21
166
represents conditions underlying the drainage blanket, whereas the standpipe piezometers SP-11 and SP-13 indicate the left and right valley contributions and the subsequent foundation response to the seepage domain, respectively. The time and response piezometeric data plots for these individual piezometers are separately presented from Figure 5.34 to Figure 5.39. Figure 5.40 compares time plots for all these selected piezometers in reference to the pseudo-steady reservoir level windows, relative to the maximum distance from the upstream end. The flattening of slopes of the response plots for these piezometers is cumulatively compared in Figure 5.41.
Figure 5.33 Layout of selected foundation piezometers for performance evaluation of impervious upstream blanket
167
Figure 5.34 Time and response piezometeric data plots for VWP-49
168
Figure 5.35 Time and response piezometeric data plots for VWP-40
169
Figure 5.36 Time and response piezometeric data plots for VWP-17
170
Figure 5.37 Time and response piezometeric data plots for VWP-21
171
Figure 5.38 Time and response piezometeric data plots for SP-11
172
Figure 5.39 Time and response piezometeric data plots for SP-13
173
Figure 5.40 Comparative time plots of selected foundation piezometers for performance evaluation of the upstream impervious blanket
Figure 5.41 Comparative response plots of selected foundation piezometers for performance evaluation of the upstream impervious blanket
174
The time plots from all the figures show that these piezometers visibly responded corresponding to the progressive phases of partial impounding, subject to the loading of the piezometer tip corresponding to the impounded lake level. The piezometers ceased to respond when lake levels dropped below their tip elevations.
The individual response plots for the piezometers located within foundations underlying the impervious zone i.e. VWP-49, VWP-40 and VWP-17 indicated delayed and divergent head plots, specifically in the falling limbs corresponding to the decreasing levels of each impounding phase. The rising limbs of the latest impounding stages (Stages 3, 4 and 5) presented lower peizometeric heads i.e., a lag
(from the time plots) / flattening of the slopes (from the response plots) for each piezometer, at the same reservoir levels through the impounding stages, substantiating the decrease in differentials with the impounding stages. The average observed percent potential for these piezometers are presented in Table 5.10. These potentials correspond to selected pseudo-steady reservoir level windows from the last three stages of partial impounding.
Table 5.10 also shows the average observed peizometeric head along with the minimum and maximum computed range of piezometeric heads between the pseudo- steady reservoir level windows. It also presents the actual observed piezometeric heads corresponding to the maximum reservoir levels for each stage, which slightly differ from the maximum of the computed range through the pseudo-steady window.
This variation in the piezometeric heads actually reflects the foundation response substantiating foundation redistribution relative to the same (pseudo-steady) reservoir levels in different impounding stages. The marked decrease in average potentials
(around 6%) corresponding to the comparative pseudo-steady reservoir levels of stage
175
3 and stage 5 (El. 2647.23 and El. 2647.90 m, respectively) reflects dissipation of
piezometeric potential with the progressive redistribution of the foundations. It also
indicates improving seepage control by the impervious blanket to effect.
Table 5.10 Piezometeric data corresponding to pseudo-steady reservoir levels from last three partial impounding stages for selected instruments installed in the foundation underlying upstream impervious blanket
Position Upstream of main dam axis Downstream of main dam axis Piezometers VWP 49 VWP 40 VWP 17 VWP 21 SP 11 SP 13 Tip Elevations (m) 2630.00 2624.00 2627.00 2627.00 2624.00 2624.00
Average Piezometeric Potential 47.53 54.13 36.48 17.09 20.65 5.12 Stage ‐ 3 (2009‐2010) Reservoir Levels (m) Piezometeric Head (m) Selected Pseudo‐Steady Average 2647.30 Average 2635.48 2636.97 2633.00 2628.63 2629.43 2625.93 Reservoir Level: 2647.23 Minimum 2647.12 Minimum 2635.27 2636.73 2632.77 2628.52 2629.29 2625.88 ± 0.15 m Maximum 2647.38 Maximum 2635.66 2637.10 2633.19 2628.69 2629.54 2626.01 Piezometeric Head at Maximum Reservoir Level (m) 2635.58 2637.04 2633.10 2628.68 2629.49 2625.99
Average Piezometeric Potential 45.05 57.40 36.35 11.46 15.23 3.70 Stage ‐ 4 (2010‐2011) Reservoir Levels (m) Piezometeric Head (m) Selected Pseudo‐Steady Average 2656.32 Average 2639.65 2643.40 2637.01 2629.46 2630.60 2627.10 Reservoir Level: 2656.34 Minimum 2656.20 Minimum 2639.59 2643.23 2636.92 2629.41 2630.58 2626.67 ± 0.15 m Maximum 2656.43 Maximum 2639.71 2643.53 2637.10 2629.50 2630.61 2627.38 Piezometeric Head at Maximum Reservoir Level (m) 2639.84 2643.75 2637.08 2629.58 2630.79 2627.44
Average Piezometeric Potential 41.51 52.93 30.44 7.04 15.02 3.84 Stage ‐ 5 (2011‐2012) Reservoir Levels (m) Piezometeric Head (m) Selected Pseudo‐Steady Average 2647.99 Average 2634.51 2637.14 2631.95 2626.56 2628.40 2625.83 Reservoir Level: 2647.90 Minimum 2647.79 Minimum 2634.28 2636.88 2631.71 2626.45 2628.30 2625.78 ± 0.15 m Maximum 2648.05 Maximum 2634.74 2637.27 2632.16 2626.65 2628.45 2625.86 Piezometeric Head at Maximum Reservoir Level (m) 2634.74 2637.26 2632.09 2626.59 2628.43 2625.86
The sectional alignment of VWP-49, VWP-17 and VWP-21 reflects a flow
path which creeps along the subsurface contact, directly below the placed fill for the
dam. Figure 5.42 shows the consistent decreasing potential drop along this alignment
at the pseudo-steady reservoir levels for each of the last three consecutive partial
impounding stages.
The data table also reflects a 2.48% potential drop differential at VWP-49
from the observed average potentials at Stage 4 in comparison to Stage 3, due to a
9.11 m higher pseudo–steady reservoir. The observed average potentials relative to
176
stage 3 further decreased to a 6.02% differential in stage 5, corresponding to only 0.67 m of reservoir increase, from the third impounding stage.
Figure 5.42 Comparative average potential drops at pseudo-steady reservoir levels from last three partial impounding stages along preferential flow path across main dam axis (VWP-49, VWP-17 and VWP-21)
The relative differentials (difference between the two piezometers) in the observed potential drops along the path were observed as 11.05, 8.7 and 11.07 % upstream of the main dam axis from VWP-49 to VWP-17 in the subsequently ensued impounding stages respectively. Downstream of the dam, from VWP-17 to VWP-21, the observed average potential drop differentials were 19.39, 24.89, and 23.4 % respectively through each successive impounding stage. This reflects a relative decrease in the observed potentials upstream of the main dam from stage 3 and stage
4 and then a sudden relative increase with stage 5. While towards the downstream, an
177
increase in potentials is evident through stage 3 and 4 which converts to a relative decrease with stage 5 impounding. This response of the foundation under imposed hydraulic loading is again presented through Figure 5.43, which along-side also identifies effects of a constricted valley on the subsurface seepage flows.
Figure 5.43 Comparative average potential drops at pseudo-steady reservoir levels from last three partial impounding stages for identifying constricted valley effects (VWP-40 and VWP-49 for upstream right side and SP-11 and SP-13 for downstream cross-valley)
VWP-40 and VWP-49 represent the subsurface cross valley flow possibility along the right abutment wraparound through its foundation contacts. The observed
6.6% differential from stage 3 changed to 12.35% in stage 4, and then to 11.42% in stage 5. SP-11 and SP-13 are positioned on either sides of the approximately 50 m natural valley constriction, where the cross-differentials continuously decreased from
15.53% to 11.53% and then to 11.18% for the impounding stages 3, 4 and 5
178
respectively. It may be noted that apart from the dam seepage SP-11 and SP-13 include downstream influences of cross-valley flows specifically from the left side.
It is presumed that from stage 3 to stage 4, under the 9.11 m higher hydraulic loading and the impervious blanket retarding the seepage flows, excessive gradients were imposed on the contact foundations upstream of the main dam axis.
Unsupported fines from these foundations might have migrated within the subsurface voids under effects of the imposed hydraulic gradients. With the arrival of the fifth impounding stage, the voids and fines had readjusted themselves forming an optimal permeation zone for the seepage flows to pass, hence resulting in an increased potential as the observed response. Towards downstream of the main dam axis, an initial increase in the observed differential in the potential drops from stage 3 to stage
4 signified that the drainage zones were attracting seepage flows through the dam and the underlying foundations. With the onset of stage 5 impounding, the unsupported fines from the foundations might also have migrated along with the flows within the drainage zones, hence decreasing the observed piezometeric potential response. For the downstream standpipe piezometers SP-11 and SP-13, an additional point of note is that SP-13 signifies no subsurface flows around 100 m downstream of the rock outcrop at right abutment of the dam axis. SP-11 on the other hand also conveys seepage flows from pervious natural left abutment foundations. The effect of foundation redistribution however, prevails for the overall foundations as explained.
The marked flattening of the slopes from the response plots in Figure 5.41 for the selected piezometers, through different phases of partial impounding, is related to the dissipation of piezometeric potential with the progressive redistribution of the foundations. The relative trends are similar in nature for the piezometers located
179
upstream of the dam axis, representing foundations underlying the impervious zones, and conversely for the piezometers located downstream of it, representing the drainage zones. The relative flatness indicated effective dissipation of the applied heads and conversely the associated pore pressures from upstream to downstream, in the direction of flow. These piezometer response patterns also indicate impediment of seepage flows by the impervious blanket upstream of the dam axis and the effective dissipation of pore pressures through the drainage zones downstream of the dam, overall improving seepage control trends to effect.
5.4.6 Piezometeric Potential Plots for Adopted Seepage Control Combination
Measures
A total of 27 responsive instruments were scrutinized to complement related inferences for the integrity and effectiveness of the proposed combination, i.e. the impervious upstream blanket and the partial cutoff wall. These instruments included thirteen (13) responsive vibrating wire piezometers (VWP-55, VWP-49, VWP-47,
VWP-44, VWP-40, VWP-37, VWP-35, VWP-33, VWP-29, VWP-28, VWP-21,
VWP-20 and VWP-17) installed within the foundations underlying the upstream impervious blanket, six (6) lower level piezometers out of the total twelve (12) relating to the cutoff wall (VWP-59, VWP-57, VWP-53, VWP-51, VWP-23 and
VWP-25) and eight (8) standpipe piezometers from downstream of the main dam axis
(SP-1, SP-2, SP-4, SP-6, SP-7, SP-11, SP-12 and SP-13). The distribution of these piezometers is shown in (Figure 5.44).
These piezometers were used to prepare the percent equi-potential contour plots for comparative performance evaluation. The contours are based on the
180
computed percent potentials from actual water pressure measurements made in only the selected responsive foundation piezometers within the study area, portrayed corresponding to their proper tip elevations. Construction of such potentiometric contours usually allows insight into regional patterns and directions of groundwater flow with identification of recharge and discharge areas. Figure 5.45 individually compares percent equi-potential contour plots computed using observed piezometeric elevations from the last three impounding stages, whereas all the respective contour plots are superimposed on two separate plans in Figure 5.46.
Figure 5.44 Layout of foundation piezometers for performance evaluation
181
Figure 5.45 Individual percent potential contour plot plans for the last three partial impounding stages, corresponding to (a) the pseudo-steady reservoir levels i.e. 2647.23, 2656.34, and 2647.90 m.a.m.s.l. for stages 3, 4 and 5 respectively and (b) the same reservoir level i.e. 2644.78 m.a.m.s.l.
182
Figure 5.46 Comparative percent potential contour plot plans for the last three partial impounding stages corresponding to (a) the pseudo-steady reservoir levels i.e. 2647.23, 2656.34, and 2647.90 m.a.m.s.l. for stages 3, 4 and 5 respectively and (b) the same reservoir level i.e. 2644.78 m.a.m.s.l.
183
Figure 5.45 (a) shows average potential plots individually corresponding to the observed piezometeric elevations achieved at pseudo-steady reservoir levels i.e. El.
2647 m (8685 ft) for stage 3, 2656 m (8715 ft) a.m.s.l. for stage 4 and El. 2648 m
(8688 ft) for stage 5 of the partial impounding, respectively. These plots clearly depict the phenomena of head distributions through prominent lateral shifts and the relative density concentrations of the potential contours. A major drop of more than 45 percent was concentrated across the cutoff wall and around 15% along the core boundary for the main dam and the wraparound. This drop reduced longitudinally towards the end of the impervious zone (blanket and main dam core) leaving a mere
20 percent potential to be dissipated through the drainage zones. The change in potentials is also identified to increase below the impervious zones like from 25% in the third stage to 35% in the fourth stage, along the right abutment and conversely decrease below the drainage zones i.e. from 10% in the third stage to only 5% in the fourth stage. The potential contours from the fifth stage, with an approximately similar pseudo-steady reservoir level to the third stage, marked a shift in the overall presentation of the contour plot. These percent equi-potential contour plots tend to laterally shift upstream to the left and constrict along the impervious barriers
(prominent along the cutoff wall, the main dam core and the wrap-around core).
Bypassing the core and entering the drainage zones, these contour plots tend to open up and shift laterally downstream to the right towards the toe drain.
