Utah State University

DigitalCommons@USU

All ECSTATIC Materials ECSTATIC Repository

Fall 1999

Water Resources Systems Analysis - Course Notes

A Bruce Bishop
Utah State University

Trevor Hughes
Utah State University

Mac McKee
Utah State University

Follow this and additional works at: https://digitalcommons.usu.edu/ecstatic_all

Part of the Civil Engineering Commons

Recommended Citation
Bishop, A Bruce; Hughes, Trevor; and McKee, Mac, "Water Resources Systems Analysis - Course Notes"
(1999). All ECSTATIC Materials. Paper 76.
https://digitalcommons.usu.edu/ecstatic_all/76

This Book is brought to you for free and open access by

the ECSTATIC Repository at DigitalCommons@USU. It
has been accepted for inclusion in All ECSTATIC
Materials by an authorized administrator of
DigitalCommons@USU. For more information, please
contact [email protected].
 CEE 6410

Water Resources Systems Analysis

Course Notes

by

A. Bruce Bishop

Trevor C. Hughes
and

Mac McKee

Fall, 1999

Civil and Environmental Engineering Department

Utah State University

 PREFACE

SUPPLEMENTAL READING

There are now many good systems analysis textbooks available, which focus on civil engineering
applications. One example is by Richard de Neufville (1990). It was not selected as a required
text for this class because most of the examples are not water resources problems. The chapters
on mathematical programming, economics, and philosophical basis for decision making, how-
ever, would make excellent supplemental reading to these course notes.

Two texts that do focus entirely on water applications are Loucks et al. (1980) and Mays and
Tung (1992). In previous years, the text by Loucks et al. (1980) has been recommended reading
for this initial course in water resources systems analysis, as well as the required text for other
classes. Unfortunately, it is now out of print. The text by Mays and Tung (1992) is therefore
suggested as supplemental reading. It and other reference materials will be placed on reserve in
the Merril Library. The lectures for this course, however, will be focused primarily upon these
course notes.

Other texts with civil engineering applications are Stark and Nicholls (1972) and Jewell (1986).
Students interested in more rigorous development of mathematical theory that supports the opti-
mization algorithms are referred to Hadley (1962 and 1964), Wagner (1975), Ossenbruggen
(1984), and Beightler et al. (1979).

BEWARE THE PITFALLS

Because the focus in these course notes is primarily upon optimization algorithms and optimiza-
tion model applications to water resources problems, an important caveat is needed. A student
who takes only this initial class in water resources systems analysis will be in danger of obtain-
ing an exaggerated perception of both:

• the importance of optimization relative to other analysis tools in the water resource planning
arena

• the credence one should place on the validity of an answer one gets from a single solution to
an optimization problem

While optimization is an extremely powerful mathematical tool for water applications, it is usu-
ally used as a screening device, the objective of which is to greatly reduce the number of alterna-
tives which should be considered before making a final decision by subsequent analyses (either
with or without a formal, mathematical).

For any complex water planning problem, a common sequence in the decision making process is
that the results from an optimization model will either become input to a more detailed simula-
tion model, or the optimization model will be revised and run iteratively to observe the effect of
changes in the structure of the objective function or constraints. The point is that a reality check
is always needed to determine:

How well does the system you modeled match the system you hoped you were modeling? If the
solutions differ from intuitively reasonable solutions, chances are the computer did only what it
was told to do, rather than what you hoped it would do.

How well does the solution from any model match the goals of the system planner? The man-
ager of a water system is often more interested in risk aversion and robustness of a solution than
in the rigorous maximization of a mathematical function which can never precisely match the
real array of (probably multiple) objectives.

In short, an optimization model is only one of several tools in our array of water resource plan-
ning approaches; a solution to an optimization problem usually represents only one piece of
information of value in the planning process.

