GENETIC ALGORITHM TECHNIQUES FOR

CALIBRATING NETWORK MODELS

Dragan A. Savic and Godfrey A. Walters

D.Savic@exeter.ac.uk
G.A.Walters@exeter.ac.uk

Report Number:95/12
1995

Centre for Systems and Control Engineering,

University of Exeter,
North Park Road,
Exeter, EX4 4QF,
Devon,
United Kingdom.

This work was funded by SERC Grant GR/J09796

Abstract
Computer models for analysing pipe flows and pressures in water distribution networks are in widespread use throughout the world as essential tools for the efficient
operation and improvement of very complex systems. Models invariably incorporate a
number of unknown parameters, the values of which must be chosen so that the modelled performance matches as closely as possible that of the real network. The process
of calibration involves both expensive data collection and a complex parameter optimisation problem.
This report presents novel Genetic Algorithm based parameter calibration procedures
developed to match hydraulic model output with observed data sets.

(Key words: Networks, Calibration, Modelling, Optimization)

TABLE OF CONTENTS
ABSTRACT .............................................................................................................................................I
TABLE OF CONTENTS ...................................................................................................................... II
LIST OF TABLES................................................................................................................................. II
LIST OF FIGURES............................................................................................................................... II
INTRODUCTION .................................................................................................................................. 1
MATHEMATICAL FORMULATION ................................................................................................ 3
STANDARD CALIBRATION PROCEDURES .................................................................................. 5
GENETIC ALGORITHMS AND OPTIMIZATION.......................................................................... 7
GENETIC ALGORITHMS AND CALIBRATION ............................................................................................ 8
GA FOR CONTINUOUS PARAMETER OPTIMIZATION ............................................................................... 9
CASE STUDY....................................................................................................................................... 11
CONCLUSIONS................................................................................................................................... 17
ACKNOWLEDGEMENT ................................................................................................................... 18
REFERENCES ..................................................................................................................................... 18
Appendix A........................................................................................................................... 22
Appendix B........................................................................................................................... 35

LIST OF TABLES
TABLE 1. INITIAL ESTIMATES OF PIPE ROUGHNESS COEFFICIENTS .......................................................... 13

LIST OF FIGURES
FIGURE 1. SUPPLY AND DISTRIBUTION ARRANGEMENTS FOR DANES CASTLE .......................................... 11
FIGURE 2. NODE PRESSURE ERROR FOR THE THREE DEMAND CONDITIONS .............................................. 13
FIGURE 3. PIPE FLOW ERRORS FOR THE THREE DEMAND CONDITIONS ...................................................... 14
FIGURE 4. COMPARISON OF DIFFERENT SOLUTIONS ................................................................................. 15

ii

Introduction
The ability to model larger water distribution systems (WDS) has improved considerably during the past decade[3,16]. Nowadays, it is widely acknowledged that design
and operation of such systems depend critically on the efficiency and accuracy of
mathematical models utilised to model the systems behaviour under a variety of conditions. Before a model is used, it must be adjusted to ensure that it will predict, with
reasonable accuracy, the behaviour of the system it models, i.e., it must be calibrated.
This is widely acknowledged by the research community and several studies on WDS
calibration have been published in the past two decades[4,13,19,30].
The problem of WDS model calibration, even if only for water quantity, (pressures
and flows) is highly complex due to the large number of parameters examined and
non-linear due to the flow equations. Several researchers have addressed this problem
developing methods to minimise the difference between the values of the observed
data and those computed by the network simulation model. These methods are based
on the use of analytical equations[30], simulation models[19], or optimisation techniques[13]. Techniques based on analytical models may be applied to very small networks or may alternatively require large network to be simplified by considering only
the skeleton network. Simulation techniques can handle larger networks but are generally restricted to a single loading condition. The most promising calibration procedures are based on optimisation. However, the success of current methods usually depends on linearizing assumptions or the unrealistic calculation of partial derivatives.
In addition, they are generally local optimisation procedures which tend to become
entrapped in local minima or suffer from numerical instabilities associated with matrix inversion.
Since models capable of simulating the hydraulic behaviour of pipe networks are
complex in terms of size, non-linearity, and discrete nature, the use of analytical
methods or classical optimization techniques requires many simplifications. These in
turn may cause unsatisfactory or unrealistic results. On the other hand, Genetic Algorithms, which belong to a class of stochastic optimisation techniques capable of dealing with complex, multi-modal and discontinuous functions, have the required robustness and efficiency as well as conceptual simplicity to handle the aforementioned

problems. Over the course of the last two decades these computer algorithms have
proved their usefulness in various domains of application[29]. Recently, they have been
applied to a broad spectrum of water resources problems[1,2,14,17,21,23,31,33].
The research described in this report combines theoretical and practical work in modelling (simulation) and Genetic Algorithms (optimisation) to develop novel, efficient
and robust calibration procedures and tools. It is believed that the availability of these
tools and an increased understanding of the data requirements for reliable model construction has great potential benefits. These include improved operation and more
purposeful monitoring of water supply systems, increased quality of supply and ultimately lower costs to water companies and consumers.

Mathematical Formulation
A distribution network may be viewed as a connected graph with arcs representing
pipes and nodes representing network elements like valves, pumps, reservoirs, demand points, etc. Two hydraulic variables are associated with network elements,
namely flows and heads. The following mathematical statement of the problem is presented for a general water distribution network. The equations of the network express
flow conservation at nodes and relations between head losses and heads for arcs:

= ci

(1)

hij = hi h j

(2)

j J (i )

ji

where J(i) is a set of nodes adjacent to node i, ci is the consumption at node i, qji is the
flow from node j to node i, hij is the head loss in the pipe connecting i and j and hi is
the head at node i. The head loss hij to friction associated with flow through a pipe
can be expressed in a general form as:
hij = Rij qij qij

n 1

(3)

where Rij and n (n > 1) depend on the flow resistance law selected. In this work the
Colebrook-White formula is used to calculate resistance coefficient Rij as a function of
the friction factor fij, the diameter dij of the pipe connecting i and j, flow qij and the
pipe length Lij:

Rij = f ij , d ij , qij , Lij

(4)

The friction factor f is a function of the roughness of the pipe k, the diameter d, the
flow q through the pipe and the viscosity of the fluid (which is for this work considered constant).
For specified pipe characteristics, demand patterns and reservoir heads the system of
non-linear equations (1)-(4) has a unique solution defined by the flows and heads in
the whole network. There are several iterative techniques[11] available for solving the
above system which are incorporated into modern simulation tools[22, 34, 35]. These
tools allow network analysts to concentrate solely on building realistic representation
of the water distribution network thus enabling easier development of models. If input

data for the model are correct, then predicted pressures and flows will match observed
values. However, two main sources of problems are associated with data collection for
real networks: (a) not all input parameters are measured directly because of the expense of data collection; (b) even if it is possible to measure all parameters a certain
amount of inaccuracy will still be associated with readings in the field.