A better insight is presented through the depiction of potential contours corresponding to a consistent reservoir level of 2644.78 m (8677.1 ft) a.m.s.l. selected from the rising limbs of each of the last three impounding stages in Figure 5.45 (b).
The constriction along impervious barriers indicates effectiveness of these barriers in
184
retarding the head distribution along the flow paths and inducing potential drops. The lateral upstream shift towards the left (notable in the plan corresponding to the 35%, the 40% and the 45% potentials) is directed along orientation of the blanket extents.
The fifth stage specifically amplifies this effect by a pronounced shift and flattening of the contours. The widening along the drainage zones on the other hand, is representative of functioning pore pressure dissipation. The downstream shift towards the right (notable in the plan corresponding to the 15% potentials) is directed along orientation of the drainage blanket extents headed for the toe drain. Localized closed contours (presented as concentric rings) are shown downstream of the cutoff wall, towards the right abutment and immediately downstream of the main dam core. Under typical conditions these concentric rings indicate three distinct sinks towards which groundwater flows are attracted from all directions and passes through to deeper zones. Typically these sinks are less distinct in the fifth impounding stage.
The effect of raising the reservoir levels through the consecutive stages of impounding appears to have resulted in an overall change in these concentric potential rings along with a lateral shift towards the left. The lateral shifting is reflective from the 40% contour downstream of the cutoff wall and the 20% contour at the right abutment, in stages 3 and 4. The fifth impounding stage however shows the eventual phasing-out and opening –up of these concentric contoured rings, due to saturation of the medium and sealing of sinks due to fines migration. The relative shifting contours in plan for the subsequent recurring stages presents foundation redistributions, corresponding to the relative percent potential variations. The concept of unsupported fines migration and their readjustment under a presumed saturation phase with repetitive impounding cycles also highlights improvement and natural healing of the
185
foundations in response to the excessive water head loadings experienced through the consecutive impounding from stages 3 through 5.
The increase in potentials below the impervious zones and decrease below the drainage zones, from the potential contours corresponding to the pseudo-steady reservoir levels of each stage, indicated a restraining effect in the total head distribution due to the impervious blanket and dissipation of the build-up potentials at the drainage zones. However, the comparable existence of local concentric contours as observed, marked a conflict in the conventional flow direction based on the constricted valley and presumed downstream oriented seepage flows.
The Darcy’s Law generally signifies a consistent decreasing head in the direction of flow. The origin of these aberrant potentials in sequence to the piezometeric levels indicated that the flow paths were not as linear as expected. The origin of these local incidents may be contributed to typical representations of high permeability pockets within the considered depth zone, which might be connected downstream at deeper zones, forming internal sink points. However, some influence of the foundation redistribution was also presented from a visible shift of these aberrant potentials with a decrease in the observed values for the latter impounding stage. The observed phenomenon and orientation of these sink-like sources corresponding to the same reservoir levels indicated in Figure 5.45 (b) and Figure
5.46 (b) is evidence of foundation redistribution / adjustments taking place subsequently under the imposed hydraulic loadings, even at the same reservoir levels, through the different consecutive partial impounding stages. The lateral shifts of these concentric contours indicated subsequent improvement in the foundation response due to material redistributions corresponding to the applied heads experienced by the
186
foundations. It is presumed that such a condition shall continue until an ideal steady state condition is reached, marked with consistent foundation responses being registered for the same reservoir level and eventual phasing out of the concentric percent potential rings indicating saturated foundation zones.
Local variations associated with the sink-like sources of closed potential contour plans (Figure 5.45 and Figure 5.46) were critically scrutinized in cross valley profiles along the flow direction (Figure 5.47). These profiles were aligned across sections distributed to cover the whole project area from the immediate vicinity of the cutoff wall to the seepage exit point at the toe drain. Corresponding to all three impounding stages, the differential across the cutoff wall, within the initial 50 m from the reservoir line, were maintained around 25%, 45% and 5% respectively along the left, center and right of the valley. The range of observed potentials flattened within the upstream blanket foundations from 150 to 375 m nearing the dam axis, and remained within a limit of 15% corresponding to the left, center and right of the valley. The differentials steepened around the core/filter transition and were observed to drop a further 20-25% within the drainage zones, some 450 m from the upstream.
The observed differentials near the dam toe, 565 m from the upstream, rose to around
10% indicating accumulation of valley flows.
The consecutive potentials along the piezometer sequence of each cross-valley section also represented the longitudinal distribution along three sections, in-line of the cutoff wall instrumentations. The overall potentials gradually dropped with the instruments from upstream to downstream, i.e. in the direction of flow, punctuated with local variations. The observed differentials with succession of each section steepened at the upstream end and then at the downstream end. However, the relative
187
potential drops varied across the valley with the maximum drop indicated at the center across the cutoff wall and then at the left abutment upstream of the dam axis with a reversal observed travelling downstream of the dam axis, shifting to the right.
It can be perceived in comparison with the sensitivity analyses discussed earlier that a total of 30% of the total hydraulic potential corresponding to the maximum conservation level can theoretically be reduced within the embankment dam core when provided with no other seepage reduction measure. As such, the remaining 70% would load the drainage system.
Instrumented observations from Satpara dam post construction measurements under different reservoir levels have indicated that the drainage system dissipates around 20% of the total potential, 80% of it is progressively reduced due to the adopted seepage control scheme. Out of the 80%, around 40-45% of the potential is dissipated just across the upstream cutoff wall while the remaining 35-40% is reduced along the upstream impervious blanket. This effect of the potential distributions indicated that a horizontal upstream impervious blanket, with an additional upstream partial cutoff wall to resolve space constraints for the blanket placement is the most effective seepage control measure though permeable dam foundations in the absence of bedrock or an impervious stratum.
The effects downstream of the cutoff wall were mainly due to the cross-valley contributions. These cross-valley flows were initially enhanced from seepage under the right abutment wrap-around, but shifted to the left valley slopes with the onset of the rock outcrop at the right abutment, limiting subsurface flows from the right side.
188
(a) Pseudo-Steady Reservoir Levels
(b) Same Reservoir Level
Note: Location of instruments is shown in Figure 5.44
Figure 5.47 Graphical percent potential plots for instruments installed along cross valley profiles corresponding to the last three partial impounding stages for (a) the respective pseudo-steady reservoir levels i.e. 2647.23 (stage 3), 2656.34 (stage 4), and 2647.90 (stage 5) m.a.m.s.l. and (b) the same reservoir level i.e. 2644.78 m.a.m.s.l.
189
The differentials across the cutoff wall, similar to the inferences from the earlier sections, were very prominent through these cross-valley sections. The upstream cutoff wall had consequently imparted favorable circumstances downstream by impeding the higher potentials right at the upstream end. The differentials and trends of the observed piezometeric potentials prove that the upstream cutoff wall was effective in preventing subsequent stages of piping i.e., blocking the path of continuation, progression and formation of pipes. An alternate to the upstream cutoff wall position might not have the successive effect of potential reduction or flattening nearing the core/filter contact, after the impounding cycles. Conversely, it would also have increased the gradients near the filter zones as investigated earlier through sensitivity analyses.
The relative differentials towards the drainage zones increased with the passage of time and a raise in the reservoir head with subsequent impounding stages.
The visible relative differential in potential drops associated with the two end sections reflects effective dissipation of accumulated potentials in these foundations.
The relative potential distributions corresponding to the impounding stages directly contribute to the efficiency of the seepage control combination adopted to safeguard the foundations against initiation of piping. Figure 5.47 also introduced evidences of transverse seepage flows through the left and right flanks of the valley, with dominant flows from the right side, indicated from higher values of the average potentials (VWP25, VWP33 and SP2). The cross valley flows within these foundations were emphasized through underground springs which originated and traveled along explicit natural pathways of the complex material compositions of moraines.
190
5.4.7 Inferences from Performance Evaluation
The general coverage and distribution of the responsive piezometers and the data collected through these instruments by the end of the considered period included sufficient data throughout the impounding stages, to help analyze the integrity and effectiveness of the proposed combination, i.e. the impervious upstream blanket and the partial cutoff wall. It was also considered adequate in evaluating the effects of impounding on the general foundation strata and pore water pressures in relation to the initiation phenomena for internal erosion or piping.
Head differentials and potential drops associated with the piezometers covering the project area helped realize a 3-D flow domain along with the diversity in the nature, direction and rate of flow through the foundations. Such 3-D flows are expected also with the valley constrictions, wider at the upstream and narrowed at the dam axis towards downstream, best represented with a 3-D modeling approach in preference to a 2-D seepage flow analysis.
The effectiveness of the seepage control measures was evident with the indication of improvement in the potential contours over the ensued impounding stages. The indication of local aberrant potentials although were useful as evidences of cross-flows across the valley, their origin could also be contributed to the inherent microbiological, chemical or mechanical mechanisms of foundation clogging dependant on the direction as well as the rate of flow through the foundations. The flow paths followed curved paths within this domain, through the high permeability
(weak) zones in the foundation, due to least resistance. This movement of water
191
through perceptible pervious interstices of the foundation stratum induced associated movements in soil particles.
The resultant particle migration was generally preferential as necessary foundation redistribution under the adopted seepage control combination and the progressive hydraulic pressure exerted by the increasing reservoir level depths through the ensued impounding stages. The major potential differentials across the cutoff wall also indicate that this phenomenon was continuous and initiated in this vicinity through the earliest impounding stages where the reservoir level could not exceed even 50% of the design head.
The foundations surrounding the cutoff wall were disturbed during its installation process. All panel excavations for the slurry trench were ensured to proceed while being filled with bentonite slurry, necessary in view of avoiding frequent cave-ins during the excavations.. This activity induced percolation of slurry, which accumulated within voids and cavities of the open works within the foundations. On impounding, finer particles from the impervious barriers (both the cutoff and the overlying blanket material) further leached in with the ingress of seepage pressures. This phenomenon resulted in possible formation of particle bridges trapping pore water and increasing piezometeric potential even after the reservoir level receded.
With the passage of time and repetitive filling and drawdown through the following phases of impounding, subsequent foundation redistribution was presumed to have taken place. The finer particles adjusted themselves around the cutoff wall
192
foundations. The pore pressure retention subsided and trend of aberrant potentials computed appeared to have decreased over the impounding stages.
The foundation redistribution due to reservoir level fluctuations was not only restricted to the vicinity foundations of the cutoff wall. It initiated due to the upstream impermeable barrier and the effects were transferred further downstream along the paths of least resistance through adjacent permeable K-zones.
The rearrangement of peculiar foundation constituents under enhanced reservoir loadings of the successive impounding stages assumed relative composure.
Repetition of the same reservoir levels resulted in a consistent potential distribution with visible flattening out as in the average potential plots from Figure 5.45 (a) and
Figure 5.46 (a). At some places in the fifth impounding stage virtual elimination of these unusual potential sources was also noted in comparative plots for the same reservoir level in the partial impounding stages indicated in Figure 5.45 (b) and Figure
5.46 (b). This foundation readjustment actually formed basis of an uncertainty propagation analysis as an imminent explanation for these observations. The analysis was framed to test likely-hood of a change in the assumed K-zone permeability associated to the adopted seepage control measures and aggravated under the applied hydraulic loading.
5.5 UNCERTAINTY PROPAGATION ANALYSES
5.5.1 Basic Hypothesis
The basic hypothesis for the uncertainty propagation analyses involved an assumption that these applied heads had likely caused migration of unsupported fines
193
from the silty matrices in the high K-zones (comprising cobbles and boulders) to low
K-zones (comprising sandy gravels). The basic inputs for the modeled foundation profiles pertained to pre-construction stages of the project. The original foundation profiles developed from the point sampled K-data using Rockworks were utilized for the initial trials.
The assessment of post-construction foundation representativeness was fundamental in establishing a conceptual foundation approximation with a satisfactory agreement between the predictions and observations corresponding to the installed seepage control measures. This analysis was also required to quantify the extents and qualify the K-zones (design stage / pre-construction) used to induce changes because of foundation redistributions (post construction).
The potential contour concentrations from Figure 5.46 and the potential drop trends from Figure 5.47 provided preliminary inferences on the perceptive locations for exercising a modelling approach to certify the uncertainty propagation hypothesis.
The highest potential drop was associated along the cutoff wall profile. Ensuing construction / installation of the cutoff wall, slurry loss into the foundations occurred in variable extents along the cutoff profile. This implied an existing erratic heterogeneity of the subsurface, accommodating different amounts of the slurry loss within the cutoff wall vicinity. The highly pervious drainage blanket followed an impervious core to collect seepage and safely convey it downstream. Figure 5.46 identifies a steep concentration of the potentials upstream of the drainage zone associated with higher gradients for attracting flows into the drainage blanket.
194
These concentrated potentials or higher gradients are susceptible locations for subsequently dislodging fines resulting in foundation redistributions. The relative shifting of potential contours in comparable concentrations (referring to Figure 5.46, in vicinity of the cutoff wall, upstream from the impervious core and downstream in the drainage blanket) in the successive impounding stages is noteworthy in context.
As such, the post-construction local modifications to the modelled K-zones were evocable in the cutoff wall proximity and the foundations underlying the drainage blanket. The permeability magnitudes of the original K-zones modelled through RockWorks, using the in-situ investigatory permeability data set, play a significant role in the local K-zone adjustments from the post construction scenario.
The increase or decrease in the original K-zones mainly depends on engineering appraisal of the extent of fines propagation from different adjacent K-zones.
Following the same theoretical hypothesis of foundation redistribution, relative permeability magnitudes for post construction local K-zone adjustments through uncertainty propagation analysis along both sections would vary. Considering heterogeneity of foundations in the problem domain, establishing foundation redistribution hypothesis also required a relative 2-D versus a 3-D treatment.
