You are on page 1of 84

PERFORMANCE OF WATER SUPPLY OPERATIONS MEASURED BY

RELIABILITY AND MARGINAL COST

by

Jason Russell Lillywhite

A thesis submitted to the faculty of


The University of Utah
in partial fulfillment of the requirement for the degree of

Master of Science

in

Civil and Environmental Engineering

Department of Civil and Environmental Engineering

The University of Utah

March 2008
i

Copyright © Jason Russell Lillywhite 2008

All Rights Reserved


ii

ABSTRACT

This study develops and applies a concept of combining marginal cost and

reliability in an operational water supply model. This concept can be used to gauge the

performance of various water supply strategies. Traditional methods of water supply

planning typically capture capital costs of new supplies but do not necessarily capture

measurements of reliability and operational efficiencies. Reliability and efficiency can

significantly impact performance of producing and delivering water. Rapid population

growth, climate change, extended droughts, and increasing public scrutiny are all reasons

why it is becoming more important for water supply planners to develop strategies that

provide reliable and cost-efficient solutions to the public. Based on a concept previously

developed, this study uses an approach of assessing reliability of water supply and

marginal costs by incorporating both supply and demand-side management options.

Risk-based reliability of the system is estimated as a function of shortages in flow rate

and system storage volumes. The new approach is applied to a water supply planning

model for the Washington County Water Conservancy District (District), a regional water

wholesaler located in St. George, Utah. The results of this study show that increased

operational efficiencies can be found while maintaining higher reliability in the system.

The results also show that this approach can provide better insight into timing of large

future supply acquisitions.


iii

TABLE OF CONTENTS
iv

ABSTRACT..................................................................................................................ii
Fig. 1. Hypothetical Water Supply Plan Chart. 9........................................................vii
Fig. 2. Flow of logic in a hypothetical Supply Model 15............................................vii
Fig. 4. Proposed Relationship Between Water Conservation and Non-Essential
Demands 21.................................................................................................................vii
Fig. 5. Visual Representation of the Efficiency-Reliability Modeling Framework 23vii
Fig. 6. Plot of historic streamflow in the Virgin River at the Virgin Gauge 29..........vii
Fig. 7. Exceedence Probability Plot for Annual Virgin River Flow Volumes 30........vii
Fig. 8. Schematic representation of the District’s water supply system 31.................vii
Fig. 9. Exceedence Probability of Annual Precipitation in St. George (USU, 2007) 33
.....................................................................................................................................vii
Fig. 10. Mass Balance Check Plot for the Model 38...................................................vii
Fig. 11. Screen Capture of the model 39.....................................................................vii
Fig. 12. Population growth used in the supply model 42............................................vii
Fig. 13. Supply Reliability Using WTP Peaking Compared to Well Peaking 45........vii
Fig. 14. Average Annual Additions to Supply Capacity from the New Method 46....vii
Fig. 15. Single Trace of Annual Additions to Supply Capacity from the New Method
47.................................................................................................................................vii
Fig. 16. Annual Additions to Supply Capacity from the Traditional Method 48.........vii
Fig. 17. Average Marginal Cost Results for the New Method and the Traditional
Method 49....................................................................................................................vii
Fig. 18. Reliability plot for the new method verses traditional method 50.................vii
Fig. 19. Plot of Average Marginal Cost compared to Reliability for the New Method
50.................................................................................................................................vii
Fig. 20. Plot of Marginal Cost compared to Reliability for the Traditional Method 51
.....................................................................................................................................vii
Fig. A-1. Comparison of plot for the Virgin River diversion 57................................vii
Fig. A-2. Comparison plot for the Sand Hollow Transfer Pump Station 59................vii
Fig. A-3. Comparison plot for the Sand Hollow Transfer Pump Station pumping costs.
59.................................................................................................................................vii
Fig. A-4. Comparison plot for the St. George City well pumps 60............................vii
Fig. A-5. Comparison plot for the cumulative flow at the St. George City well pumps
61.................................................................................................................................vii
v

Fig. A-6. Comparison plot for the Quail Creek WTP 61.............................................vii
Fig. A-8. Comparison plot for the Sand Hollow Reservoir 63...................................vii
Fig. B-1. Development of the Marginal Cost Curve for the New Method 65............viii
Fig. B-2. Development of the Marginal Cost Curve for the Traditional Method 66..viii
Fig. B-3. Additional SMAs in the New Method 67....................................................viii
Note: SMAs on layering the top are all conservation (demand based) SMAs 67......viii
Fig. B-4. Additional SMAs in the Traditional Method 68..........................................viii
ACKNOWLEDGMENTS............................................................................................ix
CHAPTER 1 - INTRODUCTION.................................................................................1
Background..............................................................................................................1
The Reason for Marginal Cost.................................................................................3
The Reason for Reliability.......................................................................................3
Reason for Using Uncertainty and Variability.........................................................4
Effects of Water Conservation.................................................................................5
Introduction to the Method......................................................................................6
CHAPTER 2 – METHOD.............................................................................................7
Efficiency-Reliability Modeling Approach..................................................................10
Model Framework..................................................................................................11
CHAPTER 4 - CASE STUDY Application.................................................................25
District Background...............................................................................................25
Model Development...............................................................................................34
Simulation Scenarios.............................................................................................37
CHAPTER 4 – CASE STUDY Analysis RESULTS...................................................44
CHAPTER 5 – CONCLUSIONS................................................................................54
Model Limitations..................................................................................................55
APPENDIX A – MODEL CALIBRATION.................................................................56
APPENDIX B – DETAILED Cost Curves..................................................................64
REFERENCES............................................................................................................69
vi

LIST OF FIGURES
Figure Page
vii

Fig. 1. Hypothetical Water Supply Plan Chart..............................................................9


Fig. 2. Flow of logic in a hypothetical Supply Model.................................................15
Fig. 4. Proposed Relationship Between Water Conservation and Non-Essential
Demands......................................................................................................................21
Fig. 5. Visual Representation of the Efficiency-Reliability Modeling Framework.....23
Fig. 6. Plot of historic streamflow in the Virgin River at the Virgin Gauge...............29
Fig. 7. Exceedence Probability Plot for Annual Virgin River Flow Volumes..............30
Fig. 8. Schematic representation of the District’s water supply system......................31
Fig. 9. Exceedence Probability of Annual Precipitation in St. George (USU, 2007). .33
Fig. 10. Mass Balance Check Plot for the Model........................................................38
Fig. 11. Screen Capture of the model...........................................................................38
Fig. 12. Population growth used in the supply model..................................................42
Fig. 13. Supply Reliability Using WTP Peaking Compared to Well Peaking.............45
Fig. 14. Average Annual Additions to Supply Capacity from the New Method..........46
Fig. 15. Single Trace of Annual Additions to Supply Capacity from the New Method
......................................................................................................................................47
Fig. 16. Annual Additions to Supply Capacity from the Traditional Method..............48
Fig. 17. Average Marginal Cost Results for the New Method and the Traditional
Method.........................................................................................................................49
Fig. 18. Reliability plot for the new method verses traditional method......................50
Fig. 19. Plot of Average Marginal Cost compared to Reliability for the New Method
......................................................................................................................................50
Fig. 20. Plot of Marginal Cost compared to Reliability for the Traditional Method...51
Fig. A-1. Comparison of plot for the Virgin River diversion ......................................57
Fig. A-2. Comparison plot for the Sand Hollow Transfer Pump Station.....................59
Fig. A-3. Comparison plot for the Sand Hollow Transfer Pump Station pumping costs.
......................................................................................................................................59
Fig. A-4. Comparison plot for the St. George City well pumps .................................60
Fig. A-5. Comparison plot for the cumulative flow at the St. George City well pumps
......................................................................................................................................61
Fig. A-6. Comparison plot for the Quail Creek WTP..................................................61
Fig. A-8. Comparison plot for the Sand Hollow Reservoir ........................................63
viii

Fig. B-1. Development of the Marginal Cost Curve for the New Method..................65
Fig. B-2. Development of the Marginal Cost Curve for the Traditional Method........66
Fig. B-3. Additional SMAs in the New Method..........................................................67
Note: SMAs on layering the top are all conservation (demand based) SMAs............67
Fig. B-4. Additional SMAs in the Traditional Method................................................68
ix

ACKNOWLEDGMENTS

I acknowledge the generous contributions of Steve Burian, Christine Pomeroy,

David Kiefer, and Muzaffar Eusuff and for their guidance in bringing this thesis to

completion. I also acknowledge Corey Cram, hydrologic coordinator and Melodie

Sorenson with the Washington County Water Conservancy District, and Scott Taylor,

City Engineer with St. George City for their assistance.


1

CHAPTER 1 - INTRODUCTION

Background
Populations in arid regions of the world are still growing at a rapid pace while

water supplies are limited, creating significant water scarcity issues. In the last century,

the world’s water use rate has been growing at more than twice the population growth

(UN-Water, 2007). Water supply sources will become less obtainable and people

increasingly challenge decisions to build new large water supply projects either in fear of

this causing more population growth, or detriment to the environment (Barakat, 1994).

As population grows and the natural resources around us become more of a shared asset,

efficiency of use and reliability of supplies become more important. Often times

however, reliability and efficiency are found to be in conflict (Mondal, 2007).

For the purposes of this study, reliability is defined as the probability that a

system will operate successfully for a specified period of time, under specified

conditions, when used for the manner and purpose for which it was intended

(Ramakumar, 1996). For the purposes of this study, efficiency is defined as a function of

marginal costs, which is defined as the incremental cost of an additional unit of output or

the cost of the additional inputs needed to produce that output. A very reliable water

supply system always costs more to operate than a less reliable system, so it is important

to maintain a balance between keeping costs down and being reliable.


2

The mission of water suppliers generally is multi-faceted; well stated by Contra

Costa Water District of California as:

“[To] strategically provide a reliable supply of high quality water at the lowest
cost possible, in an environmentally responsible manner.”(CCWD, 2007)

In order to continue serving the public with reliable, high quality water at low cost in the

face of increasing challenges, it will become more important to try and find what is

considered a balanced combination of reliability and efficiency.

The typical method of water supply planning tends to be less focused on

efficiency and reliability of a water supply portfolio over the long term and more focused

on the technical design aspects of given supply options. Examples of some typical

methods of supply planning can be found in many water master planning documents

developed for various water districts, cities, and water supply wholesalers (Water

Resources Master Plans: CDM, 2007, SNWA, 2006, and TC&B, 2004). Focusing on

constructability and engineering soundness is vital for any proposed water supply

infrastructure, but this thesis suggests that more effort should be placed on finding ways

to increase efficiency and reliability of the long-range future of the system.

Typical water resources master plans tend to base future reliable supplies on

perpetual, historic low flows and drought periods. These historic low flows are used as a

recurring annual supply for the future. While this is a conservative approach, it is

possible that it may not be the most economic or realistic approach because hydrology

periodically cycles between wet and dry. Water suppliers can make use of the water

available in normal to wet years if they have storage facilities or other means to do so.
3

The Reason for Marginal Cost


Economics also plays a major role in U.S. water supply decision making policies

and standards, which were formally established in the 1965 Water Resources Planning

Act (Sander, 1983). This act instituted a multi-objective planning approach that

incorporates cost-benefit analysis in the evaluation of future water resources projects.