ii
 Table of Contents

Section Page

PREFACE i

List of Tables xi

List of Figures xiv

1. WATER SYSTEM OPTIMIZATION: CONCEPTS AND METHODS 1

1.1 SYSTEMS DEFINITIONS 1

1.2 WATER RESOURCES SYSTEMS DESCRIPTIONS 3

1.3 MATHEMATICAL MODELS OF SYSTEMS:

OVERVIEW AND CONCEPTS 3

1.4 A GENERAL MODEL OF SYSTEM OPTIMIZATION 10

1.4.1 System Design and Optimization 10

1.4.2 The Design Function 11
1.4.3 Evaluation Models and Design Optimization 13
1.4.4 Summary 18

1.5 CLASSICAL OPTIMIZATION--A REVIEW 18

1.6 MODEL DEVELOPMENT APPROACH 21

1.8 PROBLEMS 23

2. LINEAR PROGRAMMING (LP)--INTRODUCTION

AND GRAPHICAL SOLUTIONS 25

2.1 INTRODUCTION 25

2.1.1 General Linear Programming Model 25

2.1.2 Development of Linear-Programming Equations 27

2.2 GRAPHICAL SOLUTION--AN EXAMPLE 30

2.3 LP PROBLEM FORMULATION 32

2.4 PROBLEMS 34

iii
 Table of Contents (continued)
Section Page

3. LP ANALYTIC SOLUTIONS--SIMPLEX METHOD 39

3.1 SOLUTION CONCEPT FOR LP PROBLEMS 39

3.2 LP DEFINITIONS AND THEOREMS 41

3.3 BASIC THEOREMS OF LP 43

3.4 SIMPLEX ALGORITHM 43

3.5 "GREATER-THAN" CONSTRAINTS 48

3.6 SHADOW PRICES 50

3.7 CANONICAL FORM 51

3.8 PROBLEMS 52

4. LP APPLICATIONS AND COMPUTER SOLUTIONS

TO SIMPLE WATER PROBLEMS 53

4.1 COMPUTER SOLUTION ALGORITHM FOR LP PROBLEMS 53

4.2 RESOURCE ALLOCATION PROBLEM 53

4.3 A WATER QUALITY WASTEWATER TREATMENT PROBLEM 54

4.4 RESERVOIR DESIGN AND OPERATION PROBLEMS 56

4.5 EXAMPLE RESERVOIR OPERATION PROBLEM 57

4.6 SYSTEMS ANALYSIS OF CONVEYANCE SYSTEMS 59

4.7 PROBLEMS 62

5. DUAL LP, SOLUTION INTERPRETATION, AND POST-OPTIMALITY 65

5.1 DUALITY 65

5.2 AN EXAMPLE OF THE PRIMAL-DUAL RELATIONSHIP 66

5.3 LP SOLUTION INTERPRETATIONS 69

iv
 Table of Contents (continued)
Section Page

5.4 SENSITIVITY ANALYSIS OF OBJECTIVE COEFFICIENTS (Cj) 71

5.5 PARAMETRIC PROGRAMMING 71

5.5.1 Changes in Objective Function Coefficients, Cj 72

5.5.2 Changes in Resource Limits, Bi 72
5.5.3 Strategy of Post-Optimality Analysis 73

5.6 PROBLEMS 74

6. INTEGER PROGRAMMING 77

6.1 BRANCH/BOUND TYPE ALGORITHM 78

6.2 BRANCH-BOUND SEARCH EXAMPLE 79

6.3 WATER SUPPLY PROBLEM 79

7. MORE APPLICATIONS OF INTEGER PROGRAMMING 85

7.1 WATER QUALITY PROBLEM 85

7.2 IRRIGATION EXAMPLE 87

7.3 INVESTMENT TIMING PROBLEM 89

7.4 PROBLEMS 91

8. PIECEWISE LINEARIZATION 93

8.1 INTRODUCTION 93

8.2 UNBOUNDED FUNCTION APPROACH 93

8.4 EXAMPLE PROBLEMS 98

8.4.1 A Simple LP Problem 98

8.4.2 Two-Reservoir Example Problem 99

8.5 SEPARABILITY 103

8.6 NON-LINEAR OPTIMIZATION SOFTWARE 104

v
 Table of Contents (continued)
Section Page

8.7 SIMULATION VERSUS OPTIMIZATION 104

8.8 PROBLEMS 106

9. OTHER TYPES OF LP MODELS 107

9.1 CRITICAL PATH METHOD 107

9.2 LINEAR REGRESSION 107

10. DYNAMIC PROGRAMMING 111

10.1 INTRODUCTION 111

10.2 DP CONCEPTS 111

10.3 A HYDROPOWER DESIGN PROBLEM 112

10.4 PROBLEMS 114

11. GRADIENT-BASED NONLINEAR OPTIMIZATION METHODS 115

11.1 INTRODUCTION TO NONLINEAR PROBLEMS 115

11.1.1 Convex and Concave Functions 115

11.1.2 Kuhn-Tucker Conditions 116
11.1.3 Example of the Kuhn-Tucker Method 120
11.1.4 Geometric Interpretation of the Kuhn-Tucker Conditions 120