The permeability assignment plans for each individual slice used for the actual
3-D FEFLOW model, presented individually from Figure 5.48 to Figure 5.53, show the complex layered distribution of permeability zones in 3-D. The same permeability assignment plans for all the slices are summarized for comparison in Figure 5.54.
The extent of K-zone changes to invoke limited the 3-D analyses under realization of multidimensional intricacies. A 3-D model domain projected inherent
195
difficulties not only in the application of the boundary conditions but also in the related parameter estimation and assignments. Sectional analyses as of the 2-D
SEEP/W modelling were rather manageable, compared to the 3-D FEFLOW modelling approach, under these presumptions. Testing of the uncertainty propagation hypothesis therefore utilised the sectional SEEP/W models. The comparative 2-D analyses involved two sections with different original rockworks foundation profiles, due to their separate orientations in plan.
5.5.2 Conditions for Comparative Analysis
A critical assessment of the model outcomes for the selected seepage control combination was possible under transient analyses for the exact replication of the impounding status. The incurring complications of the impounding scenarios with respect to the construction sequence limited this approach over selecting a steady state analysis for relative simplicity. A perfect steady state was clearly not achieved in view of the observed variations in the piezometeric potentials for same reservoir levels as discussed earlier. It was most likely, on the other hand for an existence of a ‘pseudo- steady’ state under consistent pond retention over some appreciable period of time, during either of the impounding phases. A plausible consecutive recurrence level of
+/- 0.15 m (0.5 ft) was selected for the said pseudo-steady state reservoir level. Three such levels were identified; one each for the last three partial impounding stages. The third pseudo-steady reservoir level of around 2648 m (8687 ft) a.m.s.l. from the fifth partial impounding stage was neglected in view of the minor difference of only around 1 m between the stage three and stage five pseudo-steady reservoir levels and its on-going operation cycle.
196
Figure 5.48 Permeability Assignment Plan for Slice 1
Figure 5.49 Permeability Assignment Plan for Slice 2
197
Figure 5.50 Permeability Assignment Plan for Slice 3
Figure 5.51 Permeability Assignment Plan for Slice 4
198
Figure 5.52 Permeability Assignment Plan for Slice 5
Figure 5.53 Permeability Assignment Plan for Slice 6
199
Slice1 Slice2 Slice3
Slice4 Slice5 Slice6
Figure 5.54 Permeability assignment plans for all slices (top left – slice 1 to bottom right – slice 6) from 3D FEFLOW model.
The Pseudo-steady reservoir level windows from the stages 3 and 4, with completed reservoir operation cycles, were selected for the comparative analysis.
These reservoir levels were used alternately to substantiate comparative evaluation of model results with the observed values. The accumulated reservoir during the partial impounding stages applied hydraulic heads never before experienced by the natural foundations. Veracity of modeling a representative seepage regime for the project required a 3-D simulation of the project head distribution under evidences of valley flows ascertaining a 3-D flow domain. This requirement was considered to be more relevant to the intricacies of a constricted valley flow as compared to the sectional
200
simulation (2-D) of the problem, which could not incorporate the overall lateral contributions of the foundation stratum with exception of a constant unit width. The resulting hydraulic head distributions in plan, from FEFLOW snapshot views of 3-D modeling, for the pseudo-steady reservoir level of Stage 4, i.e. El. 2656 m.a.m.s.l. are presented individually for each slice from Figure 5.55 to Figure 5.60 and for the pseudo-steady reservoir level of Stage 3, i.e. El. 2647 m.a.m.s.l. from Figure 5.61 to
Figure 5.66. The corresponding outcomes for both the pseudo-steady reservoir levels are summarized for comparison in Figure 5.67.
Observed trends from the resultant head contour plan plots of a 3-D seepage analysis for both stages indicate a continuously decreasing head distribution in the direction of flow through all model slices. An overall total head drop of 20 m (i.e.
2660-2645 corresponding to the Stage 4 Pseudo-Steady Reservoir Level and 2650-
2635 corresponding to the Stage 3 Pseudo-Steady Reservoir Level), was observed in the foundation layers upstream of the main dam axis.
The 3-D model however, showed effects of valley constrictions on the seepage flow domain, specifically emphasized over the downstream of the main dam axis where the contour concentration significantly increased and the total head contours were shown to drop. The same effect was persistent throughout the modeled foundation layers.
A specific difference with the increase in the modeled pseudo-steady state reservoir levels through the staged impounding phases was the distributive effect of the foundation permeability (Figure 5.54) on the total head contours. The higher permeability zones (10-3 m/s) for the foundation layers (between slice 2 to slice 6)
201
concentrated towards the left abutment influenced distribution of the total head contours with an increased domain towards the left at higher reservoir levels. These contours were restricted towards the right side due to presence of the rock outcrop, modeled with the lowest permeability assigned as 10-7 m/s, increasing from right to left in the lateral direction, with depth (slice 3 to slice 6).
Concentric contours as observed corresponding to the potential contour plot plans (Figure 5.46) reflecting the actual foundation response, were absent in the modeled 3-D scenario. The existence of these local concentric contours had marked a conflict in the flow direction, indicating non-linear flow paths and emphasizing the requirement of a 3-D modelling approach. In general, all 3-D models typically follow the Darcy’s Law, which signifies a consistent decreasing head in the direction of flow.
Permeability differences en-route the flow paths simply induce a change in the hydraulic grade line. Logically, specific local higher permeability pockets at location of these observed contours still could not replicate formation of the internal sink point sources. The formation of such contours represented some quantum of seepage flows leaving the modelled flow domain at lower levels. Assessment of exact amounts for these sinks was an unfortunate probability in regards of the modelling efforts.
Advancement in computational soil mechanics has equipped commercially available groundwater software, self-contained with intricacies governing specific modeling aspects like the phreatic simulation on the downstream drainage zones, without the specification of a specific flux or alternately a head boundary condition.
However, observations in this research have indicated that 3-D models should also include specific scenarios of permeable foundations for dams regarding model geometry and/or boundary conditions. Such improvements might include assigning
202
impervious properties of a cutoff wall to elements and nodes along specified alignments to user defined depths, or incorporation of local foundation sinks without specification of flux or head boundary conditions through open-work zones. Similarly flexibility for simulation result formats of essential items like seepage flow, gradients, hydraulic heads or potentials in the form of vectors, contours or labeled graphical plots should also be conditioned in the source codes / user interfaces.
Capability of 3-D models for transient hydraulic loading is quite commonly available in most commercially available groundwater software. Hydraulic loading is expected (as observed and discussed earlier) to cause foundation redistribution, which tends to subside over time. The extent, magnitude and direction of this change are all dependent on natural processes governed by the laws of flow through media and also influenced by lateral effects of the flow domain. It is further recommended to develop an actual transient foundation response source code for the dam foundations to be applied alongside of the transient runs for the reservoir levels through the impounding phases in this regard.
The lack of approximating active scenarios representative of the observed internal sinks through material parameters or the boundary conditions limited further application of the 3-D model as the ideal representative of the site conditions.
However, in view of the relative significance of a 3-D flow domain maintaining effects of the lateral flows, the comparative nature of the FEFLOW 3-D model was utilized as an appraisal to cross-verify adequacy of pre-design investigations with respect to the 2-D SEEP/W model results, when compared with the observed values.
203
Figure 5.55 Head Distributions Plots for Slice 1 at Stage 4 Pseudo-Steady Reservoir Level i.e. El. 2656 m.s.m.s.l.
Figure 5.56 Head Distributions Plots for Slice 2 at Stage 4 Pseudo-Steady Reservoir Level i.e. El. 2656 m.s.m.s.l.
204
Figure 5.57 Head Distributions Plots for Slice 3 at Stage 4 Pseudo-Steady Reservoir Level i.e. El. 2656 m.s.m.s.l.
Figure 5.58 Head Distributions Plots for Slice 4 at Stage 4 Pseudo-Steady Reservoir Level i.e. El. 2656 m.s.m.s.l.
205
Figure 5.59 Head Distributions Plots for Slice 5 at Stage 4 Pseudo-Steady Reservoir Level i.e. El. 2656 m.s.m.s.l.
Figure 5.60 Head Distributions Plots for Slice 6 at Stage 4 Pseudo-Steady Reservoir Level i.e. El. 2656 m.s.m.s.l.
206
Figure 5.61 Head Distributions Plots for Slice 1 at Stage 3 Pseudo-Steady Reservoir Level i.e. El. 2647 m.s.m.s.l.
Figure 5.62 Head Distributions Plots for Slice 2 at Stage 3 Pseudo-Steady Reservoir Level i.e. El. 2647 m.s.m.s.l.
207
Figure 5.63 Head Distributions Plots for Slice 3 at Stage 3 Pseudo-Steady Reservoir Level i.e. El. 2647 m.s.m.s.l.
Figure 5.64 Head Distributions Plots for Slice 4 at Stage 3 Pseudo-Steady Reservoir Level i.e. El. 2647 m.s.m.s.l.
208
Figure 5.65 Head Distributions Plots for Slice 5 at Stage 3 Pseudo-Steady Reservoir Level i.e. El. 2647 m.s.m.s.l.
Figure 5.66 Head Distributions Plots for Slice 6 at Stage 3 Pseudo-Steady Reservoir Level i.e. El. 2647 m.s.m.s.l.
209
Slice1 Slice2 Slice3
Slice4 Slice5 Slice6
(a) Stage 4 Pseudo-Steady Reservoir Level 2656 m.a.m.s.l.
Slice1 Slice2 Slice3
Slice4 Slice5 Slice6
(b) Stage 3 Pseudo-Steady Reservoir Level 2647 m.a.m.s.l.
Figure 5.67 Head Distribution plan plots for all slices (top left – slice 1 to bottom
right – slice 6) from 3D FEFLOW model.
210
5.5.3 Comparative modelled results
The Figure 5.68 shows the observed and simulated total head results from finite element models of different approximations regarding the dam foundations for the instrumented section X-X. Figure 5.69 shows the same foundation redistribution hypothesis tested on another instrumented section K-K. The results of the foundation models developed using the pre-construction exploratory in-situ test results through
Rockworks from either 2-D or the 3-D approach deviated considerably from the actual physical observations. The former generally presented underestimated resultant trends compared to latter overestimated predictions (the only exception being the trends from Section K-K under Stage 3 pseudo-steady reservoir levels). The best correspondence of simulated results to the observed values was achieved from the
RockWorks modeled foundation profile with post-construction adjusted K-zones.
The results of the foundation model developed using the pre-construction exploratory in-situ test results through Rockworks deviated from the actual physical observations. The attempt to improve prediction of the foundation response through inverse modelling of the post-construction scenario by adjusting the K-zones led to an improvement of the fit. Table 5.11 accumulates the overall model evaluation statistics for the resultant total heads based on the different pseudo-steady state reservoir levels, for both investigated sections. The best acceptable values from of all the three evaluation indices concluded that the observed and simulated total heads indicated a good relative correspondence for the post-construction adjusted K-zones scenario of foundation approximation.
211
Figure 5.70 shows actual K-zones from original Rockworks foundation approximations and adjusted K-zones for the uncertainty propagation hypothesis from two dimensional sectional analyses, along Section X-X (Figure 5.70a) and along
Section K-K (Figure 5.70b). The effects of permeability assumptions on the 2-D modeled results are further elaborated in Figure 5.71. The distribution of the equipotential total head contours tend to drop consistently within the homogeneous foundation profile. The distributions concentrate within the impervious zones with the rock works modeled profiles.
Table 5.11 Model evaluation statistics for resultant total heads
RockWorks modeled Post- Acceptance Foundation foundation profile from construction Criteria Approximations exploratory investigations adjusted K-zones Moriasi et : 2D Results 3D Results 2D Results al. (2007) Section: X-X Reservoir Level: Pseudo-Steady Stage 4: 2656 m (8715 ft) a.m.s.l. RSR 0.731 1.142 0.283 <0.5 NSE 0.358 -0.564 0.904 0 - 1 R2 0.710 0.677 0.934 >0.5 Reservoir Level: Pseudo-Steady Stage 3: 2647 m (8685 ft) a.m.s.l. RSR 0.673 0.663 0.353 <0.5 NSE 0.456 0.472 0.850 0 - 1 R2 0.742 0.708 0.869 >0.5 Section: K-K Reservoir Level: Pseudo-Steady Stage 4: 2656 m (8715 ft) a.m.s.l. RSR 0.993 1.478 0.419 <0.5 NSE -0.127 -1.498 0.800 0 - 1 R2 0.670 0.859 0.850 >0.5 Reservoir Level: Pseudo-Steady Stage 3: 2647 m (8685 ft) a.m.s.l. RSR 1.826 0.905 0.893 <0.5 NSE -2.810 0.064 0.089 0 - 1 R2 0.332 0.691 0.551 >0.5
212
(a)
(b)
Figure 5.68 Comparative graphical plots at Section X-X for head distribution at pseudo-steady state impounding reservoir levels from (a) Impounding Stage 4 and (b) Impounding Stage 3
213
(a)
(b)
Figure 5.69 Comparative graphical plots at Section K-K for head distribution at pseudo-steady state impounding reservoir levels from (a) Impounding Stage 4 and (b) Impounding Stage 3
214
(a)
(b) Figure 5.70 Foundation profiles for uncertainty propagation analyses (a) along Section X-X and (b) along Section K-K
215
(a)
(b)
(c) Figure 5.71 2-D (SEEP/W) Seepage analyses output showing equipotential total head contours and flow paths for Section X-X and Section K-K (a) Homogeneous foundation profile (b) Rockworks foundation profile with actual K-zones (c) Rockworks foundation profile with adjusted K-zones
216
The relative simplicity of managing the post construction scenarios of foundation redistribution through adjusting the K-zones in a 2-D model compared with the 3-D model is worth noting. The head drop distributions were easily correlated with the observed values, through a relative local adjustment of the sectional K-zones, facilitated through absence of a third dimension. However, in spite of local adjustments for model correlation, model results could not match reversals in observed total head variations from P20 to P21 in section X-X nor from P44 to P40,
P35 to P33 and P28 to P29, in Section K-K. These indicated similarities recalled from closed percent potential contour trends of Figure 5.46.