Historically, water utilities have relied on historic accounting data to develop economic

strategies rather than forward-looking marginal costs (Turvey, 1976). Incorporating

marginal costs in the decision making and rate design of water utilities is a unique and

cost-effective method of increasing efficiencies (Hall, 2006). As a function of variable

operating (or production) costs, marginal costs provide a measure of economic

performance. Marginal cost quantifies the rate at which these variable costs increase,

which allows the planner to find ways to decrease the rate of variable cost increases and

thus increase efficiency of the system. Marginal cost shows the impact of changes in

operations as a function of cost, which is directly related to increases in efficiencies.

The Reason for Reliability


Reliability is an important measure of water supply performance because it shows

the behavior of proposed supply management strategies under multiple uncertain

scenarios (Mondal, 2007). Within the scope of this thesis, the word reliability is used as a

general term that encompasses three aspects of measuring reliability, which include

reliability, resiliency, and vulnerability. These are defined by Mondal (2007) as follows:

• Reliability: the frequency of occurrence of water shortage

• Resiliency: the length of time needed to recover from water shortage

• Vulnerability: the maximum severity of the water shortage


4

These terms are clarified in the following example: In a given city, water

demands are not completely met once for a day in July and then again for a whole week

in August. The shortage in July was found to be equal to 5 cfs and the shortage in August

peaked at 1 cfs. The reliability for the year would be 2 because water shortage occurred

twice. The Resiliency would be 7 days because it is assumed that the time-span of the

longest occurrence is equal to the time required to get the system back into conformance.

The vulnerability would be 5 cfs because this is the highest peak shortage flow that

occurred.

The value of water supply reliability was estimated by Barakat et al (1994) as a

measure of public acceptance using a survey conducted by the California Urban Water

Agencies in 1994. The results of the survey show that people are willing to pay

substantial amounts of money in addition to regular water bills in order to avoid even

minor water shortages (Barakat, 1994). Since reliability is very important to the public,

this should be an integral part of the water supply planning process. It has been found

that enhancing and enlarging water supply portfolios even when not the most economical

option provides a strong argument to developing water supplies before the need

completely develops rather than deferring projects because these new supplies strengthen

reliability of the supply (Geriani, 1998).

Reason for Using Uncertainty and Variability


Variability in supplies is defined here as the recorded or synthesized change in

water source supplies through time. The reason to use variability in water supply

planning is because decisions should be made in a way that would allow the water

supplier to take advantage of the wetter to more normal periods in a way to hedge against
5

shortage in the dry years (UDWR, 2001). Uncertainty parameters can be used to estimate

reliability of the system. Uncertainty can be applied to water demands, hydrology, and

system operational parameters.

Effects of Water Conservation


Under pressure of changing policies due to environmental concerns and risk of

drought, decision makers often turn to water conservation as a means to show compliance

and to reduce scarcity. Before forming planning strategies around a particular water

conservation goal however, the effects of water conservation on reliability and finances

should be considered along with ways that conservation can interact with supply

acquisition. Water conservation is becoming commonplace in a utility’s water supply

portfolio, but implications of water conservation are far from understood (Chesnutt,

1995). Utilities have faced reductions in revenues as conservation programs are put into

place. Supply reliability can also be affected by water conservation (Thompson, 2007).

Demand hardening is defined as decreasing ability to reduce demands as more un-

necessary uses are eliminated and people begin to lose interest. Research has found

evidence that in many cases, the effectiveness of some water conservation measures may

decay somewhat over time after implementation (Little, 2002). Thus, conservation cannot

be treated as a one-time investment. Another complication in managing water

conservation stems from the public’s perception of the necessity to conserve, induced by

nightly news reports on TV stating wetter than normal conditions or reports on high

snowpack levels (Ellis, 2007).


6

Introduction to the Method


The method introduced in this study will combine measures of efficiency

(marginal cost) and reliability to help better understand the performance of given water

supply strategies. The basic concept of the proposed method is to simulate a water supply

system, capturing marginal costs associated with water production and delivery and

calculating shortages of water supplies. This method is referred to as the “efficiency-

reliability” modeling approach.

Marginal cost analysis will help the water supply planner visualize the financial

impacts of multiple long-term strategies. Reliability analysis can be used as a tool to

justify acquisition of new supplies or increased conservation efforts and will also help the

decision maker understand the risk associated with their planning strategy. Variability of

supplies provides a way to estimate reliability without designing a strategy that is overly

conservative by using only supply data from the driest year on record.

This study aims at determining whether using marginal cost analysis combined

with reliability analysis will be an effective tool to assist in the water supply planning

process. It is anticipated that the results of this study would provide a way to make an

existing strategy more reliable and more efficient by suggesting a change to the timing of

new projects, current operational schemes, and conservation plans. This concept is

intended to show usefulness based on the performance of the system as a function of

marginal cost and reliability. These performance measures can be used to help the

decision makers in choosing the right time to increase capacity and aid them in proving

to the public why the time was chosen.


7

CHAPTER 2 – METHOD

Typical Water Supply Planning Strategies

Water supply planning involves carefully choosing the timing and magnitude of

current and future water supply options to provide reliable and cost effective water

supply to the public. The planning process is usually implemented in an organized

planning framework like those developed by King County Department of Natural

Resources and Parks (DNRP, 2005), South Florida Water Management District

(SFWMD, 2006), and Irvine Ranch Water District (IRWD, 2005). Typically, the planning

framework includes some form of the following steps:

1. Conduct a regional demand forecast


2. Assess current supply portfolio
3. Evaluate new water source options (i.e. reclaimed water, conservation)
4. Evaluate the reliability of supplies
5. Consider climate change and risk of drought
6. Consider needs to protect the environment
7. Consider legal aspects such as acquisition of water rights, understand regulatory
framework in which development decisions will take place
8. Identify funding opportunities

Future supplies are based on a need calculated using population forecasts with

associated estimates of per capita water demands. Population growth forecasts are often

associated with significant uncertainty and are therefore viewed as sets of growth

scenarios. Typically, a single trace of conservative demand growth is chosen to develop a

single-line population projection. Population growth estimates are typically developed by

governmental agencies or planning groups such as a state Governer’s Office of Planning


8

and Budget (UDWR, 2001). In addition to population growth estimates, historic water

use data is also used to project future demand growth using extrapolation techniques.

Once future demand growth scenarios are selected, current supply sources are

analyzed by the water supplier to determine their adequacy in terms of annual volume

available. Often times, the driest year on record or the 25-percent probability supply

volume is chosen as the adequate water supply to use for projections (Thompson, 2007).

The 25-percentile probability refers to the lowest 25% in annual flow volume if all the

years of data were ranked from wettest to driest. This is done to ensure conservative

estimates of supply need in the future.

Potential new sources are decided on and also evaluated for adequacy to be

included in the supply portfolio. The demand projection is overlaid on a time plot

showing existing and proposed new supplies as shown in Fig. 1, which is a plot showing

a hypothetical supply and demand projection. The new supplies are staggered to

represent time of implementation in a fashion that follows the increase in demands.

Additionally, a projection of demand with and without conservation measures is shown

to provide a range of possible demand scenarios. The plot shown in Fig. 1 is a gross

simplification of a typical planning procedure, and is not intended to describe the

methodology of any one entity, but it is patterned after that used in the Drought Year

Water Supply Plan by Jordan Valley Water Conservancy District (JVWCD, 2007).

This methodology captures the timing of additional supplies under what is considered a

conservative scenario based on historic hydrologic and demand information.


9

80
New Supply 3
70 Demand - no
New Supply 2
Conservation
Supply (1,000 Acre-Feet) New Supply 1
60
Existing Supply
50

40

30
Demand with
20
Conservation
10

0
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050

Year

Fig. 1. Hypothetical Water Supply Plan Chart.

Graphs like these provide a good overall view of possibilities and is a well-known

visual tool often used by water supply planners (see SNWA, 2006, JVWCD, 2007,

TC&B, 2004, LYR&B, 2005, or IRWD, 2005). However, some of the issues that may not

be adequately captured in such a representation include supply reliability, economic

efficiency, resilience of reservoirs pulling out of a drought cycle, and variations in timing

of demand reductions. In short, this representation does not show adequately how the

system will perform under multiple uncertain and controlled scenarios.

An example of this typical approach was used by LYR&B (2005) for a capital

facilities plan for Washington County Water Conservancy District. This traditional

approach uses the driest year on record to establish adequate and available supplies and

incrementally adds new supplies as demands increase. New supplies are added in order to

maintain a supply excess of at least 10% above demands at any given time. A linearly

interpolated conservation reduction factor is imposed on demands starting at 0% in 2007

and ending with 20% in 2025. It is not known why the proposed supplies need to exceed

demands by 10% except to provide added security. This approach does not provide any
10

quantifiable measure of security in the supply. Also, there is no measure of costs on the

basis of incremental production so that efficiencies can be found. The new proposed

methodology of incorporating efficiency and reliability will be used to see if the strategy

developed using the LYR&B approach is adequate based on reliability and efficiency.

By just using the LYR&B approach to develop a strategy, some important

information is missing. It is unknown what effects conservation will have on marginal

costs and reliability and it is unknown whether using worst-case driest year on record

will provide an adequate future supply outlook. It is also not understood how acquisition

of new supplies will affect operational efficiencies.

Efficiency-Reliability Modeling Approach

A water supply system can be characterized by many cause and effect

relationships, many allocation priority schedules, storage targets that change through

time, and capacities dependant on flows calculated on feedback loops. Capturing all of

this detail, along with volumes of input data by hand (or in real-time) could become

complicated and unmanageable. Computer simulation models are often developed to

represent water supply systems and facilities to provide insight into the planning and

decision making process. Typically, logical rules, decision parameters, theoretical

simplifications, and assumptions are implemented into models to approximatereality

(Drobak, 1972). A computer model can aid decision making by integrating the complex

components of the water supply planning process. Computer models can be used to

simulate real world functions that would take months or years to observe and can also be

used to run repeated trial and error experimentation with controlled variables (Drobak,

1972).
11

Distinction should be made between the types of decision support models referred

to in this study and more specific models targeting physically based hydraulic and

hydrologic phenomena. The decision based models used for this study are best for

simulation of water flows in the context of mass balance and decision process on time

scales while running time steps no smaller than daily or monthly. Typical software used

for this kind of model needs to be fundamentally flexible enough to allow for

development of custom made rules and operating logic that can incorporate nuances that

operators deal with on a regular basis. Tools such as pipe network flow models or open

channel flow simulators will present difficulties in developing the decision science that

goes into these kinds of models because they lack the flexibility. Some examples of

software that may be appropriate for the application presented in this study may include:

AnyLogic, GoldSim, Aescl Xtreme, Extend, PowerSim, Stella, and custom applications

built in computer scripting code.

Model Framework
The efficiency-reliability model framework consists of a combination of some of

the traditional elements discussed above (i.e. water demand growth, supply portfolio)

along with the added elements of marginal cost and risk based performance measures.