11.2 GRADIENT-BASED OPTIMIZATION 123

11.2.1 Overview 123

11.2.2 Common Gradient-Based Methods 123
11.2.3 An Example of the Method of Steepest Ascent 123

11.3 PROBLEMS 131

12. GENETIC ALGORITHMS FOR SOLUTION OF

NONLINEAR OPTIMIZATION PROBLEMS 133

12.1 BACKGROUND 133

12.2 GENETIC ALGORITHMS AND GENETIC PROGRAMMING 133

vi
 Table of Contents (continued)
Section Page

12.2.1 Biological Background 134

12.2.2 Relationship of GP to Optimization 135
12.2.3 A Basic Genetic Algorithm 135
12.2.4 Illustration of Key Steps 137
12.2.5 Parameters of GA 141

12.3 EXAMPLE GA PROBLEM 142

12.4 PROBLEMS 144

13. TYPES AND USES OF RESERVOIR MODELS 145

13.1 INTRODUCTION 145

13.2 DESIGN MODELS 146

13.2.1 Identifying Limiting Resources 146

13.2.3 Integrated River Basin Planning (IRBP) Steps 146
13.2.4 Introduction to Yield Models 147

14. MULTIPLE RESERVOIR YIELD 149

14.1 BACKGROUND AND SCOPE 149

14.2 PRE-COMPUTER APPROACHES 150

14.2.1 Rippl Diagram 150

14.2.2 The Sequent Peak Method 150

14. 3 COMPUTER SIMULATION 151

14.3.1 Overview 151

14.3.2 Simulation Model Logic 154

14.4 MULTIPLE RESERVOIRS 154

14.4.1 Possible Reservoir Configurations and Yield Calculations 154

14.4.2 Series Reservoirs without Intervening Streamflows 156

14.5 SYSTEMS ENGINEERING DISCUSSION 156

14.6 PROBLEMS 158

vii
 Table of Contents (continued)
Section Page

15. YIELD/STORAGE ANALYSIS FOR MULTIPLE RESERVOIRS 160

15.1 REVIEW OF LP MODEL CONCEPTS 160

15.1.1 Yield Analysis by LP 161

15.1.2 Example Problem 162

15.2 SYSTEMS ENGINEERING DISCUSSION 163

15.3 FORESIGHT WITH LP MODELS 164

15.4 PROBLEMS 166

16. COST FUNCTIONS FOR RESERVOIR PROJECTS 167

16.1 NEED FOR CONTINUOUS COST FUNCTIONS 167

16.2 DERIVATION OF PRE-FEASIBILITY LEVEL COST FUNCTIONS 167

16.2.1 Cost of Embankment 168

16.2.2 Cost of Spillway 168
16.2.3 Cost of Outlet Structure 168

16.3 ALTERNATIVE METHODS OF PRELIMINARY

RESERVOIR COST ESTIMATION 169

16.4 STRUCTURE OF ECONOMIC OBJECTIVES 169

16.5 MULTIPLE RESERVOIR YIELD/COST CONCEPTS 170

16.6 DYNAMIC PROGRAMMING GENERAL CONCEPT REVIEW 170

16.6.1 DDP for Sizing Multiple Reservoirs 171

16.6.2 Continuation of Example Problem:
Reservoir Cost Analysis by DP 172

16.7 PROBLEMS 174

17. INVESTMENT TIMING FOR IRBP 175

17.1 INTRODUCTION 175

17.2 INVESTMENT TIMING PROBLEM MODELS 175

viii
 Table of Contents (continued)
Section Page

17.3 INVESTMENT TIMING EXAMPLE 176

17.4 CONSIDERATION OF MULTIPLE RESERVOIR SIZES AT A SITE 177

17.5 PROBLEMS 179

18. MULTIPLE OBJECTIVE PLANNING 181

18.1 INTRODUCTION 181

18.2 SYSTEMS ENGINEERING CONSIDERATIONS 184

18.3 INCLUDING GROUNDWATER IN THE IRBP ANALYSIS 185

18.4 SELECTION THE OBJECTIVES 186

18.5 CONSTRUCTION DECISIONS VERSUS OPERATING POLICY 186

19. MATHEMATICAL BASIS FOR MOP 189