Standard Calibration Procedures

Techniques and procedures for constructing a WDS model may vary but in the UK
they are summarised by the Water Research Centre (WRc plc) into the following activities[35]:

inspection of supply, distribution and consumer records and maps,

site inspection of plant and equipment,

preliminary field measurement,

field measurement exercise,

entry of network data for a computer analysis, and

calibration of the network model.

The basic aim of the inspection of supply, distribution records and maps is to select
network data which will justify their inclusion in the model. For example, pipes which
are below certain size are either ignored or grouped together and replaced by equivalent pipes. Since the demand for water is modelled to take place at nodes consumer
records and maps are inspected in order to enable allocation of the total demand to
network demand zones and finally to nodes. The demand allocation is aided by the
field measurement exercise which involves flow measurement of significant demands,
transfers to and from the network and from source works, pump stations and reservoirs. The key to meaningful calibration is having field measurements corresponding
to more than one flow rate. In addition to flows the exercise also entails pressure logging at as many key sites as possible, e.g. at pump stations, in known problem areas,
at large diameter pipes, etc.
Calibration performed using modern simulation packages commences when input data
including an initial estimate of the roughness values of all pipes is entered into the
model. The model is then analysed and the results compared with the field test measurements. Calibrations of this type proceed based on a tedious and time-consuming
trial-and-error procedure where the parameter values are adjusted based on the hydraulic results and the hydraulic analysis is repeated. This iterative process continues
until some stated operating specifications are satisfied or no viable change in input
parameters which improves agreement between observed and predicted values can be
found. In the latter case, the possibility of modelling anomalies, such as reduced pipe

diameter due to internal corrosion, an incorrectly modelled open/closed valve, etc.,

should be investigated.
The above presented calibration procedure is extremely tedious and, even assuming
enough time and resources are given for model building, depends critically on the
analysts skills and his/her understanding of the WDS being studied. Often this procedure will result in a less than optimum solution and may not be effective when a large
number of variables and operating conditions are investigated.

Genetic Algorithms and Optimization

Genetic Algorithms (GAs) are biologically motivated adaptive computer techniques
based on natural selection and genetic operators[8,15,20]. These algorithms are often
used to solve complex optimization problems[29]. There are many variations of GAs
but the following general description encompasses most of the important features.
Consider a GA for a parameter optimization problem where a set of real-valued variables x=(x1, x2,, xn) is to be found which maximise/minimise some objective function
max f ( x )
x

(5)

The analogy with nature is established by the creation within a computer of a set of
solutions called a population. Each individual in a population is represented by a set
of parameter values which completely describe a solution. These are encoded into
chromosomes, which are, in essence, sets of character strings analogous to the chromosomes found in DNA. Standard GAs (SGAs) use a binary alphabet (characters may
be 0s or 1s) to form chromosomes[9]. For example a two-parameter solution
x = (x1,x2) may be represented as an 8-bit binary chromosome: 1001 0011 (i.e., 4 bits
per parameter, x1 = 1001, x2 = 0011). In that particular form the algorithm requires an
additional mapping from bitstrings to real-valued parameters.
The initial population of solutions, which is usually chosen at random, is allowed to
evolve over a number of generations. At each generation, a measure (fitness) of how
good each chromosome is with respect to the objective function is calculated. This is
achieved by simply decoding binary strings into parameter values, substituting them
into the objective function and computing the value of the objective function for each
of the chromosomes. Next, based on their fitness values individuals are selected from
the population and recombined, producing offspring which will comprise the next
generation. This is the recombination operation, which is generally referred to as
crossover because of the way that genetic material crosses over from one chromosome
to another. For example, if two chromosomes are x = (x1,x2) = 1111 1111 and
y = (y1,y2) = 0000 0000, the two offspring may be z = 1100 0000 and w = 0011 1111.

The probability that a chromosome from the original population will be reproduced
into the new generation is dependent on its fitness value. Fit individuals will have
higher probability of being selected than less fit ones. Hence, the new population will
have more of the better solutions. Mutation also plays a role in the reproduction phase,
though it is not the dominant role, as is popularly believed, in the process of evolution.
In SGAs mutation randomly alters each bit (also called gene) with a small probability.
For example, if the original chromosome is x = (x1,x2) = 1111 1111, the same chromosome after mutation may be x = 1110 1111.
In essence, Genetic Algorithms rely on the collective learning process within a population of individuals, each of which represents a search point in the space of potential
solutions. They draw their power from the theoretical principle of implicit parallelism[9]. This principle enables highly fit solution structures (schemata) to receive increased numbers of offspring in successive generations and thus lead to better solutions.
Genetic Algorithms and Calibration
In recent years, many researchers have begun to investigate the use of evolution based
computer methods for calibration of various hydraulic/hydrologic models. Wang[33]
investigated the use of GAs combined with fine-tuning by a local search method for
calibration of a conceptual rainfall-runoff model. Models were calibrated by minimizing the residual variance defined as the sum of square of differences between
computed and observed discharges.
Duan et al.[6] introduced the shuffled complex evolution method for a similar problem by hybridising a genetic algorithm with the Simplex search method. The objective
function used was the mean daily square root of the difference between the observed
flows and simulated flows.
Babovic et al.[1] used GA and the hydrodynamic MOUSE package to fit Manning
numbers to pipes while Mohan and Lucks[17] reported on use of GAs for estimating
parameter values of some linear and non-linear flow routing and water quality prediction models.
Most of those studies[1,17,33] used the GA formulation with binary representation
requiring an additional decoding procedure from bitstrings to real-valued parameters
8