The framework of the proposed methodology consists of 4 major components,

which include a Supply Model, Reliability Forecast Model, Shortage Management

Actions, and a System Performance module. The following paragraphs describe the

components of this framework.


12

Uncertainty and Variability in Water Supply Planning

Uncertainty and variability can be incorporated in the modeling framework using

Monte Carlo analysis, which is a method of propagating uncertainties from model inputs

to the model results. Monte Carlo simulation produces random fluctuations to a given

variable within an identified probability distribution. For the purposes of this study,

uncertainty is based on random variables within a probability distribution that are applied

to a given parameter to represent unknown outcomes. Variability is a term used to

describe changes in historic patterns and recorded data. Both of these can be applied to

the inputs of a model to show probabilities of model outputs.

When water managers are faced with uncertainty of supplies while experiencing

booming population growth, they typically will make decisions based on scarce water

supply scenarios such as drought conditions within the 25th percentile (Thompson, 2007)

of probability or driest year on record. However, using uncertainty or variability

provides a wider range of outputs that can be plotted on a probability distribution curve

to assess probabilities of performance rather than performance of an isolated scenario.

Some of the most common variables that are characterized by uncertainty in water supply

models are stream flow, population growth, demands, and power costs.

Increasing supply capacity for growing populations is necessary, but it may be

possible more economical solutions can be found before capacity is increased

significantly (Money may be saved when large capital improvements are deferred to a

later time). The problem is uncertainty in supplies and demands makes it difficult to

quantify length of deferment.


13

Decisions are usually made by weighing chances of success or failure. If there is

a 10% chance of failure to meet water demands in 5 years, the decision to expand the

water supply system will most likely happen now. What if the chance of failure could be

reduced to 1% in 5 years by managing shortage? Do we decide rather than spend money

now, wait and rely on chance to get us through? How long are we willing to spend money

on implementation of water conservation measures to reduce risk? What risk level are we

comfortable with? Including uncertainty in the modeling framework will assist in

weighing the chances of success or failure.

The Supply Model

The supply model is intended to simulate long-term projected operations of water

supply being delivered to customers and is similar to the approach developed by Lund

(2000) and others. The proposed model is demand based, so demands are entered into the

model as inputs and this drives all the flows throughout the model using decisions and

controls to confine the simulation to represent reality. It is vital that this supply model be

calibrated to some period of actual time-based data. Once the model is calibrated and

operational parameters are fully understood, the model can then be expanded to run for

an extended period and again tested for continuity over time. Much of the input data for

the supply model will also be used for the forecast model. With a demand driven model,

the sequence of calculations, will in general follow the path summarized in the

hypothetical example shown in Fig. 2. Note the arrows shown in this table are intended

to show movement of logic (causes and effect) and not movement of water. In many

cases, feedback loops are caused by information driving a decision, but being constrained

on the other end by some criteria that needs to feed back information.
14

Input Data for the Supply Model

Input data used for of the efficiency-reliability modeling approach would

typically include, but may not be limited to the following items:

• Existing Supply Portfolio • System Constraints


• New Source Options • Operating Logic
• Demand Management • Water right priorities
• Hydrology • Demand Forecast

These input parameters can be characterized with or without uncertainty. If there

is not enough information available to use uncertainty, then use of multiple traces or a

range a values could be considered as input. The existing supply portfolio is a list of all

existing supply sources that the water supplier may rely on. New source options might be

completely new sources of water such as groundwater or new transfers in, but it can also

include wastewater reuse. A companion to new source options is demand management,

which is aimed at reducing demands (e.g. water conservation).

Hydrology is usually one of the driving factors in water supply, and should therefore be

carefully implemented. Hydrology, in the form of streamflow can be either synthetically

reproduced or indexed from historic data in order to represent uncertainty and/or

variability. System constraints usually include controls in the model such as pipe and

pump capacity, and reservoir stage-storage curves. Operating logic is critical for the

development stage of the model since the operating logic drives the decisions made in the

model. Usually, this kind of information must be obtained from system operators who are

very familiar with operations of the system. Examples of operating logic would be

typical reservoir fill rates and timing, modes of base-loading verses peaking at a water

treatment plant, and operation of a diversion gate on the river. Water right priorities are
15

Shortage

Population Total
Growth Demand
Total
Water rights
Deliveries

Allocation of
Supplies

Well pump Priority 1 Demand Priority 2 Demand


costs Deliveries Deliveries

Well pump Priority 1 Priority 2 Booster


Pump costs
rates Supply Supply pump rates

Aquifer WTP
WTP costs
Constraints Constraints

Reservoir
Operations

Stream Flow
Diversions

Fig. 2. Flow of logic in a hypothetical Supply Model

very important as these could significantly impact the amount of water that can be

diverted for supply to a utilities customer. Instream flows for environmental benefit also

fit into this category. The demand forecast can be laden with considerable uncertainty

while it is also usually the biggest driver in the decision process of supply planning.

Reliability Forecast Model

The reliability forecast model is used to first determine whether additional

supplies or demand management options are necessary for the next year in the supply

model, then to estimate the magnitude of the additional supply/demand management

options. The forecast model is a variation to Lund’s approach discussed in earlier and is a

part of the overall efficiency-reliability modeling approach. The forecast model runs at
16

the beginning of each year that the supply model runs while the supply model is paused.

While the supply model is paused, information is fed from the supply model to the

forecast model to be used as starting conditions. This information can include factors and

variables such as reservoir volume, population, current levels of conservation and added

supplies. This model then runs for a single year that is chosen by the user. The year

chosen by the user could include historic stream flow data, precipitation, and known

growth rates. The year that is chosen is important because if the year is associated with

very low streamflow data, then the forecast model will predict lower than normal

carryover storage. For a conservative model, dry years of record should be chosen as the

forecast year, but different years should be tested for sensitivity.

The forecast model calculates demands and facility operations the same way the

supply model does this. At the end of one year, the carryover storage volume of

reservoirs is converted to a shortage flow (if storage is lower than a target) and flow

shortages are also calculated if supplies cannot meet demands. The combination of

storage shortage and direct flow shortage are added together and sent out of the forecast

model. This shortage number is very important as it drives the next year’s estimated

action to reduce potential water supply shortages.

Shortage Management Actions

The term “Shortage Management Actions” refers to actions that can be taken by

water managers to influence the magnitude and/or occurrence of supply shortages. These

may include actions that either increase available supplies, or decrease the total demand.

Fig. 3 is a layout summarizing some possible components of SMAs that can be

incorporated into a supply model based on Mays (2002).


17

Cross-purpose diversions
Water treatment plant

Interdistrict transfers

Auxiliary emergency
Demand Reduction Measures System Improvements Emergency Water Supplies

Raw water sources

Distribution system
Public Education
Rebates and free distribution of
water saving devices
Restrictions on non-essential
uses
Prohibition of selected uses
Drought Emergency Pricing
Rationing Programs

Groundwater
Surface water
Enlarged storage
Wastewater Ruse
Desalinization Reduce system pressure
Cloud seeding Leak detection and repair
Discontinue hydrant and main flushing

Reduction of reservoir releases for Use dead storage in reservoirs


Import water by train or truck
non-depletion uses Temporary river diversions
Emergency interconnection
Relax environmental requirements

Fig. 3. Shortage Management Actions (based on Mays, 2002)

Depending on particular circumstances of a water supplier, priorities of SMA’s

will vary from one water supplier to another. In the SMA module, priorities are assigned

to SMAs so that they will be used in the order desired by the water supplier. Cost

efficiency is increased if lower unit cost SMA’s are implemented first.

The shortage values calculated in the Forecast Model are sent to the Shortage

Management Actions module, where a level of action is determined. For example, on the

first year of the simulation, it is determined that existing supplies can meet all the

demands. If this is the case, the shortage flow is zero and no SMA is used. On the

following year, the forecast model shows that there was a direct flow shortage of 3 cfs
18

and a storage shortage flow equivalent of 2 cfs. The SMA module adds them up and

applies 5 cfs to select from the highest priority SMAs until the 5 cfs is met.

After establishing the SMA flows, these values are sent back to the supply model

where the SMA flow value used to increase the supply capacity (in the case of supply-

side actions) or decrease demand (in the case of demand-side actions). The initial 5 cfs of

SMA flow is remembered and sent into the forecast model the following year as the

“new” capacity of the supply system. The forecast model repeats the process of

estimating supply shortages and if shortage is found, additional SMAs will be produced

on top of the previous 5 cfs determined earlier. This process is repeated through the

duration of the supply model simulation.

During the simulation, marginal costs of these actions are estimated. Costs to

implement the demand-side SMAs were obtained from research by Little (2002) and are

summarized in Table 1.

Table 1. Estimated Costs to Implement Demand-Side Shortage Management Actions

Conservation Measure Average Cost for Saved Water ($/AF)a

Voluntary Cut back $0


Water Audits/Surveys $1,300
Washing Machine Rebate $2,000
Landscape conversion Program $650
Toilet Distributions $200
Rate Changes (conservation pricing) $0

Notes:
a. Costs based on Little (2002)

There may be costs associated with maintaining water conservation, which would

impact marginal costs. A study by JVWCD found that the average cost to maintain their

conservation program was $39/AF (JVWCD, 2007). Cost information on the supply-side
19

is usually site-specific and is typically obtained from the water agency or utility. Supply-

side costs should include capital expenses annuitized (if money is borrowed) and

additional operational costs added to this for the duration of the life of the SMA.

Demand hardening is an important aspect of implementing SMAs. As defined in

the Introduction, demand hardening can reduce the flexibility of a water supplier to

manage supply shortages. Water conservation causes demand hardening because it

reduces the amount of non-essential demands that can be temporarily cut back in order to

reduce the risk of a shortage. Phoenix’s Water Resources Plan 2005 Update reveals an

unwillingness of the municipality to discourage outdoor water use except as a means to

reduce drought derived financial losses. It views non-essential water use as a buffer that

provides the utility flexibility in dealing with potential supply shortages (Bush, 2007).

Non-essential water use is defined as “water uses that are not required for the protection

of the public health, safety, and welfare” according to the City of Georgetown, Texas

(2007). Non-essential uses may include uses such as:

• Use of water to wash any motor vehicle, motorbike, boat, trailer, or airplane, or other
vehicle.
• Use of water to wash sidewalks, walkways, driveways, parking lots, or other hard-
surfaced areas.
• Flushing of gutters or permitting potable water to run or accumulate in any gutter,
street, or drainage culvert.
• Use of water to add to an indoor of outdoor swimming pool or hot-tub.
• Use of water in a fountain or pond except where necessary to support aquatic life.
• Use of water from fire hydrants for other than fire fighting and permitted use in
conjunction with a hydrant meter.

A simple relationship between non-essential demands and water conservation is

proposed to be used in the methodology discussed here to simulate demand hardening.

The basic concept is that there is a point when conservation measures have reduced non-
20

essential demands to a point that they are no longer considered as part of total demands

(Fig. 4).