5.5.4 “Uncertainty Propagation” Inferences
The assessment of post-construction foundation representativeness was fundamental in establishing a conceptual foundation approximation with a satisfactory agreement between the predictions and observations corresponding to the installed seepage control measures. The observed inconsistencies in the head distributions re- emphasize the comparative dependency of flows on the medium composition, as also indicated through results of sensitivity analyses under different foundation approximations. It also corresponds to actual sub-surface flow domain through foundations being 3-dimensional, dependent on the induced hydraulic gradients under the applied hydraulic heads and following highest permeability zones (not necessarily in a linear flow path) within the complex heterogeneity of the foundations of Satpara composed of regions characterized by different and contrasting hydraulic parameters.
The epistemic nature corresponding to uncertainties of the actual K-data set from the investigation stages was confirmed. Each sectional trial required a similar
217
modification approach relative to its original assumed foundation profile and orientation in plan. Similarity in results from both sectional trials over different reservoir loadings also helped establish the theoretical hypothesis of post-construction foundation redistribution.
Post construction scenarios induced by seepage control measures were also substantiated through these results. Such scenarios result in buildup of foundation scenarios difficult to contemplate at a pre-design stage and may not be foreseen at investigation / design stages. Reliable foundation seepage control in extremely erratic soils, like the foundation moraines at Satpara, would essentially require relative comparison of conventional estimates of simplified seepage analyses with the best estimates of hydraulic conductivity and anisotropy of the approximated foundation profile, from the most thorough possible investigations, in a detailed study.
218
CHAPTER 6.
CONCLUSIONS AND RECOMMENDATIONS
6.1 GENERAL
The design of embankment dams has evolved through time from empirical rules based on observed performances to a framework of consistent and logical analysis of seepage. Formidable problems involved in considerations for seepage under dams include heterogeneity of foundations and acceptability of comparative treatment / control features. The performance evaluation of foundation seepage control of Satpara Dam Project is presented in this research.
6.2 CONCLUSIONS
1. A multi-zoned approach is more reliable for representation of heterogeneity in
permeable foundations having permeability over several orders of magnitude, than
conventional layered arrangements.
2. Multivariate seepage sensitivity with selected seepage control measures (upstream
blanket with partial cutoff wall) confirms that both control measures complement
each other for an overall reduction of the total hydraulic potential (72%) and
hydraulic gradient (73.5%) at toe of the main dam core.
3. Instrumented observations at Satpara Dam prove that seepage in a constricted
valley topography with complex geological depositional environment is three
dimensional with distinct cross-valley contributions, and the adopted seepage
control measures are effective in such flow domain.
219
4. Performance evaluation at Satpara Dam showed that the complex geological
subsurface exhibited foundation redistributions which diminished with repetitive
impounding.
5. Uncertainty propagation analysis through inverse 2-D seepage modeling of the
post-construction scenario proves the phenomena of foundation redistribution.
6.3 RECOMMENDATIONS
1. Penetration depth of a partial cutoff wall should be optimized through assessment
of functioning gradient and head distributions with due considerations to
foundation permeability within its installation vicinity.
2. Transient 3-D modelling in conjunction with 2-D seepage modelling should be
done in corroboration of the uncertainty propagation hypothesis for K-zone
proportioning to improve perception of the subsurface response to externally
imposed hydraulic loading.
3. The 3-D flow modelling software (FEFLOW) is oriented for regional groundwater
flow modelling. Dam foundation seepage modelling under complex geological
conditions necessitates tailored flexibility of its application in terms of assignment
and modifications of domain parameters, boundary conditions (pre-processing)
and nodal results generation (post-processing).
4. Further research is required into methods and means of understanding and
describing particular concerns regarding flow conditions and seepage forces
through pervious strata of complex material composition from inherently
diversified depositional environments (covering wide spread origins including
alluvial, fluvial and glacial), and faced with the propensity of initiation in
relevance to localised aberrant potentials.
220
REFERENCES Aalto, J. (1984) Finite element seepage flow nets. International Journal for Numerical and Analytical Methods in Geomechanics, 8, pp 297 – 303. Aberg, B. (1993) Washout of grains from filtered sand and gravel materials. ASCE Journal of Geotechnical Engineering 119(1):36–53. Adel, D.H., Bakker, K.J. and Breteler, M.K. (1988) Internal stability of minestone. Proceedings of International Symposium of Modelling Soil–Water-Structure Interaction, pp 225–231, Balkema, Rotterdam. Agha, A. (1975) Seepage Control at Tarbela Dam through Upstream Blanket. Symposium on Tarbela Dam Problems and Solutions, Pakistan Engineering Congress, Lahore, P-117, Vol. 15, pp. 56-81. Agha, A. and Ahmed, M. (1990) Experience of Geotechnical Engineering on Dams in Pakistan. Proceedings of the First International Seminar on Soil Mechanics and Foundation Engineering of Iran, Vol. 2. Akin, J.E. and Grey, W.H. (1977) Contouring on isoparametric surfaces. International Journal for Numerical Methods in Engineering,11, pp 1893 – 1897. Arulanandan, K. and Perry, E.B. (1983) Erosion in relation to filter design criteria in earth dams. ASCE Journal of Geotechnical Engineering, 109(5):682–697 ASCE. (1993). Criteria for evaluation of watershed models. Journal of Irrigation and Drainage Engineering. 119(3): 429-442. Asianics Agro-Dev. International (Pvt) Ltd. (2000) Tarbela Dam and related aspects of the Indus River Basin, Pakistan, A WCD case study prepared as an input to the World Commission on Dams, Cape Town, www.dams.org ASTM Standard D4630-96, (1997) Standard Test Method for Determining Transmissivity and Storage Coefficient of Low-Permeability Rocks by In Situ Measurements Using the Constant Head Injection Test, ASTM International, West Conshohocken, PA, 1997, DOI: 10.1520/D4630-96, www.astm.org. Bennett, P.T. (1946) The Effect of Blankets on Seepage through Pervious Foundations. Transactions, ASCE, Vol. 3, Paper No. 2270, 1946, p. 215. Bligh, W.G. (1910) Dams, barrages and weirs on porous foundations. Engineering News 64(26):708–710 Brown, F. S. (1961) Service Behavior of Blankets as a Method of Sealing Dams. Transactions, 7th International Congress on Large Dams, Rome, Vol. 4, pp. 301-325. Bruce, D.A., di Cervia, A. Ressi and Venti, J. Amos (2006) Seepage Remediation by Positive Cut-Off Walls: A Compendium and Analysis of North American Case Histories. ASDSO Dam Safety, Boston, MA, September 10-14.
221
Bürgisser, H.M., Gansser, A. and Pika, J. (1982) Late glacial lake sediments of the Indus valley area, northwestern Himalaya. Ecologae Geologica Helvetica 75 (1): 51- 63. Casagrande, A. (1937) Seepage through dams. Contributions to soil mechanics 1925– 1940. Boston Society of Civil Engineers, 1940 (reprinted from J New Engl Water Works Assoc, vol 51, pp 131– 172, 1937) p 295 Casagrande, A. (1961) Control of Seepage Through Foundations and Abutments of Dams. Geotechnique, Vol. XI, September 1961, No. 3, pp. 161-181 Cedergren, H.R. (1972). Seepage control in earth dams, in Embankment Dam Engineering, Casagrande Volume. R.C. Hirschfeld & S.J. Poulos (eds). Wiley. Christian, J.T. (1980) Flow nets from finite element data. International Journal for Numerical and Analytical Methods in Geomechanics, 4(2), pp 191 - 196. Clibborn, J. (1902) Experiments on the passage of water through sand, Govt. of India, Central Printing Office CMTL (2003) Final Report on Subsoil Investigation of Satpara Dam Project Skardu, Central Material Testing Laboratory, WAPDA, Lahore, Pakistan. Cronin, V.S. (1989) Structural setting of the Skardu intermontane basin, Karakoram Himalaya, Pakistan. Geological Society of America Special Paper 232 (Tectonics of the Western Himalayas) Pp 213-216. Darcy, H. (1865) Les Fontaines Publiques de la Ville de Dijon, Dalmont, Paris Desai, C.S. (1975) Free surface seepage through foundation and berm of cofferdams. Indian Geotechnical Journal, vol. 5, Number 1, January 1975, pp 1 – 10. Desai, C.S. (1976) Finite element residual schemes for unconfined flow. International Journal for Numerical Methods in Engineering, 10, pp 1415-1418. (1976). Desai, C.S. and Beseghi, B. (1988) Theory of residual flow procedure for 3-D free surface seepage. Advances in Water Resources, Vol 11, No 4, pp192 – 203. Deutsch, C. (1989) Calculating effective absolute permeability in sandstone/shale sequences, SPE Formation Evaluation, 4: 343–348. Deutsch, C.V. (2002) Geostatistical Reservoir Modelling, Oxford University Press, Oxford, N.Y. Elkateb, T., R. Chalaturnyk, and P.K. Robertson, (2003) An overview of soil heterogeneity: quantification and implications on geotechnical field problems. Can. Geotech. J. 40: 1–15 Faiz-ul-Hassan (2001) Performance of Main Embankment Dam and its Foundation by Evaluation and Adoption of Seepage Control Measures at Khanpur Dam. M.Phil.
222
Thesis in Water Resources Engineering, Centre of Excellence in Water Resources Engineering, University of Engineering and Technology, Lahore, Pakistan. Fan, Y. and Tompkins F.D. (1992) Generation of flow nets using F.E.M. nodal potentials and bilinear shape functions. International Journal for Numerical and Analytical Methods in Geomechanics, Vol 16, pp 425 – 437. Farrar, J.A., Wirkus, K.E. and McClain, J.A. (1990) Foundation Treatment Methods for the Jackson Lake Dam Modification. Proceedings of the 10th Annual USCOLD Lecture Series, New Orleans, LA, March 6-7, 33 p. Fell, R., MacGregor, P., Stapledon, D. & Bell, G. (2005): Geotechnical Engineering of Dams, A. A. Balkema Publishers, Leiden, Netherlands, ISBN 04 1536 440 x. Fell, R., Wan, C.F., Cyganiewics, J. and Foster, M. (2003) Time for development of internal erosion and piping in embankment dams. J. Geotech. Geoenviron. Eng. 129(4):307–314 Foster, M., Fell, R. and Spannagle, M. (2000a) The statistics of embankment dam failures and accidents. Can. Geotech. J. 37(5):1000–1024 Foster, M., Fell, R. and Spannagle, M. (2000b) A method for assessing the relative likelihood of failure of embankment dams by piping. Can. Geotech. J. 37(5):1025– 1061 Foster, M., Fell, R. and Spannagle, M. (2002) A method for assessing the relative likelihood of failure of embankment dams by piping: reply. Can. Geotech. J. 39:497– 500 Fredlund, D.G., and Xing, Anqing. (1994) Equations for the soil-water characteristic curve. Canadian Geotechnical Journal, Vol. 31, pp: 521-532. Gagliardi, J., and Routh, R. (1993) Geotechnical Modifications at Meeks Cabin Dam. ASCE Specialty Conference on Geotechnical Practice in Dam Rehabilitation, North Carolina State University, Raleigh, NC, April 25-28, pp. 755-769. Goharnejad, H., M. Noury, A. Noorzad, A. Shamsaie and A. Goharnejad, (2010) The Effect of Clay Blanket Thickness to Prevent Seepage in Dam Reservoir. Research Joumal of Environmental Sciences 4 (6): 558-565 Golder, H. Q., and Bazett, D. J. (1967) An Earth Dam Built by Dumping Through Water. Transactions, 9th International Congress on Large Dams, Istanbul, Vol. 4, pp. 369-387. Griffiths, D.V. and Fenton, A. (1993) Seepage beneath water retaining structures founded on spatially random soil. Géotechnique, 43(4): 577–587 HEPO (2002) Volume V: Annexures to Volume I (Annex A to D), Tender Documents for Construction of Civil Works and Hydraulic Steel Works, Contract Lot: 1A Dam
223
and Powerhouse, Satpara Dam Project, Hydro Electric Planning Organization, WAPDA, Lahore, Pakistan. Hewitt, K. (1999) Quaternary Moraines vs Catastrophic Rock avalanches in the Karakoram Himalaya, Northern Pakistan. Quaternary Research 51: 220-237, 11 Fig., 4 Tab. Hewitt, K. (2002) Styles of rock avalanche depositional complex in very rugged terrain, Karakoram Himalaya, Pakistan. In: Evans, S.G., DeGraff, J.V. (Eds.), Catastrophic Landslides: Effects, Occurrence, and Mechanisms. Geological Society of America, Reviews in Engineering Geology, Vol. 15, pp. 345–378. Honjo, Y. and Veneziano, D. (1989) Improved filter criterion for cohesionless soils. J. Geotech. Eng. 115(1):75–94 Honjo, Y., P.A. Naushahi and P.H. Giao (1992) Seepage Analysis of Tarbela Dam (Pakistan) using Finite Element Method. Proceedings of the Symposium on Prediction versus Performance in Geotechnical Engineering, Bangkok, Ed. by A.S. Balasubramaniam et. al., Balkema, pp. 219-236. Izhar Ul Haq (1996) Tarbela Dam: Resolution of Seepage. Proceedings of the Institute of Civil Engineers, Geotechnical Engineering, Vol. 119, Issue 1, pp. 49-56. Izhar Ul Haq and Altaf Ur Rehman (1984) Sinkholes at Tarbela Dam Project. In Proceedings of the First Multidisciplinary Conference on Sinkholes, Orlando, Florida. Jafarzadeh, F. and Soleimanbeigi, A. (2005) Uncertainty in Steady State Seepage Analysis of Embankment Dams. In Proceedings of the Symposium on Uncertainty Assessment in Dam Engineering at the 73rd Annual Meeting of ICOLD, Tehran, IRAN, Paper no. 102-S5. Jamil, S.M., N. Nawab, M.S. Javaid and S. Ali (2010) Geotechnical Evaluation of Seepage Remedial Measures at Khanpur Dam. Proceedings of the International Conference on Geotechnical Engineering, Lahore, Pakistan, PGES, in Association with NESPAK, UET, ACE, pp. 193-197. Jansen, R.B. (1983) Dams and public safety. Water Resources Technical Publication, US Department of the Interior, Bureau of Reclamation, pp 14–16 Kenney, T.C. and Lau, D. (1985) Internal stability of granular filters. Can. Geotech. J., Ottawa, Can., 22:215–225. Kezdi, A. (1979) Soil physics selected topics—developments in geotechnical engineering, Elsevier, Amsterdam, pp 1–160 Khoury, M.A., Harris, A.L. Jr., and Dutko, P. (1989) Construction of a Cement- Bentonite Cutoff Wall to Control Seepage at the Prospertown Lake Dam, Association of State Dam Safety Officials, 6th Annual Conference, Albuquerque, NM, October 1- 5, 6 p.