being calibrated. This representation has several advantages over other encodings[5].
It is simple to create and manipulate, it is theoretically tractable, and it is widely applicable since very many problems can be encoded in binary strings. The mapping
from a binary string to a parameter can be accomplished in many different ways but
the precision of the mapping is limited to

x max x min
2n

(6)

where xmin and xmax are the lower and upper bounds on parameter x and n is the length
of the bitstring representing parameter x. To construct a multi-parameter coding one
can concatenate several bitstrings into a single chromosome representing a set of parameters. However, when dealing with a large number of parameters requiring high
accuracy representation, a solution chromosome becomes increasingly long and the
power of the GA search diminishes.

GA for Continuous Parameter Optimization

Instead of working with binary coding and applying problem-independent genetic operators, the size of WDS calibration problems dictates direct representation of decision variables. This simply means that bitstrings of SGA are replaced with real numbers. The change simplifies the algorithm in that no additional mapping is necessary
since these numbers represent unknown parameters of the WDS model. However,
other alterations to the SGA are required. Firstly, random binary initialisation is replaced with random real number initialisation. This is simply achieved by generating
lists of real numbers that fall within parameter limits (xmin, xmax).
The SGA crossover operator may also be used for real number chromosomes since it
does not depend on the representation scheme. Similarly to the SGA recombination
genetic material crosses over from one chromosome to another. Let x = (x1,...,xn) and
y = (y1,...,yn) be the parent chromosomes. Then the offspring z= (z1,...,zn) may be computed by
z i = {x i } or {y i }

(7)

where xi or yi are chosen with some probability of crossover pc. In addition to this, an
operator suitable for continuous parameter optimization may be used. Namely, average crossover[5] takes two chromosomes and produces one offspring that is the result
of averaging the corresponding parameters of two parental chromosomes
zi =

xi + yi
2

(8)

Other crossover operators like extended intermediate recombination and extended line
recombination [18] can also be used.
The mutation operator has been investigated for binary domains by many authors [8]
and there have been many suggestions on how often it should be applied to a chromosome. The authors are of the opinion that that the mutation rate should be inversely
proportional to the number of bits in the chromosome as suggested by Mhlenbein
and Schlierkamp-Voosen[18]. However, an operator analogous to binary mutation,
but suitable for continuous parameter optimization must be used since simple bit inversion is not possible with the floating-point representation.
An obvious way to mutate a real-valued parameter x is to randomly select a number
that falls within parameter limits xm[xmin, xmax]. Alternatively, the new parameter
may be given by
xm = x + z

(9)

where z is a number in the mutation range interval. This range can be a constant value
throughout the evolution process or it may be a function of the generation number. By
exploiting an analogy with annealing processes the range should become smaller with
the evolution process approaching final stages.

10

Case Study
The proposed algorithm is used to provide a calibrated network model for the Danes
Castle Zone of Exeter City (Devon, UK). This network was chosen for the study because it provides a complex calibration problem to solve and because the necessary
input and output data were readily available. Namely, the network model for this zone
was already built in 1991 when South West Water Services Limited (SWW) commissioned Ewan Associates to develop the model [7].

DANES CASTLE
PUMPS

PYNES WATER
TREATMENT
WORKS

500mm delivery
main
6" meter and
strainer

Legend
meter

18"

main

6"
12"

waste
district

18" dedicated
pumping main
10"
DANES CASTLE
SERVICE RESERVOIR

12"
10" into
distribution

18" into
distribution

10"
10"

12"

6"

12"

6"

6"

6"
16"

10"

8"

9"

6"

Figure 1. Supply and distribution arrangements for Danes Castle

The Danes Castle zone is supplied from Pynes Water Treatment Works (see Figure 1).
Water is drawn from the River Exe and after treatment is pumped via a dedicated 18

11

main directly to the Danes Castle service reservoir and via a 12 pumping main directly into distribution.
The skeleton of the network is given in Figure 1, while the full network used in the
original calibration exercise [7] and subsequently in this study consists of 197 nodes
and 242 elements (see Appendix A for the input data). Of the 197 nodes, one is a
fixed-head reservoir (at treatment works) and one is the Danes Castle service reservoir. Of the 242 network elements, two are pumps and three are modelled as throttling
valves. In 1991, the Danes Castle reservoir was known to be in poor condition and it
was reconstructed in 1992/93.
In order to monitor leakage, five waste districts have been set up as shown in Figure 1.
These zones were serving between 1559 and 4498 properties each. Information obtained from these zones were used for demand calculations.
A 48 hour field test was undertaken between 13th and 15th August 1991. Flows were
monitored into or within the system at 15 locations while pressures were monitored at
23 locations including the water level variation at Danes Castle service reservoir and
the pump suction and delivery pressures at Pynes Treatment Works.
Three loading conditions were considered in the analysis:
(a) Peak demand - at 10:00h on the 13th of August.
(b) Average demand - at 16:00h on the 13th of August (see Appendix A).
(c) Minimum (night) demand - at 03:30h on the 13th of August.
The calibration process adopted and carried out by Ewan Associates comprised of the
following tasks:
(1) to match the total system flows;
(2) to assess the predicted and observed total pressures for each snapshot and make
reasoned adjustments to pipe roughness coefficients; and
(3) to report model anomalies.
Initially theoretical roughness values from standard hydraulic tables [10] were adopted
for various materials as in Table 1. Steps (2) and (3) were carried out through a trialand-error procedure based on the consultants experience and knowledge of the system. The results of the calibration[7] in terms of prediction errors, are presented in
Figure 2 and Figure 3.

