An Introduction to Optimization
Models and Methods 4
Water resource systems are characterized by modeling approaches are illustrated by their
multiple interdependent components that toge- application to some relatively simple water
ther produce multiple economic, environmental, resources planning and management problems.
ecological, and social impacts. As discussed in The purpose here is to introduce and compare
the previous chapter, planners and managers some commonly used optimization methods and
working toward improving the design and per- approaches. This is not a text on the state of the
formance of these complex systems must identify art of optimization modeling. More realistic and
and evaluate alternative designs and operating more complex problems usually require much
policies, comparing their predicted performance bigger and more complex models than those
with desired goals or objectives. Typically, this developed and discussed in this chapter, but
identification and evaluation process is accom- these bigger and more complex models are often
plished with the aid of optimization and simula- based on the principles and techniques intro-
tion models. While optimization methods are duced here.
designed to provide preferred values of system The emphasis here is on the art of model
design and operating policy variables—values development—just how one goes about con-
that will lead to the highest levels of system structing and solving optimization models that
performance—they are often used to eliminate will provide information useful for addressing
the clearly inferior options. Using optimization and perhaps even solving particular problems. It
for a preliminary screening followed by more is unlikely anyone will ever use any of the
detailed and accurate simulation is the primary specific models developed in this or other chap-
way we have, short of actually building physical ters simply because the specific examples used to
models, of estimating effective system designs illustrate the approach to model development and
and operating policies. This chapter introduces solution will not be the ones they face. However,
and illustrates the art of optimization model de- it is quite likely water resource managers and
velopment and use in analyzing water resources planners will use these modeling approaches and
systems. The models and methods introduced in solution methods to analyze a variety of water
this chapter are extended in subsequent chapters. resource systems. The particular systems mod-
eled and analyzed here, or any others that could
have been used, can be the core of more complex
4.1 Introduction models needed to analyze more complex prob-
lems in practice.
This chapter introduces some optimization Water resources planning and management
modeling approaches for identifying ways of today is dominated by the use of optimization
satisfying specified goals or objectives. The and simulation models. Computer software is
© The Author(s) 2017 93
D.P. Loucks and E. van Beek, Water Resource Systems Planning and Management,
DOI 10.1007/978-3-319-44234-1_4
94 4 An Introduction to Optimization Models and Methods
becoming increasingly available for solving mutually exclusive alternatives (only one of
various types of optimization and simulation which can be selected) each characterized by a
models. However, no software currently exists time series of benefits and costs. Using various
that will build models of particular water methods involving the discount rate, the time
resource systems. What and what not to include series of benefits and costs are converted to a
and assume in models requires judgment, expe- single net benefit that can be compared with
rience, and knowledge of the particular problems other such net benefits in order to identify the one
being addressed, the system being modeled and that is best. The values of the decision variables
the decision-making environment—including (e.g., the design and operating policy variable
what aspects can be changed and what cannot. values) are known for each discrete alternative
Understanding the contents of this and following being considered. For example, consider again
chapters and performing the suggested exercises the tank design problem presented in the previ-
at the end of each chapter can only be a first step ous chapter. Alternative tank designs could be
toward gaining some judgment and experience in identified, and then each could be evaluated, on
model development. the basis of cost and perhaps other criteria as
Before proceeding to a more detailed discus- well. The best would be called the optimal one, at
sion of optimization, a review of some methods least with respect to the objective criteria used
of dealing with time streams of economic and the discrete alternatives being considered.
incomes or costs (engineering economics) may The optimization methods introduced in the
be useful. Those familiar with this subject that is following sections of this chapter extend those
typically covered in applied engineering eco- engineering economics methods. Some methods
nomics courses can skip this next section. are discrete, some are continuous. Continuous
optimization methods, such as the model defined
by Eqs. 3.1–3.3 in Sect. 3.2 of the previous
4.2 Comparing Time Streams chapter can identify the “best” tank design
of Economic Benefits and Costs directly without having to identify and compare
numerous discrete, mutually exclusive alterna-
All of us make decisions that involve future tives. Just how such models can be solved will be
benefits and costs. The extent to which we value discussed later in this chapter. For now, consider
future benefits or costs compared to present the comparison of alternative discrete plans
benefits or costs is reflected by what is called a p having different benefits and costs over time.
discount rate. While economic criteria are only Let the net benefit generated at the end of time
one aspect of everything we consider when period t by plan p be designated simply as Bp(t).
making decisions, they are often among the Each plan is characterized by the time stream
important ones. Economic evaluation methods of net benefits it generates over its planning
involving discount rates can be used to consider period Tp.
and compare alternatives characterized by vari-
ous benefits and costs that are expected to occur fBp ð1Þ; Bp ð2Þ; Bp ð3Þ; . . .; Bp ðTp Þ ð4:1Þ
now and in the future. This section offers a quick
and basic review of the use of discount rates that Clearly, if in any time period t the benefits
enable comparisons of alternative time series of exceed the costs, then Bp ðtÞ [ 0; and if the costs
benefits and costs. Many economic optimization exceed the benefits, Bp ðtÞ\0. This section
models incorporate discount rates in their eco- defines two ways of comparing different benefit,
nomic objective functions. cost or net-benefit time streams produced by
Engineering economic methods typically different plans perhaps having different planning
focus on the comparison of a discrete set of period durations Tp.