224
King, P.R. (1989) The use of renormalization for calculating effective permeability. Transport in Porous Media, 4: 37–58. Knight, D.J. (1990) The proven usefulness of instrumentation systems on varied dam projects, Geotechnical Instrumentation In Practice, Purpose, Performance And Interpretation, ICE, Thomas Telford, London pp. 401-414 Kovacs, G. (1981) Developments in water science—seepage hydraulics, Chap 3.2. Elsevier, Amsterdam, pp 349–379 Kuhle, M. (2001) The maximum ice age (LGM) glaciation of the Central- and South Karakorum: an investigation of the heights of its glacier levels and ice thicknesses as well as lowest prehistoric ice margin positions in the Hindukush, Himalaya and in East-Tibet on the Minya Konka-massif, GeoJournal 54 (2–4), 55 (1), Tibet & High Asia (VI): Boston/London, 109–396. Kuhle, M. (2004) The Pleistocene Glaciation in the Karakoram-Mountains: Reconstruction of Past Glacier Extensions and Ice Thicknesses, Journal of Mountain Science, Vol 1, No 1, 3-17. Kulesza, R., Pflueger, K., Gardner, C. and Yen, P. (1994) Investigations and Design of Diaphragm Wall for Seepage Control at Wells Dam, 14th Annual USCOLD Meeting, Phoenix AX, June, pp. 223-244. Lafuente, R., Granell, J., Poyales, I., and Florez, V. (2006) La Loteta Dam – A Strategic Location. In: Berga et al. (eds), Dams and Reservoirs, Societies and Environment in the 21st Century: Proceedings of the International Symposium on Dams in the Societies of the 21st Century, 22nd International Congress on Large Dams (ICOLD-SPANCOLD), Barcelona, Spain, published by Taylor & Francis, 2006, ISBN 0 415 40423 1, Volume I, p. 415-420. Lambe, T.W. and Whitman, R.V., (1979) Soil Mechanics (SI version), John Wiley, New York. Lane, E.W. (1934) Security from under-seepage masonry dams on earth foundations. Trans. ASCE 60(4):929–966 Lane, K.S. and Wohlt, P.E. (1961) Performane of Sheet Piling and Blankets for Sealing MissouriRiver Reservoirs. Transactions, 7th International Congress on Large Dams, Rome, Vol. 4, pp. 255-279. Lefebvre, G., C. Lupien, J.J. Pare and J.P. Tournier, (1981) Effectiveness of Seepage Control Elements for Embankments on Semipervious Foundations. Canadian Geotechnical Journal, 18, 572-576. Legates, D. R. and McCabe. G. J. (1999). Evaluating the use of “goodness-of-fit” measures in hydrologic and hydroclimatic model validation. Water Resources Research. 35(1): 233-241.
225
Lowe, J. III (1978) The Fourth Nabor Carrillo Lecture: Foundation Design, Tarbela Dam. Presented at the Mexican Society for Soil Mechanics, X National Meeting, Mérida, Yucatán, Mexico, November 18, 1978. México: Sociedad Mexicana de Mecánica de Suelos. Lumb, P. (1970) Safety factors and the probability distribution of soil strength. Canadian Geotechnical Journal, 7(3): 225–242. Malik, Z.M., A. Tariq, and J. Anwer, (2008) Seepage Control for Satpara Dam, Pakistan. Proceedings of the Institute of Civil Engineers, Geotechnical Engineering, Volume 161, Issue GE5, October 2008, p. 235-246. McCook, D.K. (2004) A comprehensive discussion of piping and internal erosion failure mechanisms. In: Proceedings, annual dam safety conference, ASDSO. Phoenix, Arizona Meek, J.L. and Beer, G. (1976) Contour plotting using isoparamertic element representation. International Journal for Numerical Methods in Engineering, 10, pp 954 – 957. Morgenstern, N.R. (2000) Performance in geotechnical practice. The Inaugural Lumb Lecture. Hong Kong Institution of Engineers, May 2000, 59 pp. Moriasi, D.N., J.G. Arnold, M.W. Van Liew, R.L. Bingner, R.D. Harmel and T.L. Vieth, (2007). Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE. 50:(3), 885–900. Mukhopadhyay, Suvasish (2008) Seepage Analysis through Foundation using F.E.M. and Flownet. The 12th International Conference of International Association for Computer Methods and Advances in Geomechanics (IACMAG), Goa, India Nash, J. E. and Sutcliffe. J. V. (1970). River flow forecasting through conceptual models part I-A discussion of principles. Journal of Hydrology. 10(3): 282-290. Norin, E. (1925) Preliminary notes on the late Quaternary glaciation of the Northwest Himalaya. Geografiska Annaler 7 Norris, R.J., J.M. Lewis and U. Heriot-Watt, (1991) The geological modeling of effective permeability in complex heterolithic facies. In Proceedings of the 66th Annual Technical Conference and Exhibition, SPE 22692, Dallas, Tex. Vol. W, pp. 359–374. Owen, L. (1988) Wet-sediment deformation of Quaternary and recent sediments in the Skardu Basin, Karakoram mountains, Pakistan. In: Glaciotectonics, Forms and Processes, Rotterdam (Balkema). Pagano, M.A., and Pache, B. (1995) Construction of the Meeks Cabin Cut Off Wall. Association of State Dam Safety Officials, 11th Annual Conference, Atlanta, GA, September 17-20, 19 p.
226
Peck, Ralph B. (2000) RISK ANALYSIS FOR DAMS – The Reluctant Conversion of a Skeptic, In Proceedings of the Workshop on Risk Assessment for Seepage and Piping in Dams and Foundations, Sponsored by the U. S. Army Corps of Engineers Organized by the Center for Geotechnical Practice and Research, Virginia Tech, Blacksburg, VA. Peterson, R. (1968) Recent Trends in Earth Dam Engineering. The Engineering Journal, Canada, EIC-68-Geotech 1, Apr., pp. 30-36. Phoon, K.K., and Kulhawy, F.H. (1999) Characterization of geotechnical variability. Canadian Geotechnical Journal, 36(4): 612–624. Rice, John D. (2007) A Study on the Long-Term Performance of Seepage Barriers in Dams, PhD Dissertation, Virginia Polytechnic Institute and State University, Blacksburg, Virginia. Richards, K.S. and Reddy, K.R. (2007) Critical appraisal of piping phenomena in earth dams. Bull. Eng. Geol. Environ., 66:381-402 Ripley, C. F. and Campbell, D. B. (1964) Performance of Earth Dam on Compressible and Pervious Foundation. Transactions, 8th International Congress on Large Dams, Edinburgh, Vol. 2, 431-451. Roberts, D.L. and Ho, D. (1991) Construction of Diaphragm Wall at Wells Dam East Embankment. Slurry Walls: Design, Construction, and Quality Control, ASTM STP 1129, Eds. D.B. Paul, R.R. Davidson, and N.J. Cavalli, Philadelphia, PA, pp. 235- 250, (published 1992). Santhi, C. Arnold, J. G., Williams, J. R., Dugas, W. A., Srinivasan, R. and Hauck, L. M. (2001). Validation of the SWAT model on a large river basin with point and nonpoint sources. Journal of American Water Resources Association. 37(5): 1169- 1188. Schultze, E. (1975) Some aspects concerning the application of statistics and probability to foundation structures. In Proceeding of the 2nd International Conference on the Applications of Statistics and Probability in Soil and Structure Engineering, Aachen, Germany, 15–18 Sept., pp. 457–494. Sherard, J.L. and Dunnigan, L.P. (1989) Critical filters for impervious soils. J. Geotech. Eng. Proc. ASCE 115(7): 927–947 Sherard, J.L., L.P. Dunnigan and J.R. Talbot, (1984a) Basic properties of sand and gravel filters. J. Geotech. Eng. ASCE 110(6):684–700 Sherard, J.L., L.P. Dunnigan and J.R. Talbot, (1984b) Filters for silts and clays. J. Geotech. Eng. ASCE 110(6):701–718
227
Shroder, J. F. Jr., L. Owen and E. Derbyshire, (1993) Quaternary glaciation of the Karakoram and Nanga Parbat Himalaya. In: Himalaya to the sea: Geology, Geomorphology and the Quaternary. Pp 132-158. New York: Routledge Singh, A. and Punmia, B.C. (1973) Seepage below horizontal apron with downstream cutoff, founded on anisotropic pervious medium of finite depth. Indian Geotechnical Journal, Vol. 3, Number 3, July 1973, pp 198 - 215. Singh, J., H.V. Knapp and M. Demissie, (2004). Hydrologic modeling of the Iroquois River watershed using HSPF and SWAT. ISWS CR 2004-08. Champaign, Ill.: Illinois State Water Survey. Available at: www.sws.uiuc.edu/pubdoc/CR/ISWSCR2004- 08.pdf Singh, N.K., R. Douglas, S.R. Ahlfield and M. Grant, (2005) Construction of a Roller Compacted Concrete Upstream Blanket and Plastic Concrete Cutoff Wall at Cleveland Dam. Association of State Dam Safety Officials, New Orleans, LA, September 25-29. Skempton, A.W. and Brogan, J.M. (1994) Experiments on piping in sandy gravels. Geotechnique 44(3):449–460 Stelzer, J.F. and Welzel, R. (1987) Plotting of contours in a natural way. International Journal for Numerical Methods in Engineering, 24, pp 1757 – 1769. Talbot, James R., Steve J. Poulos and Ronald C. Hirschfeld, (2000) Remediation of Seepage Problems Through Cutoff or Reduction of Flow and through Collection and Control of Seepage including the use of Geosynthetics. White Paper in The National Dam Safety Program, Research Needs Workshop: Seepage through Embankment Dams, presented in Denver, Colorado. Tariq, Sardar M. (2003) Control of seepages through deep alluvium foundations of Tarbela Dam Project - Pakistan, In: Llanos et al. (eds.), Dam Maintenance and Rehabilitation, Swets and Zeitlinger, Lisse, ISBN 90 5809 534 7, pp. 899-904. Telling, R. M., B.K. Menzies and J.M. Coulthard, (1978) A Design Method for Assessing the Effectiveness of Partially Penetrating Cut-Off Walls. Ground Engineering, Vol 11, No. 8, Nov 1978, pp 48-51. Available from: Technical Information Center, U. S. Army Engineer Waterways Experiment Station, P. O. Box 631, Vicksburg, MS 39180-0631. Telling, R. M., B.K. Menzies and N.E. Simons, (1978a) Cut-Off Efficiency, Performance and Design. Ground Engineering, Vol 11, No. 1, Jan 1978a, pp 30-43. Available from: Technical Information Center, U. S. Army Engineer Waterways Experiment Station, P. O. Box 631, Vicksburg, MS 39180-0631. Telling, R. M., B.K. Menzies and N.E. Simons, (1978b) The Effectiveness of Jointed Cut-Off Walls Beneath Dams on Pervious Soil Foundations. Ground Engineering, Vol 11, No. 4, May 1978b, pp 27-37. Available from: Technical Information Center,
228
U. S. Army Engineer Waterways Experiment Station, P. O. Box 631, Vicksburg, MS 39180-0631. Terzaghi, K. (1922) Der Grundbruch an Stauwerken und seine Verhutung (The failure of dams by piping and its prevention). Die Wasserkraft, vol 17, pp 445–449. Reprinted in (1960) From theory to practice in soil mechanics. Wiley, New York Terzaghi, K. (1939) Soil mechanics: a new chapter in engineering science. J. Instn. Civ. Eng. 12:106–141
Terzaghi, K. (1943) Theoretical soil mechanics. Wiley, New York, pp 1–510 Terzaghi, K. and Peck, R.B. (1948) Soil mechanics in engineering practice. Wiley, New York Terzaghi, K. and Peck, R.B. (1967) Soil mechanics in engineering practice, 2nd edn. Wiley, New York Terzaghi, K., Peck, R.B. and Mesri, G. (1996) Soil mechanics in engineering practice, 3rd edn. Wiley, New York Tracy, F.T. and Radhakrishnan, N. (1989) Automatic generation of seepage flow nets by finite element method. Journal of Computing in Civil Engineering, Vol 3, No 3. July (1989), pp 268 – 284. Uginchus, A. (1935) Seepage through earth dams with aprons. Moskvavolgostroi, 11: 67-79 Van Liew, M.W., J.G. Arnold and J.D. Garbrecht, (2003). Hydrologic simulation on agricultural watersheds: Choosing between two models. Transactions of the ASAE. 46(6): 1539-1551. Von-Thun, J. (1996) Understanding Piping and Seepage Failures – The No. 1 Dam Safety Problem in the West, USBR Bulletin No. 176, Water Operation and Maintenance, Adapted from a paper presented at the 1996 Western Regional Conference of the ASDSO. WAPDA (2007) Khanpur Dam Project Fourth Periodic Inspection, Technical Resource Services and DSO. WAPDA (2007) Tarbela Dam Project Report on Fifth Periodic Inspection August 2006 and March 2007, NESPAK, NDC and ACE. Warren, J.E., and Price, H.S. (1961) Flow in heterogeneous porous media. Society of Petroleum Engineers Journal, 2: 153–169. Weston, S., A. Hurley, S. Woods, D. Preston and I. Carter, (2003) Wilsons Dam: Design expectations compared with measured performance. Dams - Consents and Current Practice, NZSOLD, IPENZ Proceedings of Technical Groups, Volume 30, Issue 2 (LD)
229
Wilson, S.D. and Marsal, R.J. (1979) Current Trends in Design and Construction of Embankment Dams. prepared for the Committee on International Relations, International Commission on Large Dams (ICOLD) and ASCE Geotechnical Division, New York, NY: ASCE, 978-0-87262-197-8 or 0-87262-197-9, 1979, pp. 125. Yang, D.S. and Takeshima, S. (1994) Soil mix walls in difficult ground. In Situ Ground Improvement Case Histories, American Society of Civil Engineers, National Convention, Atlanta, GA. October. pp. 106-120.