12

Table 1. Initial Estimates of Pipe Roughness Coefficients

Material Type

k value (mm)

Assumed Pipe Condition

Cast Iron

6.00

80 years of moderate attack

Asbestos Cement

0.15

Good

Ductile Iron

0.15

Good

PVC

0.10

Good

Relined Mains

0.15

Good

It should be noted that the pressure logger at node 5050 failed for the minimum demand loading condition (peak in Figure 2). Resulting pipe roughness coefficients are
given in Appendix B (second row, under the heading Ewan). There were numerous
modelling anomalies reported by Ewan Associates [7] which were not resolved at the
time since they warranted additional investigation work. The anomalies were mainly
caused by flow restrictions (throttling) in particular pipes which may have occurred
because of pipe tuberculation, internal pipe corrosion or even because the situation in

4.00

Node Pressure Error (m)

3.00
Minimum demand
2.00

Peak demand
Average demand

1.00
0.00
-1.00

Node

Figure 2. Node pressure error for the three demand conditions [7]

13

PYP5S

PYP5D

7016

6029

6025

5133

5131

5100

5085

5075

5062

5050

5049

5012

4025

4010

3081

3077

3060

3030

3011

-3.00

PYC4

-2.00

the network had changed since the mains geographic plan was made. Without further
investigation these problems were solved by assigning high k (roughness) values to
those pipes (e.g., pipes 3, 10, 12, 16, etc.).
4.00

Pipe Flow Error (l/s)

2.00
0.00
-2.00
Minimum demand

-4.00

Peak demand
Average demand

-6.00
-8.00
-10.00
5

65

73

118

125

137

158

162

198

207

208

209

217

220

237

Pipe No.

Figure 3. Pipe flow errors for the three demand conditions [7]
Although the number of parameters to be estimated cannot exceed the number of total
observations available for all the loading conditions, the solutions obtained using the
GA technique are based on fitting friction factor values to each of the pipes. This assumption was used for several reasons:
(a) to demonstrate model capabilities to deal with a large number of variables;
(b) to obtain an initial grouping of pipes in absence of detailed knowledge of the age
and the service condition of pipes;
(c) to investigate how different and unrealistic solutions can result from attempting to
acquire more information from collected data than is available.
Once consistent results have been obtained, groups of pipe with similar friction values
can be identified and the GA can be restarted again. Starting from an initial grouping
of pipes based only on nominal pipe diameters may introduce initial bias and prevent
realistic solutions from being found.
The objective function used in this work is:

14

f ( x ) = p1 ( Hio Hip ) 2 + p2 (Q oj Q jp ) 2
i

(10)

where, p1,2 are normalising coefficients, and Hio and Hip are the observed and predicted heads at node i, respectively, and Q oj and Q jp are the observed and predicted
flows through the pipe j.
Since GAs are stochastic-search techniques, solutions obtained running the program
with different random seed values used to initialise the evolution process may be different. Therefore, several GA runs were necessary to ensure that the solutions identified were of good quality. Figure 4 shows results, in terms of sums of squared errors
for both pressure heads and flows in the network, obtained by the original calibration
study [7] and the 10 GA runs.

180.00
160.00

SSD - Q

140.00

SSD -H*

SSD

120.00
100.00
80.00
60.00
40.00
20.00

GA10

GA9

GA8

GA7

GA6

GA5

GA4

GA3

GA2

GA1

Ewan

0.00

Model

Figure 4. Comparison of different solutions

Since information regarding the internal condition of the mains was not available the
initial

values

for

calibration

parameters

were

initially

restricted

to

kmin k kmax. Resulting roughness values obtained by the GA are presented in Appendix B (under the heading kmax=20 mm). However, results obtained in the original

15

study and the GA results indicate that better calibration results may have been obtained if higher roughness values were used. The table in Appendix B shows results
obtained for kmax = 60 mm.

16

Conclusions
Various complex problems that arise in hydraulics and water resources in general have
been solved using evolution-based programs and their hybrids. The applications of
these techniques yield remarkable results with respect to the number of possible solutions an engineer may be faced with when dealing with the design and management of
hydraulic systems. This report describes the development of a nonlinear optimization
model for pipe network calibration. The model developed is based on GAs but departs
from classical GAs in its representation. The use of floating-point representation enables calibration of a large number of unknown parameters without compromising accuracy and precision of the solutions.
The capabilities of the developed model are ascertained using data of an actual water
distribution network. The results obtained by applying the developed model to the
Danes Castle network (Exeter, Devon) clearly show the advantages over trial-anderror procedures used to match hydraulic model output with observed data sets. The
efficiency of the procedure is tested by running the model several times with different
seed values for the random number generator. Each GA run produced a solution better
(with respect to the objective function used in this work) than the original study. The
high level of agreement between the results of different runs also demonstrates the
robustness of the procedure.
With respect to other optimization or analytical models, the GA-based calibration
tool: (a) is easier to use because it does not need complex mathematical apparatus to
evaluate partial derivatives or to invert matrices; (b) can handle larger networks, several loading conditions and a larger number of calibration parameters; and (c) permits
easy incorporation of additional parameter types (pipe diameters, demands, etc.) and
constraints into the optimization process.
It can be anticipated that the number of applications in this area will steadily grow
since GAs are not only effective, they are also easily realisable due to the conceptual
simplicity of the basic mechanisms. Their potential is even greater when parallel
forms of the algorithms can be developed and executed in low-cost multiprocessor
computing systems.

17

Acknowledgement
This work was supported by the U.K. Engineering and Physical Sciences Research
Council, grant GR/J09796. We are also grateful to Ewan Associates for providing us
with the report for the original calibration study and to South West Water Ltd. for
providing us with the data for the Danes Castle network.