The SMA module simulates demand hardening by allowing an initial amount of

shortage leniency. After establishing the initial amount, if additional conservation

measures are put in place, the leniency is reduced. The leniency is defined as 1- Ci / CC,

where Ci = conservation at the current time-step and CC = conservation at the point of

diminishing non-essential demands. Leniency is the ability of the public to reduce waste

or unnecessary water use on a short term basis. This term was developed for the purposes

of this study, but is based on the demand hardening concepts spoken of by TC&A (1994).

Definition of Marginal Cost

Marginal costs are defined as an increase in cost that accompanies a unit increase

in production or supply, or the partial derivative of the total cost function with respect to

production or supply. Marginal cost is calculated using a differential equation of the total

cost function:

TC = FC + f(Q) = FC + f(Q1, Q2, …, Qn), where TC = total cost, Q = production

increment, FC = fixed costs, Qn = ”nth” number of supply components for a single set of

alternatives, and f(Q) is the cost function for variable supply or production rate. Fixed

costs do not change over time, so this parameter drops out of the differential equation.

Differentiating the total cost equation yields this equation for marginal cost, MC =

dTC/dQ = d(FT + f(Q))/dQ = df(Q)/dQ.


21

Conservation

Point of diminishing non-essential


Non-Essential
Demand

demands
Time

Fig. 4. Proposed Relationship Between Water Conservation and Non-Essential Demands

The term f(Q) is the cost function to supply water, which can change depending

on economies of scale, facility changes, varying operating parameters, time of year, and

other economic drivers. The model of the new approach can be run for a number of times

equal to the desired steps in marginal cost analysis. Marginal costs are then provided as

end-of-year values.

Reliability-Based Performance

For the purposes of this study, the term reliability will be used as a general term

that encompasses reliability, resiliency, and vulnerability. These are defined by Mondal

(2007) as discussed earlier. Each of these three measures can be estimated as an average

value over the duration of the simulation to show comparisons between scenarios and

also as probability distributions at a given point in time to compare the probabilities of a

measure against other scenarios. Reliability is calculated in the supply model and then

sent to the System Performance module.


22

System Performance Module

The system performance module is a store house for results of the marginal cost

and reliability analysis after implementation of SMAs. These measures can be displayed

for selected years in the future in the form of mean values, levels at a given probability,

or in probability distributions. Understanding the reliability of a system can help in

determining future changes in rates and rate structures. A contingency fund can be

created to hedge against uncertainties in the risk of shortages. If drought measurements

are put in place and therefore reducing revenue, a nest egg or reserve fund can be

established to accommodate the change (Chesnutt, 1995).

Summary of Efficiency-Reliability Modeling Framework

To simplify the approach for this initial development, testing, and demonstration,

climate change, environmental management options, legal aspects, and funding

mechanisms are excluded from this framework. The new decision support framework

provides extensibility and flexibility to permit future incorporation of these factors as

needed for a particular application. Fig. 5 is a representation of the modeling framework

to be used for this study.

The efficiency-reliability modeling framework is similar to that used by Jenkins

and Lund (2000), which also incorporated supply and shortage management under

multiple uncertainties. The supply portion of their model estimates the probability of

supply shortages based on uncertainty of input parameters. The shortages (of their

model) are sent to the shortage management optimization model to determine appropriate

supply or demand management action at a given time.


23

Existing Supply Portfolio


New Source Options
Variability in River Flow
Demand Management
Uncertainty in population
River diversions
growth
System Constraints
Operating Logic
Water right priorities
Demand Forecast
Shortages

Supply Model
Shortage
(run for long term Reliability Forecast Model
Management
with variable stream (run for one year at a time)
Actions (SMA)
flow)

Supply Reliability

Marginal Costs of
Operations

System Performance

Fig. 5. Visual Representation of the Efficiency-Reliability Modeling Framework

The new approach introduced in this study differs in two ways. First, demand and

supply management actions are automatically fed back into the supply model on an

annual basis using a forecasting submodel. This is done to simulate long-term effects of

operational decisions on storage reservoirs (captures propagation of reservoir shortages if

necessary) and to calculate efficiency of on-going operational components of the system.

This approach assumes a water supplier has a set of options in mind for managing

supplies and demands, thus not requiring optimization to determine types of management

options automatically. This approach takes actions from the set of actions available, and

determines when they should be implemented. The forecast model also helps predict the

most economical amount of a given supply (or demand) action needed for the next year

that will still maintain reliability of the system overall. The forecast model will act

similar to a water supply manager looking ahead to the next year to make decisions

without knowing the future with certainty.


24

This change to the Jenkins and Lund (2000) approach is necessary to provide a

tool that more closely resembles day to day operational decisions and management

strategies. The second way in which this method differs from the Jenkins and Lund

(2000) approach is it incorporates reliability of the system as a function of shortages in

the supply model. Reliability is measured in a manner similar to that used by Mondal and

Wasimi (2007). For example, rather than saying, “our system may fall short of demands

in 10 years,” a reliability analysis will provide information to say, “our system has a 10%

chance of not meeting demands in 10 years.”


25

CHAPTER 4 - CASE STUDY APPLICATION

In testing the new approach against the traditional approach to water supply

planning, marginal costs will first be used for comparisons. If a new strategy that is

different from that used in the traditional approach is determined to have a lower

marginal cost, then this will demonstrate that more efficient strategies can be identified

using the new approach. Next, reliability of supplies will be compared against those of

the traditional approach to see if this new approach can find a more reliable strategy.

A model of Washington County Water Conservancy District’s (District) system

was developed using GoldSim Pro (GoldSim, 2008) software to help test the concept of

measuring performance based on marginal cost and reliability. The results of various

SMA implementation strategies will be compared against the strategy currently in place

for the District (LYR&B, 2005). The model was built using the methodology described in

Chapter 3.

District Background
The District is located in Washinton County, Utah. The largest city served by the

District is St. George, Utah. The area is characterized by arid climate, rapid population

growth, and limited and uncertain water supplies. The District began as a provider for

agricultural irrigators, but has more recently evolved into a major municipal water

supplier. In 1989, their water treatment plant (WTP) was upgraded to 10 MGD, followed

by two more upgrades to 20 MGD then 40 MGD in 1995 and 2001, respectively. It is
26

proposed that the next upgrade to 80 MGD occur in 2012. Because of the fast growth this

area has been experiencing, the District has determined that local supplies will be

exhausted within the near future.

The District has expressed interest in evaluating future operational and supply

strategies on the basis of marginal cost so that efficiencies in the system can be found.

These strategies are summarized in the Model Scenarios section of this report. The

District has been questioning the idea of changing the current operations of their WTP

and whether or not to increase conjunctive use of groundwater. They wanted to test their

ideas using marginal costs as a measure of performance. This study will help determine

whether the District’s ideas will reduce costs.

The District is currently working with the State of Utah Division of Water

Resources to design a pipeline that will carry water from Lake Powell to St. George to

add 70,000 AF/yr to the District’s current supply portfolio. Currently, the price for this

project is estimated to exceed $494 million and recent studies indicate that the pipeline

will need to be in place around year 2018 (UDWR, 2007). The Lake Powell pipeline

project has progressed through initial planning stages into preliminary design, bringing

along with it support and criticism (Wilson, 2007 and Wilkinson, 2002). It is very

important that the District can show a need for this large project and can show when the

project should be built. This case study will assist the District in showing a need for the

project through the use of reliability and efficiency performance measures.

Water Demand in the District

Current average water consumption in the District’s service area is 259 gpcpd,

with 61% of this being used for landscape irrigation (Thompson, 2007). Since 1990, only
27

four of the last 17 years has the area experienced above average surface water runoff.

Water conservation has become very important to this area because of fast growth and

dry conditions. Since 1993, the District has played an active role in promoting

conservation and was the first water district in Utah to develop a conservation plan. As

stated in the District’s “Long Term Framework for Water Resources Management,

Development, and Protection Plan,” created in 1993, states,

“The District shall develop a water conservation plan which


promotes public education and information dissemination
concerning water conservation; and promotes the adoption of
technologies, practices, and devices which will yield improvements
in the efficiency and management of water use.”(WCWCD, 1993)

Historically, the District has estimated water supply needs by determining the 25th

percentile annual streamflow magnitude. This has worked well because residents have

been able to cut back on the 61% of non-essential demand being used for landscaping. As

the 61% of non-essential water consumption is reduced, the risk of not meeting essential

demands increases. This forces the District to carefully consider reducing the percentile

of historical water supply probability to around 10 percent or less. By using the demand

hardening approach in this study, the results of the model will also assist the District in

determining the effects of conservation on reliability and marginal costs.

Population growth and increasing growth in demands is a major concern to the

District. In the past 20 years, the population has tripled and it is estimated that the current

growth rate might stay around 6 percent for some time to come (WCWCD, 2002). If

residents continue using water at the current rate, demands will exceed existing available

supplies within a few years. This is why conservation and demand management have

been essential aspects to the District’s water management philosophy. The District’s goal

is to undergo 25% demand reduction by 2015, from the starting point of year 2000. For
28

this model, it is assumed that the District will be able to conserve an additional 20% from

today. This assumption is based on the fact that conservation is currently at about the

15% level (year 2007) and 10% more will get them to year 2015. It is assumed that

another 10% could be reduced after 2015, but that non-essential demands will be

diminished substantially by the time a 10% reduction occurs from today.

Current District Operations

Approximately 75% of the District’s water supply comes from the Virgin River

and the remaining 25% is groundwater. Fig. 6 is a plot of historic flows in the Virgin

River from 1940 to 2003 and an exceedence plot of annual flow volumes is shown on

Fig. 7.

Water that is diverted from the Virgin River flows through a pipeline to an off-

stream storage reservoir called Quail Creek Reservoir (QCR). Before entering this

reservoir, water has an opportunity to pass through 2 hydro power plants. One of the two

hydro plants returns the flow back to the Virgin River. The Quail Creek WTP treats water

from QCR for delivery to St. George City and parts of Washington City. Before water

enters QCR, there is an opportunity to pump water up to a different off-stream storage

reservoir called Sand Hollow Reservoir (SHR). The purpose of this reservoir is mainly

for aquifer recharge. Many groundwater wells are located within the vicinity of SHR so

that water stored in the aquifer can be pumped back into the delivery system for St.

George City.
29

10 400

9 Daily Flow Rate


350
Annual Flow Volume
8
300
7
Flow Rate (1,000 cfs)

Volume (1,000 AF)


250
6

5 200

4
150
3
100
2
50
1

0 0
1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

Time
Fig. 6. Plot of historic streamflow in the Virgin River at the Virgin Gauge

Water can also flow by gravity from SHR to QCR, but this is currently rarely

done because this operation has not yet been needed. A regional pipeline connects these

two reservoirs and WTP with many groundwater wells located throughout the City and

plays an important role in providing essential conveyance capacity and flexibility to the

system. A schematic representation of the system is shown in Fig. 8.


30

100%
90%

Exceedence Probability
80%
70%
60%
50%
40%
30%
20%
10%
0%
0 50 100 150 200 250 300 350 400
Annual Flow Volume (1,000 AF)

Fig. 7. Exceedence Probability Plot for Annual Virgin River Flow Volumes

Future Water Development Plan

A capital facilities plan was developed for the District in 2005 in which the

timing of new projects was laid out (LYR&B, 2005). Table 2 is a summary of these

planned projects.