230
ANNEXURE - I: SUBSURFACE LOG REPORTS
231
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
Phase-I (Feasibility Stage: 1988) BH1 3437555.81 1243213.29 2636.5 18.3 0.0 5.6 Sand sand with some silt and gravels 5.6 6.2 Boulders boulder of granite 6.2 9.1 Sand medium coarse 0.13 sand 9.1 18.3 Sand medium coarse sand BH2 3437528.22 1243228.04 2633.5 18.3 0.0 1.5 Sand medium to coarse sand with gravels and silt 1.5 3.0 Sand Sand 3.0 9.1 Sand medium coarse sand 9.1 18.3 Sand Sand BH3 3437553.61 1243242.31 2633.5 18.3 0.0 15.2 Sand fine to coarse sand 0.001 with some silt and gravels 15.2 18.3 Sand fine to coarse sand with some silt and gravels BH4 3437525.55 1243194.67 2633.5 9.1 0.0 1.7 Sand medium to coarse sand with some silt and gravels 1.7 3.0 Sand Sand fine 3.0 4.6 Sand medium coarse sand 4.6 7.0 Boulders boulder of granodiorite 7.0 9.1 Sand medium coarse sand BH5 3437604.56 1243158.76 2636.5 6.1 0.0 2.1 Sand mediume to coarse sand with some silt and gravels 2.1 2.7 Boulders Boulder of Granodiorite 2.7 6.1 Sand Sand BH6 3437453.89 1243191.44 2635.0 6.1 0.0 2.4 Sand medium coarse sand 2.4 2.7 Boulders boulder of granodiorite 2.7 4.3 Sand medium coarse sand 4.3 4.6 Boulders boulder of granodiorite 4.6 6.1 Sand fine sand 0.095 P1 3437662.86 1243153.57 2656.3 6.1 0.0 3.0 Silt sandy silt 0.168 3.0 6.1 Silt sandy silt P2 3437591.26 1243203.73 2633.5 9.1 0.0 2.4 Silt sandy silt 0.168 2.4 4.6 Silt sandy silt 4.6 9.1 Clay Silty Clay P3 3437633.53 1243259.72 2633.5 0.0 A P4 3437600.60 1243244.10 2633.5 3.7 0.0 1.8 Silt sandy silt 3.027 1.8 3.7 Silt sandy silt P5 3437405.64 1243271.61 2638.1 6.1 0.0 1.8 Silt sandy silt 1.8 2.3 Gravels sandy gravel 0.168 2.3 2.7 Gravels sandy gravel 2.7 6.1 Silt sandy silt P6 3437352.33 1243286.40 2638.1 6.1 0.0 2.1 Silt sandy silt 0.2 2.1 4.3 Silt sandy silt 4.3 4.6 Sand medium coarse sand 4.6 6.1 Gravels sandy gravel P7 3437429.95 1243291.40 2636.5 6.1 0.0 2.4 Silt sandy silt 2.4 3.4 Clay Clay 3.4 4.0 Sand medium coarse sand
232
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
4.0 4.3 Silt sandy silt 4.3 6.1 Sand silty sand P8 3437387.70 1243241.03 2638.1 3.0 0.0 3.0 Silt sandy silt P9 3437432.55 1243245.52 2638.1 1.8 0.0 1.8 Silt sandy silt 1.83 9/1/1980 P10 3437431.23 1243261.86 2636.5 6.1 0.0 6.1 Silt sandy silt P11 3437476.93 1243243.37 2636.5 3.7 0.0 3.7 Silt sandy silt P12 3437640.15 1243192.39 2656.3 3.0 0.0 3.0 Silt sandy silt P13 3437627.51 1243226.26 2636.5 3.0 0.0 3.0 Silt sandy silt Phase-II (Design / Tender Stage: 2003) SD1 3437535.71 1243216.58 2633.5 19.0 0.0 0.5 Gravels Gravels with sand and silt. 0.5 4.5 Sand Medium to coarse 0.50 10/18/2001 0.196 sand with silt. 4.5 9.5 Gravels Overburden consist 0.61 10/22/2001 0.16 of medium to coarse gravel, sand and silt mixed with gravel. Occasional Boulders at various depths. 9.5 13.0 Gravels Overburden consist of medium to coarse gravel, sand and silt mixed with gravel. Occasional Boulders at various depths. 13.0 14.0 Boulders Boulders of 2.00 10/26/2001 0.019 granodiorite. 14.0 19.0 Boulders Boulders of granodiorite. SDL3 3437635.27 1243521.05 2666.1 78.2 0.0 5.5 Boulders Scree deposit 8.87E-05 consist of boulders and gravels of granodiorite, slate with large amount of silt and some cobbles of granodiorite. 5.5 7.7 Gravels Glacial drift 0.00061 deposit consist of silty gravel of slate, fine grained cobbles of hornfels. 7.7 10.4 Gravels Glacial drift deposit consist of silty gravel of slate, fine grained cobbles of hornfels. 10.4 11.0 Boulders Boulder of slate 0.00184 5'7' in length. 11.0 12.5 Boulders Boulder of slate 5'7' in length. 12.5 13.9 Boulders Boulders of 0.00365 granodiorite. 13.9 18.6 Boulders Boulders of granodiorite. 18.6 21.9 Cobbles Cobbles gravel of slate and hornfels with appreciable amount of silt. 21.9 28.0 Boulders Boulders and 27.99 8/30/2002 75% cobbles of slate Water having length 2'3'. Loss 28.0 28.7 Boulders Boulders and cobbles of slate having length 2'3'.
233
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
28.7 30.5 Boulders Boulders and gravels of hornfels with boulder size 2'3'. Gravels cobbles of hornfels and slate. 30.5 34.9 Boulders Boulders of 75% hornfels with 1020 Water ft long. Loss 34.9 37.8 Boulders Boulders of hornfels with 1020 ft long. 37.8 42.7 Cobbles Cobbles and pebbles of granodiorite and slate. 42.7 48.8 Cobbles Cobbles and pebbles of granodiorite and slate. 48.8 61.3 Boulders Boulders of granodiorite. 61.3 78.2 Gravels Gravels of granodiorte and slate. SDL5 3437578.20 1243528.59 2673.0 29.8 0.0 3.4 Boulders Glacial drift material consists of anguler to subanguler boulders, gravels of slate and hornfels with large amount of silt. Length of boulder is 1 to 2 ft. 3.4 5.2 Clay Finer material is 0.049 silty clay yellowish brown in colour. 5.2 5.5 Clay Finer material is silty clay yellowish brown in colour. 5.5 8.2 Boulders Boulders of slate 0.015 and hornfels. Boulders length is 7 ft. 8.2 9.1 Boulders Boulders of slate and hornfels. Boulders length is 7 ft. 9.1 11.4 Boulders Silty boulder 0.0113 gravel, broken pieces are of slate and hornfels, with large amount of silt. 11.4 14.0 Boulders Boulders of slate 0.0065 upto 3' length with anguler to subanguler gravel. 14.0 14.3 Boulders Boulders of slate upto 3' length with anguler to subanguler gravel. 14.3 16.0 Boulders 10 to 20 ft long 100% boulders of Water hornfels Loss
234
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
16.0 17.5 Boulders Silty boulders and gravels of hornfels and slate with large amount of silt. 17.5 19.7 Boulders Boulders of slate with 5'7' in length. 19.7 23.8 Boulders Boulders of hornfels with 1020 ft in length. 23.8 25.0 Boulders Boulder of slate upto 3 ft in length with angular to subangular gravels. 25.0 26.5 Boulders Boulders of slate 26.49 6/30/2002 100% with 5'7' in length. Water Loss 26.5 26.7 Boulders Boulders of slate with 5'7' in length. 26.7 28.7 Cobbles Cobble gravel of slate and hornfels with appreciable amount of silt. 28.7 29.8 Boulders Boulders and gravels of hornfels. Boulder size is 23 ft, gravels cobble of slate and hornfels. SDR1 3437702.68 1243461.74 2629.6 61.5 0.0 2.1 Gravels Angular to 2.00 4/17/2002 subangular gravels of granodiorite and slates with silty sand. Sand is brownish grey, fine to coarse grained with little silty clay. 2.1 5.0 Gravels Angular to subangular gravels of granodiorite and slates with silty sand. Sand is brownish grey, fine to coarse grained with little silty clay. 5.0 11.2 Boulders Boulders of 2.00 4/23/2002 0.109 granodiorite. Maximum size of boulder is 3.25 ft. (with little silt from 36.08 to 36.74 ft) 11.2 14.5 Boulders Boulders of 1.90 4/26/2002 0.031 granodiorite and slate with angular to subangular fine gravels and little silt. 14.5 15.4 Boulders Boulders of granodiorite and slate with angular to subangular fine gravels and little silt. 15.4 17.7 Boulders Boulders of 2.10 4/28/2002 0.02 granodiorite with little angular to subangular gravels.
235
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
17.7 20.0 Boulders Boulders of 2.50 4/29/2002 0.15 granodiorite with little angular to subangular gravels. 20.0 23.0 Boulders Boulders of 2.40 5/1/2002 0.1 granodiorite and slate with some angular gravels. 23.0 25.0 Boulders Boulders of granodiorite and slate with some angular gravels. 25.0 26.0 Boulders Boulder of slate 2.10 5/4/2002 0.014 26.0 29.0 Boulders Boulders of 1.95 5/6/2002 0.15 granodiorite and slate with anguler gravels 29.0 30.6 Boulders Boulders of granodiorite and slate with anguler gravels 30.6 32.0 Boulders Boulders of 1.95 5/8/2002 0.013 granodiorite with washed sampling brownish colour, sticky silty clay between 108.6 ft and 109.2 ft 32.0 36.1 Boulders Boulders of 2.15 5/11/2002 0.017 granodiorite with washed sampling brownish colour, sticky silty clay between 108.6 ft and 109.2 ft 36.1 39.0 Boulders Boulders of granodiorite with washed sampling brownish colour, sticky silty clay between 108.6 ft and 109.2 ft 39.0 39.6 Boulders Boulders of slate 39.6 40.0 Boulders Boulders of 1.93 5/13/2002 0.023 granodiorite 40.0 41.0 Boulders Boulders of granodiorite 41.0 45.4 Boulders Boulders of slate 1.96 5/17/2002 0.017 and granodiorite 45.4 49.5 Boulders Boulders of slate and granodiorite 49.5 57.7 Rock Granodiorite light 1.90 5/26/2002 0.0001196 9.2 grey, brownish on weathered surface, coarse grained, moderately to highly weathered, jointed. 57.7 61.5 Rock Granodiorite light 8.58E-05 6.6 grey, brownish on weathered surface, coarse grained, moderately to highly weathered, jointed.