References
1. Babovic, V., Larsen, L.C. and Z. Wu, (1994), Calibrating Hydrodynamic Models
by Means of Simulated Evolution, Hydroinformatics 94.
2. Cieniawski, S.E., Eheart, J.W. and Ranjithan, S., (1995), Using Genetic Algorithms to Solve a Multiobjective Groundwater Monitoring Problem, Water Resources Research, Vol. 31, No. 2, 399-409.
3. Coulbeck, B. (ed.), (1993), Integrated Computer Applications in Water Supply,
Proc. of the International Conference on Integrated Computer Applications for
Water Supply and Distribution, Research Studies Press, Taunton, UK.
4. Data, R.S.N., and Sridharan, K., (1994), Parameter Estimation in WaterDistribution Systems by Least Squares, Journal of Water Resources Planning and
Management Div., ASCE, 120(4), 405-421.
5. Davis, L., (1991), Handbook of Genetic Algorithms, Van Nostrand Reinhold, New
York.
6. Duan, Q., Sorooshian, S., and Gupta, V.K., (1994), Optimal Use of the SCE-UA
Global Optimization Method for Calibrating Watershed Models, Journal of Hydrology, 158, 265-284.
7. Ewan Associates, (1991), Network Analysis Report for the Danes Castle Zone of
Exeter City, Vol. I, November 1991, p.29.
8. Goldberg, D.E., (1989), Genetic Algorithms in Search, Optimization, and Machine
Learning, Addison-Weley, Reading, MA.
9. Holland H. J. , (1975), Adaptation in natural and artificial systems, University of
Michigan Press, Ann Arbor.

18

10. Hydraulic Research Wallingford, (1983), Tables for the hydraulic design of pipes
and sewers (fourth edition), Hydraulic Research Station Ltd.
11. Jeppson, R.W., (1983), Analysis of flow in pipe networks, Ann Arbor Science,
Michigan.
12. Khomsi, D., G.A. Walters, A.R.D. Thorley and D. Ouazar, (1996), A Reliability
Tester for Water Distribution Systems, Journal Computing in Civil Engineering,
ASCE, 10(1), 10-19.
13. Lansey, K.E., and Basnet, C., (1991), Parameter Estimation for Water Distribution
Networks, Journal of Water Resources Planning and Management Div., ASCE,
117(1), 126-144.
14. McKinney, D.C. and Lin, M-D., (1994), Genetic Algorithm Solution of Groundwater Management Models, Water Resources Research, Vol. 30, No. 6, 18971906.
15. Michalewicz, Z. , (1992), Genetic Algorithms + Data Structures = Evolutionary
Programs, Springer-Verlag.
16. Miller, D. (edt.), Water Pipeline Systems, Proc. of the 2nd International Conference on Water Pipeline Systems, Mechanical Engineering Publications, London,
UK, 1994.
17. Mohan, S. and Loucks, D.P., (1995), Genetic Algorithms for Estimating Model
Parameters, in M.F. Domenica (ed.) Integrated Water Resources Planning for the
21st Century, ASCE, New York, pp.460-463.
18. Mhlenbein, H. and D. Schlierkamp-Voosen, (1993), Predictive Models for the
Breeder Genetic Algorithm, I. Continuous Parameter Optimization, Evolutionary
Computation, Vol. 1, No. 1, 25-49.
19. Rahal, C.M., Sterling, M.J.H., and Coulbeck, B., (1980), Parameter Tuning for
Simulation Models of Water Distribution Networks, Proc. Instn Civ. Engrs, Part
2, 69, pp. 751-762.
20. Rechenberg, I., (1973), Evolutionsstrategie, Frommann-Holzboog, Problemata 15.

19

21. Ritzel, B.J., Eheart, J.W. and Ranjithan, S., (1994), Using Genetic Algorithms to
Solve a Multiple Objective Groundwater Pollution Containment Problem, Water
Resources Research, Vol. 30, No. 5, 1589-1603.
22. Rossman, L.A., (1993), The EPANET water quality model, in Coulbeck, B. (ed.),
Integrated Computer Applications in Water Supply, Vol. 2, Research Studies
Press, Taunton, Somerset, pp. 79-93.
23. Savic, D.A. and G.A. Walters, (1995), An Evolution Program for Optimal Pressure
Regulation in Water Distribution Networks, Engineering Optimization, Vol. 24, No.
3, pp. 197-219.
24. Savic, D.A. and G.A. Walters, (1995), Genetic Operators and Constraint Handling
for Pipe Network Optimization, Lecture Notes in Computer Science 993,
Evolutionary Computing, Fogarty,T.C. (ed.), Springer-Verlag, pp. 154-165.
25. Savic, D.A., G.A. Walters and J. Knezevic, (1995), Optimal Opportunistic
Maintenance Policy Using Genetic Algorithms, 1: Formulation, Journal of Quality
in Maintenance Engineering, Vol. 1, No. 2, pp. 34-49.
26. Savic, D.A., G.A. Walters and J. Knezevic, (1995), Optimal OpportunisticMaintenance Policy Using Genetic Algorithms, 2: Analysis, Journal of Quality in
Maintenance Engineering, Vol. 1, No. 3, pp. 25-34.
27. Savic, D.A. and G.A. Walters, (1995), Place of Evolution Programs in Pipe Network
Optimization, Integrated Water Resources Planning for the 21st Century, M.F. Domenica (ed.), American Society of Civil Engineers, New York, USA, pp. 592-595.
28. Savic, D.A. and G.A. Walters, (1994), Evolution Programs in Optimal Design of
Hydraulic Networks, in Adaptive Computing in Engineering Design and Control
94, edited by I.C. Parmee, University of Plymouth, Plymouth, UK, pp. 146-150.
29. Savic, D.A. and G.A. Walters, (1994), Genetic Algorithms and Evolution Programs for Decision Support, Proceedings of an International Symposium on Advances in Logistics, edited by J. Knezevic, University of Exeter, United Kingdom,
pp. 72-80.
30. Walski, T.M., (1983), Technique for Calibrating Network Models, Journal of
Water Resources Planning and Management Div., ASCE, 109(4), 360-372.

20

31. Walters, G.A. and Lohbeck, T.K., (1993), Optimal layout of tree networks using
genetic algorithms, Engineering Optimization, 22, pp.27-48.
32. Walters, G.A. and D.A. Savic, (1994), Optimal Design of Water Systems Using Genetic Algorithms and Other Evolution Programs, Keynote paper in Hydraulic Engineering Software V Vol. 1: Water Resources and Distribution, edited by W.R. Blain
and K.L. Katsifarakis, Computational Mechanics Publications, pp.19-26.
33. Wang, Q.J., (1991), The Genetic Algorithm and its Application to Calibrating
Conceptual Rainfall-Runoff Models, Water Resources Research, Vol. 27, No. 9,
2467-2471.
34. Wood, D.J., (1980), Computer Analysis of Flow in Pipe Networks Including Extended Period Simulations, University of Kentucky, Lexington, Kentucky (Revised 1986).
35. WRc Engineering, (1989), WATNET, Analysis and Simulation of Water Networks
and a Guide to the WATNET3 Computer Program, Swindon, UK.