4.2 Comparing Time Streams of Economic Benefits and Costs 95
4.2.1 Interest Rates each amount in the time series to what it is worth
today, its present worth, that is, a single value at
Fundamental to the conversion of a time series of the present time. This present worth will depend
incomes and costs to an equivalent single value, so on the prevailing interest rate in each future time
that it can be compared to other equivalent single period. Assuming a value V0 is invested at the
values of other time series, is the concept of the beginning of a time period, e.g., a year, in a
time value of money. From time to time, individ- project or a savings account earning interest at a
uals, private corporations, and governments need rate r per period, then at the end of the period the
to borrow money to do what they want to do. The value of that investment is (1 + r)V0.
amount paid back to the lender has two compo- If one invests an amount V0 at the beginning
nents: (1) the amount borrowed and (2) an addi- of period t = 1 and at the end of that period
tional amount called interest. The interest amount immediately reinvests the total amount (the
is the cost of borrowing money, of having the original investment plus interest earned), and
money when it is loaned compared to when it is continues to do this for n consecutive periods at
paid back. In the private sector the interest rate, the the same period interest rate r, the value, Vn, of
added fraction of the amount owed that equals the that investment at the end of n periods would be
interest, is often identified as the marginal rate of
return on capital. Those who have money, called Vn ¼ V0 ð 1 þ r Þ n ð4:2Þ
capital, can either use it themselves or they can
lend it to others, including banks, and receive This results from V1 ¼ V0 =ð1 þ r Þ at the end of
interest. Assuming people with capital invest their period 1, V2 ¼ V1 =ð1 þ rÞ ¼ V0 ð1 þ rÞ2 at the end
money where it yields the largest amount of of period 2, and so on until at the end of period n.
interest, consistent with the risk they are willing to The initial amount V0 is said to be equivalent
take, most investors should be receiving at least the to Vn at the end of n periods. Thus the present
prevailing interest rate as the return on their capital. worth or present value, V0, of an amount of
Any interest earned by an investor or paid by money Vn at the end of period n is
a debtor depends on the size of the loan, the
duration of the loan, and the interest rate. The V0 ¼ Vn =ð 1 þ r Þ n ð4:3Þ
interest rate includes a number of considerations.
One is the time value of money (a willingness to Equation 4.3 is the basic compound interest
pay something to obtain money now rather than discounting relation needed to determine the
to obtain the same amount later). Another is the present value at the beginning of period 1 (or end
risk of losing capital (not getting the full amount of period 0) of net benefits Vn that accrue at the
of a loan or investment returned at some future end of n time periods.
time). A third is the risk of reduced purchasing The total present value of the net benefits
capability (the expected inflation over time). The generated by plan p, denoted V0p , is the sum of
greater the risks of losing capital or purchasing the values of the net benefits Vp(t) accrued at the
power, the higher the interest rate compared to end of each time period t times the discount
the rate reflecting only the time value of money factor for that period t. Assuming the interest or
in a secure and inflation-free environment. discount rate r in the discount factor applies for
the duration of the planning period, i.e., from
t = 1 to t = Tp.
4.2.2 Equivalent Present Value
X
V0p ¼ V p ðtÞ=ð1 þ r Þt ð4:4Þ
To compare projects or plans involving different t¼1;Tp
time series of benefits and costs, it is often con-
venient to express these time series as a single The present value of the net benefits achieved
equivalent value. One way to do this is to convert by two or more plans having the same economic
96 4 An Introduction to Optimization Models and Methods
planning horizons Tp can be used as an economic This factor is often used to compute the
basis for plan selection. If the economic lives or equivalent annual end-of-year cost of engineer-
planning horizons of projects differ, then the ing structures that have a fixed initial construc-
present value of the plans may not be an appro- tion cost C0 and annual end-of-year operation,
priate measure for comparison and plan selection. maintenance, and repair (OMR) costs. The
A valid comparison of alternative plans using equivalent uniform end-of-year total annual cost,
present values is possible if all plans have the TAC, equals the initial cost times the capital
same planning horizon or if funds remaining at recovery factor plus the equivalent annual
the end of the shorter planning horizon are end-of-year uniform OMR costs.
