Water Resources Systems Modeling
A Model:
A mathematical description of some system.
Model Components:
Variables, parameters, functions, inputs,
outputs.
A Model Solution Algorithm:
A mathematical / computational procedure for
performing operations on the model – for
getting outputs from inputs.
Water Resources Systems Modeling
Model Types:
• Descriptive (Simulation)
• Prescriptive (Optimization)
• Deterministic
• Probabilistic or Stochastic
• Static
• Dynamic
• Mixed
Water Resources Systems Modeling
Algorithm Types:
• Descriptive (Simulation)
• Prescriptive (Constrained Optimization)
• Mathematical Programming
• Lagrange Multipliers
• Linear Programming
• Non-linear Programming
• Dynamic Programming
• Evolutionary Search Procedures
• Genetic Algorithms, Genetic Programming
Water Resources Systems Modeling
Simulation: System Design and
Operating Policy
System WATER RESOURCE System
Inputs SYSTEM Outputs
Optimization: System Design and
Operating Policy
System WATER RESOURCE System
Inputs SYSTEM Outputs
Water Resources Systems Modeling
Modeling Example
• Problem.
Need a water tank of capacity V.
• Performance Criterion.
Cost minimization.
• Numerous alternatives.
Shape, dimensions, materials.
• Best design not obvious.
Water Resources Systems Modeling
Modeling Example Continued
Consider a cylindrical tank V.
having radius R and height H.
Average costs per unit area:
R
Ctop
Cside H
Cbase
Water Resources Systems Modeling
Modeling Example Continued
Model:
Minimize Total_cost (Objective)
subject to: (Constraints)
Volume = (R2H) V.
Total_cost = $_Side+$_Base+$_Top
$_Side = Cside(2RH)
$_Base = Cbase(R2)
$_Top = Ctop(R2)
Water Resources Systems Modeling
Modeling Example Continued
Solution: a tradeoff between cost and volume.
Total
Cost
Tank Volume
1. Optimization Models:
• Mathematical models may be used to identify
management schemes, water development plans and
investments (land, reservoirs, pipes, canals, P.S., etc)
• Economic and other objectives are expressed
mathematically
• The design and operating variables are called decision
variables (variables to be determined) such as
capacities of reservoirs and pipelines, cropping
pattern, location and capacity of flood control
reservoirs along a river, etc.
• To find the optimum plan or management scheme with
decision variables (xj, j = 1, 2, ….., n) or (X as a vector),
which maximize the net-benefits; the objective function may
be expressed as:
Maximize NB (X)
• Physical, technical, legal, financial and other restrictions on
the values of decision variables can be expressed as
constraints as follows:
Gi (X) bi (i =1, 2, …, k)
Gi (X) = bi (i =k+1, …, l)
Gi (X) bi (i =l+1, …, m)
• The problem is to determine the value of each xj, which
maximizes the value of the objective function, while
satisfying each of the constraint equations.
• Solution methods of the above optimization model are
called mathematical programming algorithms.
Mathematical Optimization Models
Linear Models Nonlinear Models
( Objective Function and Constraints )
Real Variables Integer Variables Unconstrained Constrained
[Linear Programming] [Integer Programming]
Nonlinear
Constraints Constraints
[Seperable Programming] Equality Inequality
Constraints Constraints
[Lagrangian] [Nonlinear
Programming]
2. Definitions:
• Feasible Solution: It satisfies every one of the
constraints Well defined objectives
• Optimum Solution: A feasible solution which is better
(larger or smaller) than all of the set of feasible solutions
to which it is being compared
*
• Local Optimum: It is a feasible solution x for which
f ( x ) f (x ) for a maximization problem, or f ( x ) f (x )
* *
for a minimization problem for all x over the closed set
within a distance , 0 for all x
*
• Global Optimum: It is a feasible solution x for which
f ( x *) f ( x) for a maximization problem, or f ( x ) f (x)
*
for a minimization problem for all x over the closed set of
feasible solutions
• Unbounded Solution: A solution without upper or lower
limit
f(x)
Global
max
Local
max
Local
Local max
max
Local
min
Local
min
Global
min
X
a b