21

Appendix A
Nodes
ID

Elevation

Demand

(m)

(l/s)

3000

10.4

0.438211

3001

8.38

3002

7.92

3003

10.06

1.704568

3004

8.7

1.842017

3010

1.267655

3011

18.99

0.641376

3015

8.23

3020

12.8

1.154087

3025

22

0.428796

3026

23.7

1.154466

3027

24.49

3030

24.49

3035

23

0.125274

3040

12.19

2.166761

3041

20.97

3045

12.19

3050

12.19

0.122498

3051

0.419707

3052

0.082232

3053

0.079397

3054

0.028356

3060

6.07

0.532309

3061

0.04537

3062

6.4

0.079397

3063

6.4

0.053876

3064

6.4

3070

9.67

0.325956

3071

9.67

0.098264

3072

9.5

0.102082

3073

6.4

0.039698

3074

9.5

3075

6.4

0.04537

22

Nodes
ID

Elevation

Demand

(m)

(l/s)

3076

5.7

0.065219

3077

5.25

0.412519

3080

10.5

0.318678

3081

9.08

0.324231

3082

3083

19.81

0.290322

3084

19.81

0.706857

3085

11.89

0.276816

3090

0.393162

3095

6.68

4000

7.5

1.932201

4010

6.89

1.334978

4015

7.32

0.686687

4016

0.50752

4020

7.62

0.732494

4021

0.548623

4022

0.443823

4023

4025

6.83

0.395239

4030

6.53

0.457622

5000

21

0.399252

5005

22.19

0.345727

5010

15.5

0.10029

5012

14.19

0.166443

5015

14.63

0.258059

5020

16.15

0.334326

5021

26

0.254054

5025

14.33

0.305591

5030

14.33

5035

12.19

0.431554

5040

32.31

0.702543

5045

40

5049

43.79

5050

43.68

5060

39.5

0.532204

5062

39.586

0.508377

23

Nodes
ID

Elevation

Demand

(m)

(l/s)

5065

36.25

0.86899

5070

9.8

0.857269

5075

8.41

5080

10.36

0.288257

5085

12.93

5090

39.56

1.148491

5095

38.5

5100

27.43

0.502038

5105

28

1.447096

5110

28

5115

18

1.574201

5120

26

0.697161

5125

13.41

0.767554

5130

9.14

0.220817

5131

29.78

5133

33.08

5135

8.53

5140

7.92

0.766705

5200

54.89

5205

55.21

5210

46

5215

43

5216

43

5220

43

5225

44.5

5226

44.5

5230

33.4

0.331111

5235

38

0.256451

5240

37

0.862208

5241

37

5245

34.5

5246

34.5

5247

34.5

5300

54.89

5301

54.89

5303

55.21

24

Nodes
ID

Elevation

Demand

(m)

(l/s)

5305

34.7

5310

34.7

5315

41

0.295332

5320

41

5325

41

0.099179

5326

34.16

0.102082

6000

24.5

0.250994

6005

24.3

0.108573

6010

21.79

1.408546

6015

19.5

0.340329

6016

19.5

6020

15

6025

19.86

1.228856

6024

19.86

6026

28

0.163063

6027

28

0.146429

6028

21.3

0.303642

6029

28.35

0.072383

6030

21.3

0.170721

6035

21

0.098955

6040

21

0.056042

6041

20.42

0.764201

6042

8.84

0.530838

6043

21.03

0.238455

6045

16.03

1.318714

6050

10.03

0.078609

6054

10

6055

10

1.069331

6060

9.5

6061

7000

19.51

-0.35878

7005

19.2

0.380454

7010

19.2

0.568682

7009

19.2

7011

17.1

0.81789

7015

19.2

0.068484

25

Nodes
ID

Elevation

Demand

(m)

(l/s)

7014

19.2

7016

21.74

0.800935

7017

25

1.021349

7018

25

0.461837

7020

18.38

7021

18.38

0.22786

8000

10.4

8002

15

0.02

8004

15.24

8006

15.24

8008

34.444

8100

8104

61.3

8106

61.3

8110

61.3

8112

61.3

8114

7.5

1.93

8116

7.5

1.93

8118

60.01

8122

60.01

8124

34.44

8130

15

8140

6.72

4.15

8150

16

8160

36

8200

16

8202

16

8204

16

8206

16

8208

15.8

8210

27.1

0.103681

8212

20

8214

18.26

8216

15.8

8218

17.18

8220

17

26

Nodes
ID

Elevation

Demand

(m)

(l/s)

8222

17.17

8224

17.19

8226

17

8228

17.17

8230

16

8232

16

8234

16

8236

16.5

8238

13.78

8240

13.93

8300

34.44

1.79

8310

12

1.87

8312

8.53

8314

7.5

8316

17.5

0.28

8318

9.75

8320

2.27

Pipes
ID

Head Node Tail Node

Length

Diameter

(m)

(mm)

(mm)

3000

3001

110

203

1.5

3000

3002

100

300

0.5

3000

3010

290

254

70

3000

3010

290

102

10

3000

8000

10

254

10

3002

8314

315

300

0.5

3003

3004

410

100

3003

3020

130

300

0.5

3003

8314

420

300

0.5

10

3010

3011

590

102

80

11

3010

3015

500

152

20

12

3010

3020

760

305

70

13

3015

8314

10

152

20

14

3020

3025

300

152

1.5

27

Pipes
ID

Head Node Tail Node

Length

Diameter

(m)

(mm)

(mm)