Table 2. Summary of Planned Projects


Project Date Project is to be Estimated Project Estimated Capital
Completed Yield (1,000 Costa (M$)
AF/yr)

Well Development (recharge) 2008 9 4.3

Ash Creek 2009 6 12.7

Agricultural Land Conversion 2015 15 11.1

Wastewater Reuse 2014 10 19.5

Quail Creek WTP upgrade 2015 See note b 42.5

Lake Powell PL WTP #1 2020 See note b 21.5

Lake Powell PL WTP #2 2025 See note b 21.5

Lake Powell Pipeline 2018 70 392.2

a. Cost shown in year 2007 dollars


b. These projects increase conveyance capacity and are not considered an added source of supply
31

Fig. 8. Schematic representation of the District’s water supply system

The Lake Powell Pipeline is by far the most expensive project on a per AF basis

so delaying this project will likely help reduce costs to the District over the long term.

Even though projects that are delayed will cost more in the future, money will always be

saved if purchases are delayed. This is because money that is saved now can most likely

be invested at a higher return than inflation and because the overall life of the project can

be prolonged.

The well development project consists of building new wells and a transmission

pipe to pump water from below the Sand Hollow Reservoir and convey it to the District’s

regional pipeline. It has been estimated that up to 9,000 AF/yr of reliable yield can be

obtained from these wells. Work has already begun on this project, but for the purposes

of this study it is assumed that full implementation of this project will begin as a
32

completely new project. The Ash Creek project would capture water from a

watershed located to the north of Quail Creek Reservoir, but is intended to serve

customers mostly located outside the scope of this model so this project has been

ignored. The agricultural land conversion project would allow water that is currently

diverted away from the Districts main Virgin River supply system and allow this water to

be directed to Quail Creek Reservoir. The land conversion will mostly occur in the

Hurricane Irrigation company’s land area, where currently irrigated lands will be

fallowed. The District has estimated that buying these water rights could cost up to

$1,000/AF.

Input Data

Data was obtained from the District, St. George City, and Utah DWR for daily

operations in 2006 of all supply facilities. The year 2006 was chosen because the District

did not own the WTP prior to this year and data prior to this time is not available. This

year is also considered by the District to be a relatively normal year for hydrology when

compared to other recent years. Fig. 9 is a graph of exceedence probability of annual

precipitation for the period of record from 1893 thru 2006. Precipitation in 2006 totaled

about 6.5 inches, which is just over the 60% probability level.
33

120%

100%
Exceedence Probability Year 2006

80%

60%

40%

20%

0%
0 2 4 6 8 10 12 14 16
Total Annual Rainfall (inches)
Fig. 9. Exceedence Probability of Annual Precipitation in St. George (USU, 2007)

Only one year was used for calibration because of time constraints on data

collection. Much of the daily flow records were obtained from hand-written data sheets,

with each day’s record on individual sheets. Table 3 is a summary of the information

obtained and used to develop the model for the case study.

Table 3. – Summary of Existing Conditions Model Input


Parameter Input Value

Virgin River flows Varies

Maximum Diversion Capacity 165 cfs

Minimum instream flow 3 cfs


requirement

LaVerkin Irrigation Demand Avg flow: 5 cfs, Total: 1,200 AF/yr (demand pattern based on 2006 flow rates)

Hurricane Hydro Power Plant 30 cfs, 590 kW


Capacity

Hurricane Hydro Plant operational Maintain minimum 86 cfs in downstream Virgin River for Washington Fields
objective Canal Diversion

Quail Creek Pipeline capacity 150 cfs


downstream of Hurricane Hydro
Plant

Hurricane Irrigation Demand Avg flow: 25 cfs, Total: 12,000 AF/yr (demand pattern based on 2006 flow rates)
34

Table 3 (cont’d). Summary of Existing Conditions Model Input

Parameter Input Value

Other Irrigation Demands on Quail Avg. flow: 6 cfs in summer, 3 cfs otherwise, Total: 1,500 AF/yr
Creek Pipeline

Quail Creek Hydro Power Plant 110 cfs when pumping to SHR, 150 cfs otherwise, 243 kW
Capacity

Quail Creek Hydro Plant Reduce power production at plant when maintaining high flows is critical
operational objective

Sand Hollow transfer pump station 110 cfs


capacity

Sand Hollow transfer pump station Pump water to SHR only when water diverted out of the Virgin River is clean.
operational objective This typically occurs in the winter and in April. Water can flow by gravity from
SHR to QCR when not pumping. Gravity flow to fill QCR when needed. Run for
at least 1 month duration to avoid demand charges.

Sand Hollow Reservoir (SHR) Capacity: 50,000 AF, drought pool: 20,000 AF, avg. seepage rate to aquifer:
parameters 10,000 AF/yr

SHR operational objectives This reservoir takes second priority to fill behind QCR. Purpose of reservoir is to
recharge aquifer.

SHR groundwater recharge wells Ultimate sustainable capacity in the future is estimated to be 9,000 AF/yr. The
capacity District has water rights to pump up to 15,000 AF/yr. Currently, capacity is
limited by a conveyance pipe from the wells to the delivery point

Quail Creek Reservoir (QCR) Capacity: 40,000 AF, drought pool: 20,000 AF
parameters

Quail Creek Reservoir operational Filling this reservoir takes priority over SHR. June 1 storage target is 40 AF.
objectives

Quail Creek Water Treatment Plant 40 MGD


(WTP) capacity

Quail Creek WTP operational Currently, the District desires to base-load the WTP instead of peaking off the
objectives well supply base flows.

Notes: Data source is WCWCD except for Virgin River Flows are from Utah DWR. Utah DWR = Utah Division of
Water Resources (a division of the Utah Department of Natural Resources. WCWCD = Washington County Water
Conservancy District

Model Development
A model was developed to represent existing conditions and operations that occurred in

year 2006. The development process required frequent conversations with District and

City staff, including the City engineer, District hydrology coordinator, District O&M

supervisor, WTP manager, and District accountant.


35

The existing conditions model was built in GoldSim software through an iterative

process of on-going calibration and discussions with operations staff, including St.

George City engineer and the District’s O&M supervisor in order to capture a significant

amount of the operations logic. The model was built in a way that would allow for future

additions to the model in preparation for building the case study model. This was done by

building the model in compartments representing individual facilities. These

compartments can be edited, removed, or new compartments can be added.

The hurricane hydro power plant flow rate is calculated as the flow required to

achieve 86 cfs downstream in the river. Because of return flows downstream of the

power plant and the recursive nature of this kind of operation, an iterative approach is

used to calculate this flow. If the downstream flow is found to be less than 86 cfs in the

river, then a flow increment is added to the power plant flow. This process is repeated

until the minimum flow in the river is achieved. Sometimes the flow cannot be achieved

because of insufficient flows upstream in the river. If this is the case, flow into Quail

Creek reservoir are reduced to zero so that the remaining flow can be conveyed through

the power plant and back to the river.

Variability of Streamflow and Uncertainty of Demands

Streamflow variability was incorporated into this model in an effort to capture

hydrologic uncertainty that appears to drive operations of much of the system. The

streamflow variability was captured by indexing the historic streamflow data and running

various years of data using Monte Carlo simulation. Monte Carlo typically involves

numerous simulations performed with variables changed based on random sampling

from assigned probability distributions. For this model, the Monte Carlo capabilities of
36

GoldSim were used only to randomly sample from historic streamflow data rather than

assigning probability distributions to random variables.

Uncertainty in water demands was estimated using a stochastic parameter based

on a log-normal statistical distribution with a geometric mean of 1 and a standard

deviation of 1.06. The stochastic function provides a multiplier that is applied to the daily

demand function to represent daily fluctuations in demand. The stochastic function was

developed to produce a similar range of fluctuation as that which actually occurred in

2006. The Monte Carlo analysis was performed by running 100 realizations in order to

capture uncertainty variations in demand in combination with variability in the Virgin

River streamflow.

Model Calibration and Validation

The existing conditions model was calibrated using observed data from 5 main

areas of the system. These areas were chosen for calibration because of their overall

effect on the system as a whole. The calibration elements are: Diversion from the Virgin

River, Pumping flows and costs, WTP production, reservoir levels, and hydro plant flows

and revenues. The calibration was performed by adjusting parameters in the model until

simulated values matched observed values within reason. Details of how the calibration

was quantified and qualified are discussed in Appendix A of this report. The results of the

calibration show that the existing conditions model simulates actual operations with a

reasonable level of accuracy based on scatter plot line fit comparisons and qualitative

measures based on visual plot comparisons. After calibration, the model was converted to

be used as the Supply Model to run for multiple years as described in the next section.
37

The model was validated by performing a daily mass balance check on the entire

system. The mass balance was calculated using the equation MB = Qin – Qout - ∆

storage, where Qin is total inflow, Qout is total outflow, and ∆storage is the change in

volume of all storage elements in the model. Fig. 10 is a graph showing the mass balance

error value, which is very small.

Simulation Scenarios
Two modeling scenarios were developed and incorporated in the model:

1. New Method: This scenario is called the new method because it incorporates the

methodology of using reliability and efficiency as discussed above. Shortage

management actions are chosen based on forecasted shortages each year.

2. Traditional Method: This scenario uses the supply and conservation strategy based on

the District’s capital facilities plan. This scenario is used to compare results against

Scenario 1.

Performance will be measured using reliability and marginal cost results for both

scenarios.

The Forecast Model – Case Study

The forecast model was developed using the methodology discussed in Chapter 2

and is based on the calibration model. Fig. 11 is a screen capture showing how the

forecast model interacts with the rest of the model. The forecast model receives inputs

from the supply model, such as reservoir volume, population, supply capacity with

SMAs, and all the other duplicate input parameters that are used to control the supply
38

model. The forecast model runs for 365 days at the beginning of each year the supply

model is run and then results from the forecast model are fed back into the SMA module.

0.0002
Mass Balance Error (gpm)

0.0001

0.0000

-0.0001

-0.0002
Oct Dec Feb Apr Jun Aug Oct
Time

Fig. 10. Mass Balance Check Plot for the Model

Fig. 11. Screen Capture of the model

The purpose of the forecast model is to predict end of year storage carryover and

flow shortages given some starting conditions from the supply model, specified

hydrology, and scenario inputs based on the type of scenario that is run. This model runs

using the same operations logic as the existing conditions model.


39

The user can enter a hydrologic year in the forecast model so that it can select that

year’s historic streamflow for the forecast simulation. By selecting a dry year, the

forecast model will project a more conservative shortage value to be used to calculate the

SMA, which is then put in the supply model. For this model, the year 2002 was used as

the hydrologic year of the forecast model. The year 2002 is the driest year on record

(1940 to 2007) for the Virgin River. Note that using the driest year on record in the

forecast model is not the same as using the driest year on record for all years run in the

supply model. Using the driest year on record in the forecast model will force the next

year’s SMAs to be conservative, but will not propagate shortages in reservoirs of the

supply model. Since the supply model is running on indexed historic record flows, it will

be able to rely on wet to normal years to assist in recovering from droughts by

replenishing storage.