19.1 IDENTIFYING BAD SOLUTIONS 189

19.2 SUMMARY OF MATHEMATICAL LOGIC FOR MOP 189

19.3 EXAMPLE MOP PROBLEM 191

19.4 SELECTING A DECISION FROM THE NON-DOMINATED SET 192

19.5 PROBLEMS 193

20. GROUNDWATER IN IRBP 195

20.1 INTRODUCTION 195

20.2 CONJUNCTIVE USE 195

20.3 TYPES OF GROUNDWATER RESOURCES 196

20.4 SOIL MOISTURE 196

20.5 AQUIFER MODEL 199

20.6 EFFECT ON SURFACE RESERVOIR MODEL 199

ix
 Table of Contents (continued)
Section Page

21. RISK ANALYSIS 201

21.1 BACKGROUND 201

21.2 HYDROLOGIC RISK 201

21.3 SENSITIVITY TO SEASONAL USE PATTERN 201

21.4 SENSITIVITY TO EVAPORATION LOSS ASSUMPTIONS 206

21.5 PROBLEMS 207

REFERENCES 209

APPENDIX 1: INTRODUCTORY NOTES ON LINGO 211

APPENDIX 2: NOTES ON Mac’sGA 2

APPENDIX 3: LINGO MODELS FROM SECTION 15 2

x
 List of Tables

Table Page

1.1 Elements of Water Resources Systems 6

1.2 An Overview of Approaches to Systems Modeling 8
1.3 Types of Models in “Modeling Space” 9

3.1 Basic Solutions to the Example Problem 41

3.2 Illustration of Gauss-Jordan Pivoting 44
3.3 Illustration of Simplex Tableau Calculations 47
3.4 Simplex Calculations for a “Greater-Than” Constraint 49

4.1 Example Farm Problem Information 53

4.2 Example Wastewater Treatment Problem Data 55
4.3 Data for the Example Reservoir Irrigation Supply Problem 58
4.4 Pipeline Network Elevation and Demand Data 60
4.5 Hydropower and Irrigation Problem Data 63

5.1 Example of Primal and Dual Solutions from LINGO 68

7.1 BOD Waste Loading and Cost Data 85

7.2 Water Demand and Hydrologic Data 87
7.3 Facility Capacities and Costs 88
7.4 Project Scheduling Problem Data 89
7.5 Seasonal Flow and Demand 91
7.6 Wasteloads, Flows, Treatment Costs, and Water Quality Constraints 92

8.1 Calculation of Coefficients for Use in Piecewise Linearization 98

8.2 Piecewise Linear LP Model Coefficients 99
8.3 Summary of Constraints for the Two-Reservoir Problem Model 102
8.4 Calculation of Coefficients for Piecewise Linearization of the Two-Reservoir
Hydropower and Irrigation Problem 103
8.5 Reservoir Capacity and Capital Cost Data 106

9.1 Example Data for Regression 108

10.1 Hydropower Benefits of Alternative Dam Sites 113

10.2 Stage 3 Tableau 113
10.3 Stage 2 Tableau 113
10.4 Stage 1 Tableau 113

11.1 Distance in the Direction of vPo from Point Po to Each Constraint 127

12.1 Terminology Used in GA and Corresponding Synonyms from Systems Analysis 136
12.2 Genetic Algorithm Parameters Used in Solving Example Problem 142

xi
 List of Tables (continued)
Table Page

14.1 Inflow Data for Problem 1 158

15.1 Example Streamflow Data 163

16.1 Example Reservoir Cost Data 172

17.1 Example Demand Data 177

17.2 Reservoir Cost, Yield, and Storage Data 177
17.3 Problem Demand Data 179
17.4 Project Cost, Yield, and Storage Data 179