15

3020

3030

360

305

60

16

3020

3035

420

229

100

17

3025

3026

180

152

1.5

18

3025

3027

190

152

1.5

19

3030

3035

75

305

30

20

3030

3040

1170

406

20

21

3040

3041

200

229

1.5

22

3040

3045

200

229

35

23

3045

3050

30

152

35

24

3045

3060

520

229

45

25

3050

3051

720

152

35

26

3050

3070

440

152

40

27

3051

3052

190

152

10

28

3051

3052

190

102

10

29

3052

3053

105

102

10

30

3052

3063

330

152

10

31

3053

3054

200

150

10

32

3053

3064

315

152

10

33

3060

3061

50

152

10

34

3060

3063

405

152

10

35

3061

3062

315

152

10

36

3061

3073

320

152

30

37

3062

3063

200

152

10

38

3062

3076

405

152

10

39

3063

3064

150

102

10

40

3070

3080

110

152

30

41

3070

8318

100

152

10

42

3071

3072

295

152

10

43

3071

3072

305

152

10

44

3071

8318

10

152

10

45

3072

3073

185

152

35

46

3072

3074

190

152

35

47

3072

3077

300

152

40

48

3073

3075

100

152

10

49

3074

3075

185

152

10

50

3074

3076

150

152

10

28

Pipes
ID

Head Node Tail Node

Length

Diameter

(m)

(mm)

(mm)

51

3075

3076

295

152

10

52

3076

8114

95

152

90

53

3077

8100

1250

180

0.05

54

3077

8100

1210

180

0.05

55

3077

8116

490

152

40

56

3080

3081

295

102

10

57

3080

3085

105

152

25

58

3081

3082

85

76

59

3082

3083

405

150

0.1

60

3083

3084

100

76

61

3084

3085

250

102

62

3085

3090

375

152

20

63

3090

3095

1015

152

20

64

4000

4010

300

152

1.5

65

4000

8314

50

102

30

66

4010

4015

160

152

67

4015

4016

505

102

68

4015

4020

105

152

69

4020

4021

185

152

1.2

70

4020

4025

310

152

1.2

71

4021

4022

40

152

1.2

72

4021

4023

40

152

1.2

73

4023

8312

100

150

1.2

74

4025

4030

370

152

1.2

75

4030

8140

300

152

1.2

76

5000

5005

385

305

40

77

5000

8216

190

305

40

78

5005

5010

400

305

40

79

5010

5012

95

305

40

80

5010

8150

20

152

81

5012

5015

395

305

40

82

5015

5020

100

305

650

83

5015

5025

695

254

55

84

5020

5021

205

305

650

85

5020

8002

152

1.5

86

5021

8104

1090

305

650

29

Pipes
ID

Head Node Tail Node

Length

Diameter

(m)

(mm)

(mm)

87

5025

5030

40

150

25

88

5025

5050

505

254

60

89

5025

5070

950

254

40

90

5030

5035

100

150

20

91

5035

5040

300

150

0.1

92

5035

8310

50

152

0.1

93

5040

5045

315

150

0.1

94

5040

5065

360

150

0.1

95

5045

5050

150

457

12

96

5045

5060

30

254

13

97

5045

5226

500

305

98

5049

8110

270

457

99

5049

8206

3000

457

0.75

100

5050

5200

125

457

101

5060

5062

245

254

13

102

5062

5095

400

254

103

5070

5075

180

254

40

104

5075

5080

200

254

105

5075

5125

380

150

10

106

5080

5085

45

254

107

5080

5130

195

300

0.5

108

5080

8000

100

381

10

109

5085

5090

400

152

0.5

110

5090

5095

110

178

111

5090

5100

240

152

112

5095

5100

200

102

113

5100

5105

170

150

114

5105

5110

50

150

115

5110

5115

90

150

116

5110

5125

305

152

0.5

117

5110

5125

265

254

0.3

118

5110

5131

650

254

0.1

119

5115

5120

705

102

120

5120

5241

400

102

121

5125

5130

125

254

0.1

122

5130

5135

100

300

0.5

30

Pipes
ID

Head Node Tail Node

Length

Diameter

(m)

(mm)

(mm)

123

5131

5133

20

305

0.1

124

5131

5226

820

305

1.5

125

5133

5235

350

305

0.1

126

5135

5140

215

150

127

5135

8312

105

300

0.5

128

5200

5205

490

254

129

5200

5300

457

130

5205

5210

190

254

131

5210

5215

40

254

132

5215

5216

290

150

133

5215

5220

40

254

134

5220

5225

340

203

135

5220

5320

15

203

136

5225

5226

10

152

137

5225

5230

290

203

55

138

5225

5325

100

203

139

5230

5235

350

203

55

140

5235

5240

350

305

141

5235

5241

400

203

142

5240

5245

600

305

0.5

143

5241

8160

150

203

144

5245

5246

10

203

0.5

145

5245

8008

10

254

146

5245

8300

185

305

0.5

147

5247

8124

10

203

0.5

148

5300

5301

10

457

149

5300

5305

610

254

150

5301

5303

405

152

151

5301

8118

457

152

5305

5310

100

254

153

5310

5315

40

254

154

5315

5320

25

203

155

5320

5325

100

203

120

156

5325

5326

300

152

157

6000

6005

300

152

158

6000

8160

395

152

31

Pipes
ID

Head Node Tail Node

Length

Diameter

(m)

(mm)

(mm)

159

6005

6010

535

152

160

6010

6015

480

152

10

161

6010

6016

400

254

40

162

6010

8008

525

254

1.5

163

6015

6016

152

1.5

164

6015

8004

102

25

165

6015

8006

102

166

6015

8130

315

102

10

167

6015

8316

55

102

10

168

6016

6020

400

254

40

169

6020

6025

320

203

40

170

6025

6024

152

10

171

6025

6026

300

150

60

172

6025

6035

605

203

10

173

6024

6030

45

102

10

174

6024

8130

500

102

10

175

6026

6027

70

102

60

176

6027

6028

255

102

60

177

6027

6029

200

102

60

178

6028

6029

300

102

60

179

6028

6030

80

102

60

180

6030

6035

485

102

1.5

181

6035

6040

90

203

1.5

182

6035

6043

125

102

1.5

183

6040

6041

45

102

1.5

184

6040

6045

420

152

1.5

185

6041

6042

495

102

1.5

186

6041

6045

400

102

1.5

187

6042

8320

300

102

1.5

188

6043

6054

1000

102

1.5

189

6045

6050

300

102

1.5

190

6045

6050

320

152

1.5

191

6050

6054

110

152

1.5

192

6050

6055

215

102

1.5

193

6054

6055

100

152

1.5

194

6055

6060

575

150

1.5

32

Pipes
ID

Head Node Tail Node

Length

Diameter

(m)