236
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
SDR2 3437721.18 1243445.67 2636.8 35.0 0.0 3.0 Gravels Silty sandy gravels,anguler to subanguler gravels of granodiorite and slates with brownish grey, fine to medium grained silty sand. 3.0 3.9 Boulders Material consist of 100% anguler to Water subanguler gravels, Loss boulders of granodiorite and slate having little silt. 3.9 10.1 Boulders Material consist of 10.00 5/2/2002 anguler to subanguler gravels, boulders of granodiorite and slate having little silt. 10.1 14.0 Boulders Boulders of 10.05 5/6/2002 0.28 granodiorite and slate with minor amount of silt and fine gravels, silt is brownish grey in colour. Maximum size of boulder is 0.55 ft. 14.0 17.0 Boulders Boulders of 10.03 5/9/2002 0.31 granodiorite and slate with minor amount of silt and fine gravels, silt is brownish grey in colour. Maximum size of boulder is 0.55 ft. 17.0 19.2 Boulders Boulders of granodiorite and slate with minor amount of silt and fine gravels, silt is brownish grey in colour. Maximum size of boulder is 0.55 ft. 19.2 20.0 Rock Slate and 9.80 5/12/2002 0.0055 granodiorite, grey fine grained moderately hard slates intruded by igneous rock (granodiorite) which is light grey to grey, coarse grained and jointed.
237
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
20.0 22.0 Rock Slate and granodiorite, grey fine grained moderately hard slates intruded by igneous rock (granodiorite) which is light grey to grey, coarse grained and jointed. 22.0 23.0 Rock Slate, grey to dark 10.30 5/15/2002 0.068 grey, weathered surface is brownish, fine grained, highly fractured and jointed, cleavage developed, weathering effect is visible on joint surfaces. 23.0 25.0 Rock Slate, grey to dark 9.85 5/17/2002 0.000364 28 grey, weathered surface is brownish, fine grained, highly fractured and jointed, cleavage developed, weathering effect is visible on joint surfaces. 25.0 27.6 Rock Slate, grey to dark grey, weathered surface is brownish, fine grained, highly fractured and jointed, cleavage developed, weathering effect is visible on joint surfaces. 27.6 28.0 Rock Slate and 10.05 5/20/2002 0.000663 51 granodiorite, grey to dark grey, brownish grey at places, fine grained slates continued and intruded by granodiorite, jointed. 28.0 30.3 Rock Slate and granodiorite, grey to dark grey, brownish grey at places, fine grained slates continued and intruded by granodiorite, jointed.
238
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
30.3 32.0 Rock Granodiorite, light 10.05 5/24/2002 0.000572 44 grey coarse grained, hard and moderately jointed. Joint surfaces are weathered. 32.0 35.0 Rock Granodiorite, light 9.86 5/27/2002 0.000598 46 grey coarse grained, hard and moderately jointed. Joint surfaces are weathered. SDR6 3437824.76 1243602.18 2629.8 40.7 0.0 2.6 Boulders Scree material deposit consisting of boulder (14 ft) and gravels of granodiorite and slate, with appreciable amount of silt from depth 1' to 4'. 2.6 5.2 Cobbles Disintegrated 0.07 material consists of 15% Cobbles and Pebbles (10% of granodiorite and 5% of slate). 5.2 8.0 Cobbles Disintegrated material consists of 15% Cobbles and Pebbles (10% of granodiorite and 5% of slate). 8.0 8.5 Boulders Scree material 5.20 8/6/2002 0.0015 consists of 25% silty gravel of granodiorite, with 15% cobbles of slate (size 3"4"), and boulders of granodiorite size 2'3' 8.5 9.6 Boulders Scree material consists of 25% silty gravel of granodiorite, with 15% cobbles of slate (size 3"4"), and boulders of granodiorite size 2'3' 9.6 10.3 Boulders Boulders of 100% granodiorite (size 5' Water to 7') Loss 10.3 15.5 Boulders Boulders of granodiorite (size 5' to 7') 15.5 16.7 Cobbles Cobbles and gravels of slate and hornfels, with amount of silt. 16.7 17.4 Boulders Boulders,Gravels,& 5.30 8/12/2002 0.0075 Cobbles of granodiorite 17.4 20.4 Boulders Boulders,Gravels,& 5.30 8/15/2002 0.05 Cobbles of granodiorite
239
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
20.4 21.8 Boulders Boulders,Gravels,& Cobbles of granodiorite 21.8 22.7 Gravels Granodiorite gravels with medium coarse sand 22.7 23.2 Sand Gravelly coarse sand 23.2 23.4 Sand Medium coarse 5.25 8/17/2002 0.1 sand 23.4 24.2 Sand Medium coarse sand 24.2 25.3 Gravels Granodiorite gravels with medium coarse sand 25.3 27.0 Boulders boulders of 4.99 8/19/2002 0.38 granodiorite 27.0 29.4 Boulders boulders of 5.20 8/21/2002 0.16 granodiorite 29.4 32.1 Boulders boulders of granodiorite 32.1 32.6 Sand Gravelly coarse 5.15 8/23/2002 0.1 sand 32.6 32.9 Sand Gravelly coarse sand 32.9 33.7 Sand Medium coarse sand 33.7 35.3 Boulders Silty gravels and 100% boulders of Water granodiorite with Loss cobbles of slate 35.3 36.0 Sand Gravelly coarse sand 36.0 37.0 Sand Medium coarse sand 37.0 40.7 Boulders Boulders, gravels and cobbles of granodiorite SDSB1 3437793.50 1243652.88 2630.5 25.0 0.0 3.0 Boulders Silty gravels with cobbles and boulders. Size of gravels ranges from 3'' to 4''. 3.0 3.7 Boulders Overburden consist 3.66 6/20/2002 of boulders of granodiorite with appreciable amount of gravels of various materials. Maximum boulder size is 9''. 3.7 4.6 Boulders Overburden consist of boulders of granodiorite with appreciable amount of gravels of various materials. Maximum boulder size is 9''.
240
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
4.6 4.9 Boulders Overburden 3.60 6/22/2002 0.0093 consists of boulders of slate. Maximum size of boulder is 4 ft. 4.9 10.1 Boulders Overburden consists of boulders of slate. Maximum size of boulder is 4 ft. 10.1 11.0 Boulders Overburden 3.75 6/30/2002 0.015 consists of boulders of granodiorite and slate with clayey silty snad. Maximum size of boulder is 3 ft. 11.0 13.0 Boulders Overburden consists of boulders of granodiorite and slate with clayey silty snad. Maximum size of boulder is 3 ft. 13.0 14.0 Boulders Overburden 3.69 7/3/2002 0.028 consists of boulders of granodiorite with appreciable amount of sand. 14.0 17.0 Boulders Overburden 3.69 7/7/2002 0.32 consists of boulders of slate and appreciable amount of gravels and silty clay. 17.0 20.0 Boulders Overburden 3.66 7/11/2002 0.0095 consists of boulders of slate and appreciable amount of gravels and silty clay. 20.0 23.0 Boulders Boulders of 3.66 7/15/2002 0.01 granodiorite with brownish colour sand. 23.0 25.0 Boulders Boulders of granodiorite with brownish colour sand. TP1 3437722.26 1243505.19 2658.3 3.7 0.0 3.7 Gravels Silty gravel; 3.66 7/23/2002 angular to subangular boulders (20%), cobbles/gravels (56%) mixed with silty clay (24%) TP2 3437709.77 1243430.12 2660.0 6.1 0.0 6.1 Gravels Silty gravels/Boulders; angular to subangular gravels (45%), boulders (30%), mixed with clayey silt (25%).
241
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
TP2A 3437644.63 1243499.42 2666.1 3.0 0.0 3.0 Gravels Silty gravel; angular to subangular boulders (19%), cobbles/gravels (52%) mixed with silty clay (29%) TP3 3437411.87 1243367.99 2646.4 3.0 0.0 3.0 Silt Gravely silt; angular to subangular gravels (35%), boulders (25%), mixed with clayey silt (40%). TP4 3437330.80 1243328.06 2670.8 3.0 0.0 3.0 Gravels Silty gravel; angular to subangular gravels (40%), boulders (20%), mixed with clayey silt (40%). TP5 3437484.42 1243303.68 2642.9 3.0 0.0 3.0 Silt Gravely silt; angular to subangular boulders (15%), cobbles/gravels (35%), mixed with clayey silt (50%). TRB1 0.0 0.0 Gravels Scree material consist of angular to subangular gravel, cobble of granodiorite 4060% size about 48'' and boulder of 1' to 2', mixed with fine matrix of silty clay. 1.4 1.4 Gravels Scree material consist of gravels, cobbles (40% size 4"6") and boulders of granodiorite (30% size 2'4'), mixed with fine matrix of silty clay. TRB2 0.0 0.0 Gravels Scree material consist of angular to subangular gravel, cobble of granodiorite 4060% size about 48'' and boulder of 1' to 2', mixed with fine matrix of silty clay. 1.4 1.4 Gravels Scree material consist of gravels, cobbles (40% size 4"6") and boulders of granodiorite (30% size 2'4'), mixed with fine matrix of silty clay.
242
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
TLB1 0.0 0.0 Gravels Glacial drift deposit consists of mainly angular to subangular gravels of slate(15%) and hornfels(20%), boulders(4'7') mainly of hornfels, with varying amounts of silty clay of slate. 1.4 1.4 Gravels Silty gravels and cobbles (15% slate, 25% hornfels) and boulders of hornfels (1'5'). TLB2 0.0 0.0 Gravels Glacial drift deposit consists of mainly angular to subangular gravels of slate(15%) and hornfels(20%), boulders(4'7') mainly of hornfels, with varying amounts of silty clay of slate. 1.4 1.4 Gravels Silty gravels and cobbles (15% slate, 25% hornfels) and boulders of hornfels (1'5'). TLB3 0.0 0.0 Gravels Glacial drift deposit consists of mainly angular to subangular gravels of slate(15%) and hornfels(20%), boulders(4'7') mainly of hornfels, with varying amounts of silty clay of slate. 1.4 1.4 Gravels Silty gravels and cobbles (15% slate, 25% hornfels) and boulders of hornfels (1'5'). BAP1 3437580.20 1243458.35 2670.7 3.0 0.0 3.0 Gravels Silty gravels; Angular to subangular boulders (15%), gravels (50%), mainly of granodiorite and slate, mixed with sandy clayey silt (35%)
243
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
BAP2 3437667.91 1243595.18 2672.5 6.1 0.0 6.1 Gravels Silty gravels; Angular to subangular, rounded to subrounded boulders (20%), gravels (45%), mainly of granodiorite and slate, mixed with clayey silt (35%) Phase-III (Construction Stage: 2004) SBH1 3437671.93 1243488.39 2638.1 20.0 3.0 5.0 Gravels Gravels of slate in silty matrix. 5.0 5.5 Gravels Gravels of granodiorite. 5.5 7.5 Gravels Slate gravels in silty matrix. 7.5 10.5 Gravels Granodiorite gravels in silty matrix. 10.5 20.0 Gravels Granodiorite gravels and some boulders in silty matrix. SBH2 3437477.07 1243270.74 2638.1 20.0 0.0 1.0 Clay / Clayey silt with 0.00414 Silt boulders and gravels. Slate hornfels gravels and pebbles of granodiorite and marbled limestone. 1.0 3.0 Clay / Clayey silt with 0.00136 Silt boulders and gravels. Slate hornfels gravels and pebbles of granodiorite and marbled limestone. 3.0 4.0 Boulders Boulders and 0.0158 gravels of slate/hornfels in silty matrix. 4.0 5.0 Boulders Boulders and 0.0123 gravels of slate/hornfels in silty matrix. 5.0 6.0 Boulders Boulders and 0.0191 gravels of slate/hornfels in silty matrix. 6.0 7.0 Boulders Boulders and 0.0341 gravels of slate/hornfels in silty matrix. 7.0 8.0 Boulders Boulders and 0.00857 gravels of slate/hornfels in silty matrix. 8.0 9.0 Boulders Boulders and 0.0105 gravels of slate/hornfels in silty matrix. 9.0 10.0 Boulders Boulders and 0.0115 gravels of slate/hornfels in silty matrix.
244
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
10.0 11.0 Gravels Slate hornfels 9.10 5/10/2003 0.0109 gravels in silty matrix 11.0 12.0 Gravels Slate hornfels 0.00774 gravels in silty matrix 12.0 13.0 Gravels Slate hornfels 0.0086 gravels in silty matrix 13.0 14.0 Gravels Slate hornfels 0.0124 gravels in silty matrix 14.0 15.0 Sand Wash sample silty 0.00953 sand. 15.0 16.0 Gravels Gravels of slates. 0.00936 16.0 17.0 Boulders Slate boulders and 0.00823 gravels in silty matrix. 17.0 18.0 Boulders Slate boulders and 0.00852 gravels in silty matrix. 18.0 19.0 Boulders Slate boulders and 0.00828 gravels in silty matrix. 19.0 20.0 Boulders Slate boulders and 0.00742 gravels in silty matrix. SBH3 3437251.98 1243313.75 2657.9 25.0 0.0 1.0 Boulders Overburden mainly 0.00235 consist of boulders and gravels of granodiorite in silty matrix. 1.0 3.0 Boulders Overburden mainly 0.0692 consist of boulders and gravels of granodiorite in silty matrix. 3.0 4.0 Boulders Boulders of 0.064 granodiorite in silty matrix. 4.0 5.0 Boulders Boulders of 0.0246 granodiorite in silty matrix. 5.0 6.0 Boulders Boulders of 0.00933 granodiorite in silty matrix. 6.0 7.0 Boulders Boulders of 0.0107 granodiorite in silty matrix. 7.0 8.0 Boulders Boulders of 0.021 granodiorite in silty matrix. 8.0 9.0 Boulders Boulders of 100% granodiorite in silty Water matrix. Loss 9.0 10.0 Boulders Boulders of 0.0233 granodiorite in silty matrix. 10.0 11.0 Boulders Boulders of 0.0187 granodiorite in silty matrix. 11.0 12.0 Boulders Boulders of 0.00309 granodiorite in silty matrix. 12.0 13.0 Boulders Boulders of 0.00332 granodiorite in silty matrix.