19.1 Definition of Solutions in Objective Space 192

21.1 Calculation of Composite Demand Fractions 205

21.2 Reservoir Yield and Demand Shifts 206
21.3 Variation in Evaporation 208

xii
 List of Figures

Figure Page

1.1 Representation of a System 1

1.2 Detailed Representation of a System 3
1.3 Levels of Specificity in Water Resources Systems Modeling 4
1.4 Hierarchy of Systems and Systems Functions 5
1.5 Modeling Space (Cube) 7
1.6 Model of Systems Design and Optimization 10
1.7 Example of Design Function 13
1.8 System Performance and Optimization 14
1.9 Cost-Effectiveness Curves of Reservoir Sediment Control Measures 15
1.10 Joint Optimization of Multiple Outputs 17
1.11 Possible Behaviors of a Nonlinear Function 21

2.1 Graphical Solution of a Linear Programming Problem 27

2.2 Graphical Solution of a Production Problem with Constraints on Resources 32

3.1 Graphical Illustration of Example Problem 41

3.2 Examples of Convex Sets 42
3.3 Examples of Nonconvex Sets 42
3.2 Graphical Solution to Example Problem 46
3.3 A Linear Programming Problem with a “Greater-Than” Constraint 50

4.1 Schematic Showing Location of Cities Discharging

Wastewater into A River System 55
4.2 Example Reservoir Irrigation Supply Problem 57
4.3 Example Pipeline Network Problem 60
4.4 Schematic of a Hydropower and Irrigation Supply Problem 62

5.1 Illustration of LP Solution Sensitivity to Uncertainty in Resource Availability 70

6.1 Relation between Project Cost and Quantity of Water Produced 77

7.1 Location of Five Cities Discharging Treated Wastewater 85

7.2 Irrigation Supply from Reservoir and Well Water 87
7.3 MIP and Irrigation Supply from a Reservoir and/or River Diversion 91
7.4 Wastewater Treatment and Water Quality Control Problem 92

8.1 Typical Non-linear (Concave) Cost Function 93

8.2 Convex Function 94
8.3 Unbounded Approach to Piecewise Linearization 94
8.4 Bounded Variable Approach 96
8.5 Restricted Basis Entry 97
8.6 Two-Reservoir Irrigation and Power Problem 100

xiii
 List of Figures (continued)
Figure Page

9.1 Illustration of a CPM Problem 108

9.2 Linear Regression Using LP 109

11.1 A Convex Function of Two Variables 116

11.2 Lagrangian Solution to an Optimization Problem with an Equality Constraint 121
11.3 Lagrangian Solution of a Problem with Multiple Nonlinear Equality Constraints 122
11.4 Kuhn-Tucker Solution of a Problem with Multiple
Nonlinear Inequality Constraints 122
11.5 Steepest Ascent Method Search 126
11.6 A Simple Groundwater Optimization Problem 132

12.1 A Simple Representation of the Relationship between

an Organism’s Genotype and Phenotype 134
12.2 Alternative Chromosome Encoding Methods 137
12.3 Evaluation of “Fitness” for a Water Resources Problem 138
12.4 The Process of Crossover to Produce a New Offspring 139
12.5 Value Encoding Crossover 139
12.6 The Process of Mutation 140
12.7 Value Encoding Mutation 141
12.8 GA Convergence in Solution of a Simple Quadratic Problem 143

13.1 Typical Reservoir Yield-Capacity Curves 147

13.2 Typical Reservoir Cost-Capacity Curves 148

14.1 Typical Rippl Diagram 150

14.2 A Simple Reservoir with Inflows and Releases 152
14.3 Surface Area versus Storage Function 153
14.4 Linearized Surface Area Function 153
14.5 Yield/Storage Function 154
14.6 Possible Reservoir Configurations 155

15.1 Example Reservoir Configuration 163

16.1 Reservoir Cost Functions 167

16.2 Example Reservoir Cost Functions 172

17.1 Demand Function 177

18.1 A Two-Objective Tradeoff Result 182

18.2 A Three-Objective Problem Result 182

19.1 Non-Dominated Solutions 189

19.2 A Two-Objective Problem in Objective Space 190

xiv
 List of Figures (continued)
Figure Page

19.3 A Simple Two-Objective MOP Problem 193

20.1 Conjunctive Use Schematic 197

20.2 Conceptual Model for Conjunctive Use 197

21.1 Lower Iowa River Weekly Flood Damages

Related to Coralville Reservoir Operation 202
21.2 Trade-off Relationships between Flood Control
on Coralville Reservoir for 10 Inflow Sequences 203
21.3 Estimated Probability Quantiles for Trade-offs between
Flood Control and Recreation on Coralville Reservoir 204

xv