invested for the remaining time up until the longer
planning horizon at the same interest rate r. TAC ¼ CRFn C0 þ OMR ð4:8Þ
For private investments requiring borrowed
4.2.3 Equivalent Annual Value
capital, interest rates are usually established, and
hence fixed, at the time of borrowing. However,
If the lives of various plans differ, but the same
benefits may be affected by changing interest
plans will be repeated on into the future, then one
rates, which are not easily predicted. It is com-
need to only compare the equivalent constant
mon practice in benefit–cost analyses to assume
annual net benefits of each plan. Finding the
constant interest rates over time, for lack of any
average or equivalent annual amount Vp is done
better assumption.
in two steps. First, one can compute the present
Interest rates available to private investors or
value, V0p , of the time stream of net benefits,
borrowers may not be the same rates that are used
using Eq. 4.4. The equivalent constant annual
for analyzing public investment decisions. In an
benefits, Vp, all discounted to the present must
economic evaluation of public-sector invest-
equal the present value, V0p .
ments, the same relationships are used even
X though government agencies are not generally
V0p ¼ V p =ð 1 þ r Þ t or free to loan or borrow funds on private money
t¼1;Tp
X ð4:5Þ markets. In the case of public-sector investments,
V p ¼ V0p = 1=ð1 þ r Þt the interest rate to be used in an economic anal-
t¼1;Tp
ysis is a matter of public policy; it is the rate at
which the government is willing to forego current
Using a little algebra the average annual
benefits to its citizens in order to provide benefits
end-of-year benefits Vp of the project or plan p is
to those living in future time periods. It can be
V p ¼ V0p rð1 þ rÞTp = ð1 þ rÞTp 1 ð4:6Þ viewed as the government’s estimate of the time
value of public monies or the marginal rate of
The capital recovery factor CRFn is the return to be achieved by public investments.
These definitions and concepts of engineering
expression rð1 þ rÞTp = ð1 þ rÞTp 1 in Eq. 4.6
economics are applicable to many of the prob-
that converts a fixed payment or present value V0p
lems faced in water resources planning and
at the beginning of the first time period into an
management. Each of the equations above is
equivalent fixed periodic payment Vp at the end
applicable to discrete alternatives whose decision
of each time period. If the interest rate per period
variables (investments over time) are known. The
is r and there are n periods involved, then the
equations are used to identify the best alternative
capital recovery factor is
from a set of mutually exclusive alternatives
CRFn ¼ ½rð1 þ rÞn =½ð1 þ rÞn 1 ð4:7Þ whose decision variable values are known. More
detailed discussions of the application of
4.2 Comparing Time Streams of Economic Benefits and Costs 97
engineering economics are contained in numer- as well as other constrained optimization proce-
ous texts on the subject. In the next section, we dures is highly dependent on the mathematical
introduce methods that can identify the best structure of the model that in turn is dependent
alternative among those whose decision variable on the system being analyzed. Individuals tend to
values are not known. For example, engineering construct models in a way that will allow them to
economic methods can identify, for example, the use a particular optimization technique they think
most cost-effective tank from among those whose is best. Thus, it pays to be familiar with various
dimension values have been previously selected. types of optimization methods since no one
The optimization methods that follow can iden- method is best for all optimization problems.
tify directly the values of the dimensions of most Each has its strengths and limitations. The
cost-effective tank. remainder of this chapter introduces and illus-
trates the application of some of the most com-
mon constrained optimization techniques used in
4.3 Nonlinear Optimization Models water resources planning and management.
and Solution Procedures Consider a river from which diversions are
made to three water-consuming firms that belong
Constrained optimization involves finding the to the same corporation, as illustrated in Fig. 4.1.
values of decision variables given specified Each firm makes a product. Water is needed in
relationships that have to be satisfied. Con- the process of making that product, and is the
strained optimization is also called mathematical critical resource. The three firms can be denoted
programming. Mathematical programming tech- by the index j = 1, 2, and 3 and their water al-
niques include calculus-based Lagrange multi- locations by xj. Assume the problem is to deter-
pliers and various methods for solving linear and mine the allocations xj of water to each of three
nonlinear models including dynamic program- firms (j = 1, 2, 3) that maximize the total net
P
ming, quadratic programming, fractional pro- benefits, j NBj ðxj Þ, obtained from all three
gramming, and geometric programming, to firms. The total amount of water available is
mention a few. The applicability of each of these constrained or limited to a quantity of Q.
Fig. 4.1 Three water-using firms obtain water from a river. The amounts xj allocated to each firm j will depend on the
river flow Q