245
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
13.0 14.0 Boulders Gravels and 0.00265 boulders of granodiorite in silty matrix. 14.0 15.0 Boulders Boulders and 100% gravels of Water granodiorite. Loss 15.0 16.0 Boulders Boulders of 100% granodiorite in silty Water matrix. Loss 16.0 17.0 Boulders Boulders of 100% granodiorite Water weathered along Loss joints. 17.0 18.0 Boulders Boulders of 100% granodiorite Water weathered along Loss joints. 18.0 19.0 Boulders Boulders of 100% granodiorite Water weathered along Loss joints. 19.0 20.0 Boulders Boulders of 0.00279 granodiorite weathered along joints. 20.0 21.0 Boulders Slate and hornfels 0.00235 boulders in silty matrix. 21.0 22.0 Boulders Slate and hornfels 0.0054 boulders in silty matrix. 22.0 23.0 Boulders Slate and hornfels 0.00245 boulders in silty matrix. 23.0 24.0 Boulders Boulders and 0.00573 gravels of slate, hornfels and granodiorite. 24.0 25.0 Boulders Boulders and 0.0095 gravels of slate, hornfels and granodiorite. SBH4 3437596.79 1243184.72 2633.5 20.0 0.0 1.0 Clay / Overburden clayey 0.00205 Silt gravely silt with boulders and gravels. 1.0 3.0 Clay / Overburden clayey 0.135 Silt gravely silt with boulders and gravels. 3.0 4.0 Gravels Gravels of slates 0.0157 and granodiorite. 4.0 5.0 Boulders Boulders and 4.50 6/25/2003 0.00933 gravels of granodiorite with some gravels of slate. 5.0 6.0 Boulders Boulders of 4.50 6/25/2003 0.0105 granodiorite. 6.0 7.0 Gravels Gravels of slate and 100% granodiorite. Water Loss 7.0 8.0 Gravels Gravels of slate and 4.50 6/25/2003 0.0179 hornfels.
246
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
8.0 9.0 Gravels Gravels of slate and 100% hornfels. Water Loss 9.0 10.0 Gravels Gravels of slate and 100% hornfels. Water Loss 10.0 11.0 Gravels Gravels of slate and 100% hornfels. Water Loss 11.0 11.2 Gravels Gravels of slate and hornfels. 11.2 12.0 Gravels Gravels of slate and 100% granodiorite. Water Loss 12.0 13.0 Gravels Gravels of slate and 4.80 7/11/2003 0.0825 some pieces of granodiorite. 13.0 14.0 Gravels Gravels of 100% granodiorite and Water hornfels. Loss 14.0 15.0 Gravels Gravels of 100% granodiorite and Water hornfels. Loss 15.0 16.0 Gravels Gravels of 100% granodiorite and Water hornfels. Loss 16.0 16.5 Gravels Gravels of granodiorite and hornfels. 16.5 17.0 Boulders Boulders and 4.50 7/15/2003 0.0592 gravels of hornfels and 2 pieces of granodiorite. 17.0 18.0 Boulders Boulders of 4.50 7/15/2003 0.085 hornfels. 18.0 19.0 Boulders Boulders of 4.80 7/16/2003 0.0653 hornfels. 19.0 20.0 Boulders Boulders and 5.00 7/17/2003 0.0556 gravels of hornfels with coarse grained sand. SPD5A 3437587.14 1243538.80 2671.7 80.0 0.0 5.0 Boulders Overburden consist of boulders and gravels with some traces of sand. 5.0 14.3 Boulders Slate and hornfels boulders and gravels in silty matrix. 14.3 16.5 Sand Fine to medium silty sand with gravels of slate. 16.5 17.3 Sand Coarse grained sand with gravels of slate. 17.3 18.0 Cobbles slates /horfels and pieces of granodiorite. 18.0 20.0 Boulders Slate /hornfels boulders and gravels in silty matrix. 20.0 21.0 Boulders Boulders of hornfels. 21.0 30.0 Boulders Boulders and gravels of slate and hornfels.
247
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
30.0 40.0 Boulders Boulders and gravels of slates and hornfels in silty matrix. 40.0 42.0 Boulders Gravels and 0.0036 boulders of slate and hornfels in clayey silty matrix with some and in slate boulders. 42.0 52.0 Boulders Gravels and boulders of slate and hornfels in clayey silty matrix with some and in slate boulders. 52.0 53.0 Sand Fine to medium 44.80 6/11/2003 0.01 sand of granodiorite and slates. 53.0 54.0 Boulders Piece of slate(7 cm), 44.80 6/12/2003 0.00548 remaining core is boulders and gravels of granodiorite. 54.0 55.0 Boulders Piece of slate(7 cm), 44.80 6/12/2003 0.123 remaining core is boulders and gravels of granodiorite. 55.0 56.0 Boulders Piece of slate(7 cm), 44.78 6/15/2003 0.0045 remaining core is boulders and gravels of granodiorite. 56.0 57.0 Boulders Piece of slate(7 cm), 44.80 6/15/2003 0.004139 remaining core is boulders and gravels of granodiorite. 57.0 58.0 Boulders Piece of slate(7 cm), 44.80 6/16/2003 0.003825 remaining core is boulders and gravels of granodiorite. 58.0 59.0 Boulders Boulders and 44.80 6/16/2003 0.00355 gravels of granodiorite in silty sand matrix. 59.0 60.0 Boulders Boulders and 44.80 6/16/2003 0.00415 gravels of granodiorite in silty sand matrix. 60.0 61.0 Boulders Boulders and 44.80 6/17/2003 0.00623 gravels of granodiorite in silty sand matrix. 61.0 62.0 Boulders Boulders and 44.80 6/17/2003 0.00403 gravels of granodiorite in silty sand matrix. 62.0 63.0 Boulders Boulders and 44.80 6/20/2003 0.00173 gravels of granodiorite in silty sand matrix.
248
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
63.0 64.0 Boulders Boulders and 44.78 6/21/2003 0.00229 gravels of granodiorite. 64.0 65.0 Boulders Boulders and 44.78 6/21/2003 0.00218 gravels of granodiorite. 65.0 66.0 Boulders Boulders and 44.75 6/21/2003 0.00297 gravels of granodiorite with some gravels of slates. 66.0 67.0 Boulders Boulders and 44.75 6/22/2003 0.00327 gravels of granodiorite with some gravels of slates. 67.0 68.0 Boulders Boulders and 44.80 6/22/2003 0.00244 gravels of granodiorite with some gravels of slates. 68.0 69.0 Boulders Boulders and 44.80 6/23/2003 0.00313 gravels of granodiorite with some gravels of slates. 69.0 70.0 Boulders Boulders and 44.80 6/26/2003 0.003 gravels of granodiorite with some gravels of slates. 70.0 71.0 Boulders boulders and 44.80 6/26/2003 0.00265 gravels of granodiorite. 71.0 72.0 Boulders Boulders and 44.80 6/27/2003 0.00278 gravels of granodiorite with some gravels of slates. 72.0 73.0 Boulders Boulders and 44.80 6/27/2003 0.00268 gravels of granodiorite. 73.0 74.0 Boulders Boulders and 44.80 6/28/2003 0.00237 gravels of granodiorite. 74.0 75.0 Sand Wash sample of 44.80 6/28/2003 0.00313 coarse sand. 75.0 76.0 Boulders Boulders and 44.75 6/29/2003 0.00303 gravels of granodiorite. 76.0 77.0 Sand Wash Sample 44.75 6/29/2003 0.00274 Medium to Coarse Sand 77.0 78.0 Gravels Gravels of 44.75 7/3/2003 0.000949 granodiorite. 78.0 79.0 Boulders Boulders and 44.75 7/3/2003 0.00147 gravels of granodiorite. 79.0 80.0 Gravels Gravels of 44.75 7/4/2003 0.00161 granodiorite. SB1 3437278.04 1243445.44 2667.0 3.0 0.0 0.9 Silt gravely silt 0.0112 0.9 1.8 Silt gravely silt 0.0521 1.8 2.7 Silt gravely silt 0.0309 0.0 3.0 Silt gravely silt SB2 3437278.04 1243366.33 2651.8 3.0 0.0 0.9 Silt gravely silt 0.00637 0.9 1.8 Sand silty sand 0.0738
249
Depth (m) K-Test Results
Description Date Bore / Pit / Trench Easting (m) (m) Northing (m) Elevation Total Start End Lithology GWL (m) (cm/sec) (Lugeons)
1.8 2.7 Sand silty sand 100% Water Loss 2.7 3.0 Sand silty sand SB3 3437275.48 1243286.88 2650.2 3.0 0.0 0.9 Silt gravely silt 0.0105 0.9 1.8 Boulders boulder of 100% granodiorite Water Loss 1.8 2.7 Boulders boulder of 100% granodiorite Water Loss 2.7 3.0 Boulders boulder of granodiorite SB4 3437357.62 1243445.41 2662.4 3.0 0.0 0.9 Gravels silty gravel 0.00668 0.9 1.8 Silt gravely silt 0.0314 1.8 2.7 Silt gravely silt 100% Water Loss 2.7 3.0 Silt gravely silt SB5 3437436.24 1243525.25 2660.9 3.0 0.0 0.9 Silt gravely silt 0.00376 0.9 1.8 Silt gravely silt 0.0112 1.8 2.7 Silt gravely silt 100% Water Loss 2.7 3.0 Silt gravely silt SB6 3437357.03 1243366.06 2643.5 3.0 0.0 0.6 Sand silty sand 0.6 0.9 Silt gravely silt 0.0037 0.9 1.8 Silt gravely silt 0.00965 1.8 2.7 Silt gravely silt 0.0229 2.7 3.0 Silt gravely silt SB7 3437515.80 1243366.14 2651.8 3.0 0.0 0.1 Silt clayey silt dry 0.1 0.9 Gravels silty gravel 100% Water Loss 0.9 1.8 Gravels silty gravel 100% Water Loss 1.8 2.7 Gravels silty gravel 100% Water Loss 2.7 3.0 Gravels silty gravel SB8 3437596.01 1243365.91 2636.5 3.0 0.0 0.3 Silt clayey silt dry with some gravels 0.3 0.6 Gravels sandy gravel 0.6 0.9 Gravels silty gravel 100% Water Loss 0.9 1.8 Gravels silty gravel 100% Water Loss 1.8 2.7 Gravels silty gravel 100% Water Loss 2.7 3.0 Gravels silty gravel SB9 3437651.99 1243431.95 2633.5 3.0 0.0 0.7 Clay / clayey silt with Silt gravels and sand 0.7 0.9 Silt gravely silt 100% Water Loss 0.9 1.8 Silt gravely silt 0.00317 1.8 2.7 Silt gravely silt 100% Water Loss 2.7 3.0 Silt gravely silt
250
ANNEXURE - II: SEEP/W INPUT FUNCTIONS
251
Water Content Functions
252
Hydraulic Conductivity Functions
253
254
VITA ZAHEER MUHAMMAD MALIK Candidate for the Degree of Doctor of Philosophy 2005-PhD-CEWRE-02 Dissertation Performance Evaluation of Selective Control Measures of Foundation Seepage for Embankment Dams over Permeable Strata. Major Field Water Resources Engineering Biographical Information Personal Information Date / Place of Birth: 23-06-1979 / Lahore Father’s Name: Dr. Ghulam Muhammad Malik National I.D. Card No.: 38401-0167585-1 Education M.Sc. Hydropower Engineering: 2004, Centre of Excellence in Water Resources Engineering, University of Engineering and Technology, Lahore. B.E. (Civil): 2001, Military College of Engineering, Risalpur, National University of Sciences and Technology. H.S.S.C.: 1996, Pakistan Embassy School, Jeddah, K.S.A. S.S.C: 1994, Pakistan Embassy School, Jeddah, K.S.A. Professional Experience Geotechnical Engineer, from January 2002 to-date, Pakistan Engineering Services (Pvt.) Ltd. Affiliations Pakistan Engineering Council (PEC Registration No.: CIVIL/22261), Pakistan Geotechnical Engineering Society. Publications Malik, Z. M., A. Tariq and J. Anwer, Seepage Control for Satpara Dam, Pakistan, in Proceedings of the Institute of Civil Engineers, Geotechnical Engineering, Volume 161, Issue GE5, October 2008, pp. 235-246. Gulrez, W., and Z. M. Malik, Upgrading Capacity at Mangla Dam – Alternative Resolutions, in Proceedings of the International Symposium on Dams in the Societies of the 21st Century, ICOLD-SPANCOLD, Barcelona, Spain, June 2006, Taylor & Francis, Berga et al (eds), 2006, Volume I, pp. 415-420. Gulrez, W., and Z. M. Malik, Hidden Troubles of Protection Dykes (A Planning Perspective), in Proceedings of International Symposium on Dam Safety and Detection of Hidden Troubles of Dams and Dikes, Xian, China, 1-3 Nov. 2005, p. 24. Anwer, J., M. Mushtaq Ch. and Z. M. Malik, “Varied Foundations of Chotiari Reservoir” in Proceedings of IrCOLD Symposium on Uncertainty Assessment in Dam Engineering, Tehran, Iran, 1-6 May, 2005, 019-S5, p. 67.
255