The SMA module – Case Study

The SMA module receives supply shortage information from the Forecast model

and uses the magnitude of the shortage to determine what level of additional SMA needs

to take place for the following year in the supply model. For example, it is found in the

forecast model that there is a direct flow shortage of 3 cfs and a carryover shortage of

2,000 AF. Carryover shortage is calculated for QCR, SHR, and the aquifer below SHR.

The aquifer below SHR is simulated as a contained system with 10% outflow. The

District has stated that if 10,000 AF are injected into the aquifer from SHR, then they are

entitled to 9,000 AF to pump out. Knowing this, the aquifer can be modeled as a simple

reservoir. The carryover shortage is converted to a flow by calculating what flow rate it

would take to eliminate the 2,000 AF shortage and match the carryover target. It is
40

assumed that if a constant flow rate over the year is added, it would be enough to allow

the reservoirs to meet their targets.

Depending on the scenario, the individual SMAs are allocated to the shortage that

was calculated. Following the example, it was determined that a 3 cfs direct flow

shortage and a 2,000 AF carryover shortage (found to be equal to an equivalent 2.8 cfs) is

5.8 cfs. If conservation measures are required to be used before supply-side actions, then

up to 5.8 cfs of conservation will be applied to the supply model to reduce demand by

that much. This process is repeated each year until SMA capacity is met, and then no

more additions can be made to the supply model and shortages begin to increase.

The unit costs of SMAs are located inside the SMA module and are used to

calculate the annuitized cost to implement the SMA. For all SMA costs, it is assumed

that projects are paid for using a 30-yr loan with a 5.5% interest rate. Costs and ultimate

capacities of each SMA are summarized in Table 4.3.

Table 4. Cost to Implement Water Conservation Measures and capacities at build-out

Conservation Measure Average Cost for Saved Assumed SMA Capacity at Build-out
Water ($/AF)b (AF/yr)c

Voluntary Cut back $0 7,000a

Water Audits/Surveys $1,300 7,000

Washing Machine Rebate $2,000 7,000

Landscape conversion Program $650 15,000

Toilet Distributions $200 3,000

Rate Changes (conservation pricing) $0 7,000

b. Voluntary cut backs are based on the public’s leniency to supply shortage induced mandates to cut water
use on a short-term basis. This SMA is reduced to zero as all other demand based SMAs reach full
capacity.

c. Costs based on Little (2002)

d. Total build-out capacity estimated as 20% of demand at growth rate documented in the District’ CFP
41

The SMA module also simulates demand hardening by allowing a small amount

of shortage leniency on the first year of the simulation. After the first year, if additional

conservation measures are put in place, the leniency is reduced. Conservation capacity

(CC) is 20% of demand, which corresponds with the District’s conservation plan.

The Supply Model – Case Study

The supply model is very similar to the forecast model except that it runs for 20

years and makes changes to supplies and demands based on results of the SMA module.

The supply model is run as a Monte Carlo simulation for river indexing. River indexing

is a method of applying a Monte Carlo type of approach to select various years of record

to represent the uncertainty in the system. Daily river flow data was obtained from Utah

Division of Water Recourses for the years 1941 thru 2006. Every time the supply model

runs, it selects a different starting year to index the historic data. Since each model run

takes 1 minute to run thru the 20 years, it was found that running more than 20

realizations of this model would be quite time consuming. In order to capture all the

years on record, the supply model jumps forward 5 years on every realization when it

indexes the historic stream flow.

Population projections drive supply forecasts, so it is important to be careful not

to over or under estimate projections. For this model, the population projection used in

the capital facilities plan (CFP) was used to calculate future demands. Fig. 12 shows a

plot of the population growth rate used in the model based on the projected rate from the

CFP.
42

6.5

6.0

Annual Growth Rate (%)


5.5

5.0

4.5

4.0

3.5

3.0
2006 2009 2012 2015 2018 2021 2024

Year

Fig. 12. Population growth used in the supply model

The supply model pauses at the beginning of each year while the forecast model

and SMA module perform their tasks, then it continues until year 2025. While the supply

model is running, it sends cost data and reliability data to a results module that

categorizes the information into each year. When the simulation is complete, this data is

exported to MS excel so that costs can be plotted against production for a marginal cost

plot and reliability measures can be plotted against production as well.

System Performance Module – Case Study

The system performance module is used to summarize results of the reliability

and marginal cost analysis. Reliability is calculated as a probability of not completely

meeting demands as either a storage or flow shortage. Using Monte Carlo analysis, 100

realizations are run and the percent chance of not meeting demands is calculated.

Marginal cost is calculated by plotting variable costs against increasing production

levels, curve fitting the scatter points, and taking the derivative of this curve. Marginal
43

cost shows the rate at which variable costs are increasing as production increases. Lower

marginal costs show that the system is more efficient.


44

CHAPTER 4 – CASE STUDY ANALYSIS RESULTS

The model results are organized to demonstrate two main points:

1. To show that operational costs can be reduced while maintaining reliability

2. To show that this new approach can provide additional insight into the timing

of large future supply acquisitions

The District was planning on operating the WTP with a base flow and peaking off

the groundwater wells. Using the new modeling approach, it was found that this

operational scheme would pose greater risk of shortage as opposed to using the wells as

base supply and peaking off the WTP. Fig. 13 shows the reliability of the system with

and without the peaking operation of the WTP using the New Method. On this graph,

reliability is represented as the probability that no shortage occurs. If no shortage occurs

for the entire Monte Carlo analysis (100 realizations), the system is assumed to be

completely reliable for the purposes of this demonstration.

The reason peaking wells is less reliable than WTP peaking is because base

loading the WTP draws heavily on the reservoirs and prevents them from recharging in

normal to wet years. It was found to be very important that the reservoirs stay as full as

possible during the normal to wet years so that they can be relied upon during the dry

years. This issue came to light using the new method of supply planning, which relies on

operational details, hydrologic variability, and demand uncertainty. It was also found that

peaking off the WTP is less expensive because in this case the WTP is treating less
45

100%

90%
Probability of Perfect Reliability 80%

70%

60%

50%
40%

30%

20%
Peaking WTP
10% Peaking Wells

0%
2007 2009 2011 2013 2015 2017 2019 2021 2023 2025

Year

Fig. 13. Supply Reliability Using WTP Peaking Compared to Well Peaking

volume of water and expensive chemical usage is reduced. Base loading makes

operations of the WTP a little less complicated, but does not mean less man-power is

required to run the plant in base-loading operations (Childers, 2007).

It was also found that maximizing diversions to Quail Creek Reservoir (QCR) is

useful even if the reservoir is full and the water needs to be spilled back to the Virgin

River because of increased hydropower revenues. The traditional approach to water

supply planning has no way of analyzing the financial and risk impacts of these

operational modifications.

The new supply planning method uses an optimal timing approach to adding new

SMAs as described in the forecast and SMA model methodology. This approach assumes

that SMA’s can be built in phases so that capital costs can be minimized while

maintaining reliability of the supplies. For example, the new approach will tell the supply

planner just how many wells to build for each succeeding year rather than assuming that
46

70

60
Conservation
Additional SMA Volume (1,000 AF)

Reuse
50 Ag Conversion
New Wells
40

30

20

10

0
2009

2014

2016

2021

2023

2024
2008

2010

2011

2012

2013

2015

2017

2018

2019

2020

2022

2025
Time
Fig. 14. Average Annual Additions to Supply Capacity from the New Method

the ultimate number of wells for maximum capacity would be built all at once. Fig. 14 is

a plot of average annual additions to system supply capacity using the New Method. This

plot represents the average of 100 Monte Carlo realizations. As seen in Fig. 14, SMAs

are added gradually over time in phases. This is considered to be practical for the types

of SMAs used in this case study. For example, wells can be added to a system one at a

time rather than all at once as can conservation practices. It is also important to consider

the more aggressive projections of SMA additions that may come out of the Monte Carlo

analysis. Fig. 15 is a plot of a single trace from the Monte Carlo analysis showing one of

many possible scenarios of supply acquisition.


47

70

60
Additional SMA Volume (1,000 AF) Conservation
Reuse
50 Ag Conversion
New Wells
40

30

20

10

0
2010

2014
2008

2009

2011

2012

2013

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025
Time
Fig. 15. Single Trace of Annual Additions to Supply Capacity from the New Method

This scenario shows aggressive SMA additions from 2011 to 2016 and again from 2021

to 2024, which occurred during drought years. SMAs did not increase from 2016 to 2021

because these were wetter than normal years and the reservoirs were full in 2016. Fig. 16

is a plot showing annual additions to the system supply using the traditional approach.

The traditional approach does not provide a way to differentiate between drought and

non-drought years, so it will produce the same plot for all realizations of the Monte Carlo

analysis. The supply acquisition plan shown in Fig. 16 is not dynamic like it is using the

efficiency-reliability method.
48

70

60 Conservation
Reuse
Added Capacity (1,000 AF/yr)

50 Ag Conversion
New Wells
40

30

20

10

0
2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025
Time
Fig. 16. Annual Additions to Supply Capacity from the Traditional Method

Fig. 17 is a graph showing the difference in average marginal cost between the

new method as compared to the traditional method. The marginal cost curves shown in

Fig. 17 were developed based on variable costs as shown in plots in Appendix B. Note

that using the traditional method, it was found that marginal costs rise sharply in the

range of 35,000 to 45,000 AF of production and then drop off after 50,000 AF of

production. This shape is due to a much more aggressive SMA implementation than is

required for a completely reliable supply system. The average marginal costs of the new

method are much flatter and more stable through the range of future additional SMAs. A

higher marginal cost means that more money is being spent to produce the same unit of

water. The marginal curve for the new method follows closely to the growth rate of water

demands.
49

2.5
New Method
Marginal Cost ($1M) 2.0 Traditional Method

1.5

1.0

0.5

0.0
35 40 45 50 55 60 65
Production (1,000 AF)

Fig. 17. Average Marginal Cost Results for the New Method and the Traditional Method

The District would save money by delaying those expenditures until they are

needed. Note that a flatter marginal cost curve does not necessarily mean less total

money was spent by year 2025, but rather that there is a difference in the rate at which

the money is being spent. It is also important to note that the reliabilities for these two

options are very similar, as shown in Fig. 18.

The results of the model also demonstrate that the new approach can provide

additional insight into the timing of large future supply acquisitions. The future Lake

Powell Pipeline is considered a very large supply acquisition and the need for this project

should be considered carefully. The model was run without including the Lake Powell

pipeline to see when it would be needed based on a reliability standpoint. Fig. 19 is a plot

of marginal cost compared to reliability using the new modeling approach.


50

100%
90%
Probability of Shortage 80%
70% New Method
60% Traditional Method

50%
40%
30%
20%
10%
0%
2007 2012 2017 2022
Year
Fig. 18. Reliability plot for the new method verses traditional method

1.8 100%

1.6 90%

1.4 80%

70%
Marginal Cost ($M)

1.2
Marginal Cost

Reliability (%)
60%
1.0 Reliability
50%
0.8
40%
0.6
30%
0.4 20%
0.2 10%

0.0 0%
30 35 40 45 50 55 60 65 70
Production (1,000 AF)

Fig. 19. Plot of Average Marginal Cost compared to Reliability for the New Method

Reliability begins to drop off around the production level of about 55,000 AF. At this

point, the marginal cost of production is just over $1 million. Compare this result to that

of the traditional approach shown in Fig. 20.


51

2.5 100%
Marginal Cost
90%
Reliability
2.0 80%
70%
Marginal Cost ($M)

Reliability (%)
1.5 60%
50%
1.0 40%
30%
0.5 20%
10%
0.0 0%
30 35 40 45 50 55 60 65 70
Production (1,000 AF)

Fig. 20. Plot of Marginal Cost compared to Reliability for the Traditional Method

The traditional approach shows reliability starting to break down around the 50 to 55,000

AF production level, similar to the reliability of the new method. However, the marginal

cost at this point is closer to $2 million. More money was spent to produce the same

amount of water for essentially the same reliability for both scenarios. These results also

show that the SMAs non-inclusive of Lake Powell can reliably supply up to 55,000 AF.

After this point, it appears that the Lake Powell pipeline should be put in use. Using the

new modeling approach, it was found that 55,000 AF/yr of production would occur

around year 2020 based on the assumed demand growth.

Table 5 is a summary of results of a comparison made between the traditional

approach as described in Chapter 2 and the new approach of incorporating marginal cost

and reliability.
52

Table 5. Comparison of Traditional Approach vs. New Approach


Traditional Supply Planning
Description Approach New Approach

Maximize use of wells, maximize


What operational changes can be made in Virgin diversions, and peak off the
the system to increase efficiency? Unknown WTP while base loading the wells

Demands are projected to The probability of shortage rises above


Reason for building Lake Powell Pipeline exceed supply somewhere 20% in year 2020 or at a production
in year 2018 between 2018 and 2024 rate of 50,000 AF/yr

Reservoirs cannot provide supply after


Long-term effect of using driest year on a repeat of 3 of the driest years on
record to forecast supplies Unknown record

Decreases demands, increases


marginal cost, and increases risk of
Effects of water conservation Decreases demands supply shortages

Rather than just finding the point where annual demand crosses the supply line to

mark a time to build a new project, a more rigorous approach to supply planning should

be considered. Using this new approach, the limitations of the reservoirs, pipelines, pump

stations, and WTP are incorporated for a more operational focus. The reliability of the

combination of all these components can be different than the reliability of an isolated

component when viewed independently of the system. This new method ties the

operations of all the facilities into one whole, providing a quantitative view of system

reliability. Putting more pressure on the WTP puts more pressure on the reservoirs, which

increases risk of shortages. Putting more pressure on Sand Hollow Reservoir and

drawing it down decreases the supply available to the recharge wells. These kinds of

cause and affect issues in the system cannot be visualized using the traditional approach

of supply planning.

Marginal cost provides a way to compare efficiencies of operations of the system.

The traditional method of supply planning does not consider the marginal costs of base-

loading the WTP and therefore would assume a higher risk scenario. The new method not
53

only provides a way to find operational efficiencies but also quantifies the level of

efficiency as a function of marginal cost.


54

CHAPTER 5 – CONCLUSIONS

The results of this study show that increased operational efficiencies in a water

supply system can be found using the new method while maintaining equal or higher

reliability in the system. The framework of the model is set up to easily facilitate multiple

operational scenarios for comparison.

The results also show that this approach can provide better insight into timing of

large future supply acquisitions. Quantifying reliability provides useful decision support

for water suppliers who are considering acquiring significant water projects in the future.

The new method demonstrated in the case study shows that risk of not meeting demands

increases at a known point in time, related to a given level of production. As shown in the

results, this may not be exactly the same as the point at which a projected demand line

crosses the supply limit.

The approach developed in this thesis could be applied to real-world water supply

applications. Using this concept can help a water supply agency or utility develop more

reliable supply strategies at the lower costs while fitting their specific operational needs.

This approach can also help the water agency provide a basis for a decision to build a

new water supply project that can be backed up quantitatively. The traditional method of

water supply planning that many utilities rely on shows the need for future projects but

fails to effectively quantify the need. In developing a strategy using this approach, the

decision maker will be empowered to justify their decisions by these measures.


55

Model Limitations
The model limitations are summarized in the list below.

• Unit costs of SMAs are based on a single study and should be verified using

additional research

• Calibration of the existing conditions model is based on a single year. Operational

logic in the model could be further refined by calibrating to multiple years under

various hydrologic conditions.

• Shortage Management Actions may look good on the basis of economics and

reliability, but this approach does not incorporate public perception issues, public

policies, and political pressures to build new large projects

• This approach is data intensive and time consuming in the development phase of the

model. Depending on the modeling software used, simulation run times can become

very long, making it difficult to iterate through scenarios and tuning strategies

through trial and error. The GoldSim Pro software is extremely visual and easy to use,

but the run time for 100 realizations is nearly 3 hours, so a more efficient tool might

be a better choice for this application.


56

APPENDIX A – MODEL CALIBRATION

All calibration plots in this appendix use thick solid lines to represent model results and

thin, dashed lines to represent observed data.

Calibration of Diversion from the Virgin River

Calibration of the diversion was accomplished by validating outflows along the

Quail Creek Pipeline, which include irrigation diversions, Hurricane Hydro Plant, and

pumping up to SHR. In order to achieve some form of calibration, these other elements

were adjusted to match historic data and operational strategies set forth by the District.

The irrigation demands are calculated using an annual volume multiplied by a demand

pattern. The Hurricane Hydro plant flows are calculated using a more complex algorithm

as discussed in the section on calibrating hydro plant flows. Adjustments made to

pumping rates to SHR are also discussed later. Fig. A-1 is a calibration plot showing a

qualitative comparison.

Three areas of this plot need to be considered separately. The first portion is from

January through June. This is a critical filling period for QCR and SHR and is where

most of the inflows occur. The model produced a diversion flow almost 20 cfs lower than

the observed flows in February and March. This is because in 2006, more water was sent

to QCR even though it was already filled to capacity. According to the District, flows

into QCR are typically shut down if QCR fills to capacity. Therefore, this difference
57
Validation Plotfor Virgin River Diversion

200

150
(cfs)

100

50

0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Time

Legend:
QC_divert QCPobserved

Fig. A-1. Comparison of plot for the Virgin River diversion

in February and March is due to an operational anomaly. The area between July and

October is a period when most of the water flows back to the river through the Hurricane

Hydro plant and is diverted to irrigation users. The difference between the model results

and the observed data can be explained as a result of flow fluctuations in canals that

divert water to agricultural lands. The available flow records did not show all of these

fluctuations because flow measurements were not taken with enough accuracy or enough

frequency. The area between November and December is a time when turbidity in the

Virgin River is highly variable (late summer also has this problem). When turbidity is

high, the District avoids diverting water to the reservoirs because the higher turbidity

causes problems at the WTP and could reduce the infiltration capacity at SHR due to

siltation of the reservoir bottom.

Calibration of Pump Stations

The Sand Hollow Reservoir Transfer pump station is operated on a monthly basis

to reduce demand charges induced by multiple occurrences of pump start-ups. It is also


58

important that dirty water is not pumped into SHR so that siltation of the reservoir

bottom can be minimized. The pump will typically be turned on during the months that

turbidity in the river is low. Fig. A-2 is a plot showing a comparison of the model results

and observed data. As shown in this plot, a significant discrepancy occurs in the months

of April and May. It was determined that the pumps were likely run at a lower speed in

April possibly in anticipation of filling the reservoir. Since the model cannot simulate

with perfect foresight, this operation was not reproduced. Rather, the model uses a

ramping-down equation that reduces pump flow as the reservoir fills to capacity. This is

why the model shows higher flows in April and lower flows in May. If the April-May

time period is disregarded, the R^2 value for the comparison is 0.9. On the basis of

annual flow volume, the two are only 3% different.

The pumping cost comparison of the Sand Hollow Transfer pump station is

shown in Fig. A-3. The annual pumping costs difference is 3%.


59
Validation Plotfor Pump Station to Sand Hollow

100

80

Flow Rate (cfs) 60

40

20

0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Time

Legend:
Pump SHpumpObs

Fig. A-2. Comparison plot for the Sand Hollow Transfer Pump Station
Pump CostValidation

50

40

30
(k$)

20

10

0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Time

Legend:
SHPCcostObs TCshp

Fig. A-3. Comparison plot for the Sand Hollow Transfer Pump Station pumping costs.

The City well pumps were also compared against observed data as shown in Fig.

A-4. The observed data was provided in monthly flow increments, but the comparison

seems qualitatively pretty accurate on a monthly basis. On the basis of annual flow

volume, the two are 4% different.


60
Simulation of St. GeorgeCity Wells

20

18

16

Flow Rate (mgd)


14

12

10

6
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Time

Legend:
Qcw CWobs

Fig. A-4. Comparison plot for the St. George City well pumps

The main significant difference between the model results and the observed data

occurs in July, where actual flows were reduced from the previous month even though

demands increased. This is considered to be an anomaly of operations because the wells

have capacity to maintain the flow observed in June for three months time.

The pumping cost comparison of the City wells is shown in Fig. A-5. The annual

pumping costs difference is 4%.

Calibration of the Quail Creek WTP Production

The WTP production is a function of total demand and flows of competing

supplies, which consist of the City wells and the Sand Hollow recharge wells. The

demand has a stochastic component that produces uncertainty on a daily basis in an

attempt to represent reality. Fig. A-6 shows a comparison of the model results and the

observed flows at the Quail Creek WTP. When these are plotted against each other, an

R^2 value of 0.84 is achieved.


61
CumCW

16

14

12

10
(KAF) 8

0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Time

Legend:
CumCW CumCWobs

Fig. A-5. Comparison plot for the cumulative flow at the St. George City well pumps
Validation Plotfor WTP Production

30

20
(MGD)

10

0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Time

Legend:
Qwtp WTP2006

Fig. A-6. Comparison plot for the Quail Creek WTP

Calibration of the reservoirs

The Quail Creek Reservoir comparison plot is shown on Fig. A-7. Many

discussions with the District occurred because of this plot. After many thoughts were

relayed back and forth and multiple conjectures pointed out, no one reason can be

considered as the source of the error in this comparison. The most significant error in this
62

comparison is related to the difference in drawdown in the summer. It is not understood

why the observed data shows such a fast drop in the first part of July because no large

spill actually occurred. One thought is that the water level readings could have been

made erroneously. But even if this were the case, the volume later in the summer would

still be able to match very closely. It is concluded that something is amiss with the

observed data in June and July. The demand on this reservoir is the WTP, and that

variable calibrates well. The only other variables are seepage and evaporation.

The District noted that seepage from this reservoir is near zero, so this leaves

evaporation. However, the SHR calibrated well using the same evaporation, so changing

evaporation for this lake would require changing evaporation at SHR, which would just

shift the error from one reservoir to the other. Fig. A-8 is a plot showing the comparison

of model results and observed volume at SHR. This reservoir calibrated well, with an

R^2 value of 0.97.


63
Validation PlotFor Quail CreekReservoir

40

38

Storage Volume (kaf)


36

34

32

30

28

26

24
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Time

Legend:
QCR QCRVobs

Validation Plotfor Sand HollowReservoir


Fig. A-7. Comparison plot for the Quail Creek Reservoir
52

50
Storage Volume (kaf)

48

46

44

42
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Time

Legend:
SHR SHRVobs

Fig. A-8. Comparison plot for the Sand Hollow Reservoir


64

APPENDIX B – DETAILED COST CURVES


65

9 1.8

8 1.6
y = 103814e0.067x
7 R2 = 0.8479 1.4
Variable Cost ($1M)

Marginal Cost ($1M)


6 1.2

5 1.0

4 0.8

3 0.6

2 Cost x Prod. 0.4


Marginal Cost
1 0.2
Variable Cost

0 0.0
30 35 40 45 50 55 60 65 70
Year

Fig. B-1. Development of the Marginal Cost Curve for the New Method
66

Scenario 7
9 3.0
4 3 2
y = 41.047x - 8074.6x + 578350x - 2E+07x + 2E+08
8 R 2 = 0.8762
2.5
7

Marginal Cost ($1M)


Variable Cost ($1M)

6 2.0
Cost x Prod.
5
Marginal Cost 1.5
4 Variable Cost

3 1.0

2
0.5
1

0 0.0
30 35 40 45 50 55 60 65 70
Year

Fig. B-2. Development of the Marginal Cost Curve for the Traditional Method
67

70

Conservation Rates
60 Rebates
Audits
Additional SMA Volume (1,000 AF)

50 Landscape Conversion
Toilet Distributions
Reuse
40
Ag Conversion
City Wells
30 ASR Wells

20

10

0
2009
2008

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025
Time

Fig. B-3. Additional SMAs in the New Method

Note: SMAs on layering the top are all conservation (demand based) SMAs
Added Capacity (1,000 AF/yr)

0
10
20
30
40
50
60
70

2008

2009
2009

2010

2010
Reuse

2011
New Wells

2011
Conservation

Ag Conversion

2012

2012

2013

Fig. B-4. Additional SMAs in the Traditional Method


2013

2014

2014
2015

2015

2016

2016
2017
Year

2017

2018

2018
2019

2019

2020

2020
2021

2021

2022

2022
2023

2023

2024

2024
2025

2025
68
69

REFERENCES

Barakat & Chamberlin, Inc. (1994). The Value of Water Supply Reliability: Results of a
Contigent Valuation Survey of Residential Customers. California Urban Water
Agencies. August, 1994.

Bishop, D. B., Weber, J.A. (1996). Impacts of Demand Reduction on Water Utilities. AWWA
Research Foundation.

Bockman, S., Longville, S.L., Sirotnik, B., Ruiz, C., Hanna, G. (2007). Statewide Market
Survey: Landscape Water Use Efficiency, Final Report. California Urban Water
Agencies.

Bush, J.C. (2007). Wringing Water-Thrifty Urban Design from Southwestern Water Plans.
Southwest Hydrology Journal, May/June 2007. Webpage:
http://www.swhydro.arizona.edu

CDM, Meurer & Associates. (2007). South Metro Water Supply Authority Regional Water
Master Plan. For Denver South Metro Water Supply Authority. (June 2007).

Chesnutt, T.W., Bamezai, A., Hanemann, W.M. (1995). Revenue Instability and Conservation
Rate Structures. AWWA Research Foundation.

Chesnutt, T.W., Fiske, G. Beecher, J. AWWA Research Foundation (2007). Water Efficiency
Programs for Integrated Water Management. California Urban Water Conservation
Council.

Childers, Hank, Washington County Water Conservancy District Water Treatment Plant
Operator. (2007). Telephone interview with Hank Childers, Water Treatment Plant
Manager. (November, 2007).

City of Georgetown, Texas. (2007). Drought Contingency Plan. City website:


http://www.georgetown.org/departments/gus/water/drought.contingency.plan.php

Conkling, Roger. (2004). Marginal Cost in the New Economy

Contra Costa Water District, (2007). Mission Statement. Web site: http://www.ccwater.com

CUWA (California Urban Water Agencies). (1994). The Value of Water Supply Reliability:
Results of a Contingent Valuation Survey of Residential Customers, prepared by
Bakarat and Chamberlin, Inc., Oakland, CA, August 1994.
70

Drobak, J.N. (1972). Computer Simulation and Gaming: An Interdisciplinary Survey with a
View toward Legal Applications. Stanford Law Review, Vol. 24, No. 4. (Apr., 1972),
pp. 712-729.

Ellis, Scott. (2007) Telephone interview discussing financial impacts of water conservation.
Sandy City Public Utilities, Utah. November, 2007.

Geriani, A.M., Essamin, O., Gijsbers, P.J.A., Loucks, D.P. (1998). Cost-Effectiveness
Analyses of Libya’s Water Supply System.

GoldSim Inc. (2008). GoldSim webpage at www.GoldSim.com.

Hall, D.C., MacEwan, D., Garcia, M., Norris, C. (2006). Final Report. Integrating Marginal
Cost Water Pricing and Best Management Practices. Prepared for Metropolitan Water
District of Southern California

Irvine Ranch Water District (IRWD), (2005). 2005 Urban Water Management Plan. Irvine
Ranch Water District, Irvine, CA. November, 2005.

Jenkins, M.W., Lund, J.R. (2000). Integrating Yield and Shortage Management Under
Multiple Uncertainties. In press, J. of Water Resources Planning and Management,
ASCE, New York, NY). Feb. 28, 2000

Jordan Valley Water Conservancy District. (2007). Jordan Valley Water Conservancy District
Drought Year Water Supply Plan. Power Point presentation. (2006).

King County Department of Natural Resources and Parks (DNRP). Seattle, WA. (2005).
King County Water Supply Planning Process. Planning Framework Summary,
October 31, 2005

Lewis Young Robertson & Burningham, Inc.(LYR&B). (2005). Capital Facilities Paln and
Impact Fee Analysis for Water Facilities for Washington County Water Conservancy
District, Washington County, Utah. (May, 2005).

Little, V. L. Water Conservation Alliance of Southern Arizona. Evaluation and Cost Benefit
Analysis of Municipal Water Conservation Programs. (2002).

Marques, G.F., Lund, J.R., Leu, M.R., Jenkins, M., Howitt, R., Harter, T., Hatchett, S., Ruud,
N., Burke, S. (2006). Economically Driven Simulation of Regional Water Systems:
Friant-Kern, California. Journal of Water Resources Planning and Management.

Mays, L. W. (2002). Urban Water Supply Handbook. McGraw-Hill Companies, Inc., 2002.

Moncur, J.E.T., Pollock, R.L. (1988). Scarcity Rents for Water: A Valuation and Pricing
Model. Land Economics, Vol. 64, No. 1. pp. 62-72.

Mondal, M.S., Wasimi, S.A. (2007). Evaluation of Risk-Related Performance in Water


Management for the Ganges Delta of Bangladesh. Journal of Water Resources
71

Planning and Management, March 2007.

Peter H. Gleick, P.H. (1998). Water in Crisis: Paths to Sustainable Water Use. Bulletin of the
Ecological Society of America. Pacific Institute for Studies in Development,
Environment, and Security, 654 13th Street, Oakland, California 94612 USA

Pulido-Velazquez, M., Andreu, J., Sahuquillo, A. (2006). Economic Optimization of


Conjunctive Use of Surface Water and Groundwater at the Basin Scale. Journal of
Water Resources Planning and Management.

Ramakumar, R. Engineering Reliability: Fundamentals and Applications. Prentice Hall; 1st


edition (December 2, 1996)

Sander, W., (1983). Federal Water Resources Policy and Decision-Making: Their
Formulation Is Essentially a Political Process Conditioned by Government Structure
and Needs. American Journal of Economics and Sociology, Volume 42 Issue 1 Page
1-12, January, 1983.

Sneddon, C., Harris, L., Dimitrov, R., Ozesmi, U. (2002) Contested Waters: Conflict, Scale,
and Sustainability in Aquatic Socioecological Systems. Society and Natural Resources
15 (no. 8, 2002): v. 13, #4.

SNWA (2007). Drough Plan. April 2007.

South Florida Water Management District (2006). 2005-2006 Lower East Coast Water Supply
Plan

Southern Nevada Water Authority (SNWA). (2006). SNWA 2006 Water Resource Plan.

Tabors Caramanis and Associates (TC&A). (1994). Long-Term Water Conservation &
Shortage Management Practices: Planning that Includes Demand Hardening.
California Urban Water Agencies. (June 1994)

Thompson, R. (2003). Water Conservation Plan. Washington Water Conservancy District


Water Line, 2003.

Thompson, R. (2007). Water Conservation, How Long Can We Go? Washington Water
Conservancy District Water Line, Fall, 2007.

Turner Collie & Braden Inc. (TC&B). (2004). City of San Marcos Water Supply Master Plan,
Draft Report, November 2004.

Turvey, Ralph. (1976). Analyzing the Marginal Cost of Water Supply. Land Economics, Vol.
52, No. 2 (May, 1976), pp. 158-168.

UN-Water (2007). Coping With Water Scarcity, Challenge of the Twenty-First Century. World
Water Day, March 2007
72

Utah Dept. of Water Resources (UDWR). (2007). Lake Powell Pipeline General Information
webpage.
http://www.water.utah.gov/LakePowellPipeline/GeneralInformation/default.asp

Utah Division of Water Resources (UDWR). (May, 2001). Utah's Water Resources: Planning
for the Future. Utah Department of Natural Resources, Division of Water Resources.

Utah State University (USU). (2007). Utah Climate Center, GIS Climate Data Search Engine
on website: http://climate.usurf.usu.edu/index.php

Vista Consulting Group, Inc., Chesnutt, T. W., Beecher, J. A. (1996). Managing the Revenue
and Cash Flow Effects of Conservation

Vista Consulting Group, Inc., Chesnutt, T. W., Beecher, J. A. (1997). Long-Term Effects of
Conservation Rates. AWWA Research Foundation.

Washington County Water Conservancy District (WCWCD). (1993). A Long Term


Framework for Water Resources Management, Development and Protection. August,
1993.

Washington County Water Conservancy District (WCWCD). (2002). Conservation Plan for
Washington County Water Conservancy District, Washington County, Utah. (2002)

Watkins, D.W., McKinney, D.C. (1997). Finding Robust Solutions to Water Resources
Problems. Journal of Water Resources Planning and Management, Jan/Feb 1997.

Wilkinson, T. (2002). Roman aqueducts of new West: water pipes. Special report to The
Christian Science Monitor, 2002.

Wilson, B. (2007). Lake Powell Pipeline: Is It Inevitable Too? Newspaper article published
by The Salt Lake Tribune Mar. 18, 2007

Worthington, S., MacPherson, M., Subramanian, S., Somani, M. (year). Marginal Cost
Analysis. University of Calgary

You might also like