(mm)

(mm)

195

6060

6061

150

150

1.5

196

7000

7005

420

152

197

7000

7014

130

152

198

7000

8124

450

203

199

7005

7010

60

102

200

7010

7011

605

102

201

7010

7016

275

102

202

7009

7014

245

102

203

7015

7016

100

102

10

204

7015

7017

385

102

10

205

7016

7018

300

102

206

7016

7021

500

80

207

7017

8004

300

102

15

208

7018

8006

300

102

209

7020

8130

360

80

80

210

8000

8312

125

300

0.1

211

9000

8106

305

212

8104

8106

305

213

9000

8112

457

214

8110

8112

457

215

8114

8116

102

1.5

216

8118

8122

457

217

8120

8122

457

218

8200

8204

152

0.1

219

8200

8238

500

457

220

8202

8216

305

221

8204

8240

500

457

5.5

222

8208

8210

410

102

223

8208

8216

10

102

224

8210

8212

195

102

225

8210

8214

830

102

226

8218

8220

500

227

8220

8226

500

228

8220

8236

500

229

8222

8242

500

230

8224

8226

500

33

Pipes
ID

Head Node Tail Node

Length

Diameter

(m)

(mm)

(mm)

231

8228

8242

500

232

8230

8232

450

233

8230

8234

450

234

8230

8236

130

500

235

8232

8234

450

236

8234

8238

25

457

237

8234

8240

25

457

243

8102

8120

500

1E-07

244

8108

8120

500

1E-07

34

Appendix B
Roughness Values (mm)
kmax = 20 mm

kmax = 60 mm

Pipe Ewan
No.

GA

GA

GA

GA

GA

GA

GA

GA

GA

Run

Run

Run Run

Run

Run

Run

Run

Run Run

1.5 12.1 14.9 12.8

0.5 20.0 19.9 20.0 20.0 20.0 41.5 59.9

GA

3.1 14.0 42.8 39.0 34.5 12.7 42.6

8.9 13.9 59.5

70.0 20.0 20.0 20.0 20.0 20.0 60.0 60.0 60.0 59.9 59.9

10.0

10.0 20.0 20.0 20.0 19.9 20.0 59.9 59.8 55.9 59.3 59.8

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

0.5 20.0 20.0 20.0 20.0 20.0 15.7

1.0

0.5 19.8 19.9 19.8 19.9 19.7 59.7 59.7 59.3 59.7 60.0

0.5 20.0 20.0 20.0 20.0 20.0 60.0 60.0 59.9 59.9 60.0

10

80.0 20.0 20.0 20.0 20.0 20.0 23.7 23.9 25.0 23.9 24.9

11

20.0 19.6 18.8 19.9 18.3 18.6 56.2 59.7 59.6 58.5 49.9

12

70.0 20.0 20.0 20.0 20.0 20.0 59.9 59.9 59.8 59.9 59.9

13

20.0

14

1.5

7.6 15.6

6.3 36.4 32.6 14.0

7.0 14.7 12.5 14.8 42.3 46.2 25.9 57.7

7.9 13.8 11.2 13.2

6.4 31.4 49.4 38.8 43.3 41.3

5.4

8.7 17.7 30.3 33.9 26.4 36.4

4.7 11.6 11.0

15

60.0 20.0 20.0 20.0 20.0 20.0 60.0 59.9 59.9 59.9 60.0

16

100.0 19.9 19.9 19.8 19.9 19.8 59.7 59.9 59.7 60.0 59.8

17

1.5

7.2

5.1

3.9 15.5 11.9 29.8 43.8 31.7 11.2 25.2

18

1.5

6.3

6.2

9.0

4.8 11.9 30.4 22.0 34.7 25.1 29.7

19

30.0 17.6 17.7 18.3 16.0 10.9 50.8 57.8 47.6 52.6 52.8

20

20.0 19.1 18.1 20.0 19.3 19.0

21

1.5

3.9

6.5

6.6 10.7

0.5

0.5

0.1

0.2

0.5

5.7 39.7 28.4 27.5 26.7 20.7

22

35.0 19.6 20.0 19.9 19.3 19.8

0.9

0.6

0.1

23

35.0 19.8 19.8 19.6 19.7 19.8 21.5 23.3

8.0

5.6 30.4

24

45.0

0.0

0.0

25

35.0 18.1 18.4 18.8 19.0 17.3 46.0 47.9 15.5 30.6 42.9

26

40.0 19.9 19.9 19.9 20.0 19.9

27

10.0 14.7 10.6 12.2 12.7

28

10.0

7.9

8.3

8.8 11.7

5.7

2.6

8.9

5.9

0.5

0.1

6.2

0.1

4.1 23.7 31.9

0.8

0.0

8.2

8.3 35.4 11.0 31.7 24.8 38.2

9.3 19.4 40.6 42.5 27.7 45.9 34.7

35

29

10.0 12.1 17.9 12.8

9.1 13.1 46.5 45.7 54.6

7.5 19.5

30

10.0

3.6

6.0 17.2

8.1 14.9

31

10.0

8.1 12.7 15.9

32

10.0 11.5 15.2 16.4 10.6 13.2 37.5 19.0 32.9 41.6 28.0

33

10.0 19.9 19.6 19.9 19.8 20.0 59.8 56.9 46.8

34

10.0 19.7 19.8 19.8 19.5 18.9 56.7 59.2 53.8 55.4 52.5

35

10.0 17.4 16.7 18.6 17.6 18.3 57.2 57.2 36.1 56.1 52.6

36

30.0 19.9 19.9 20.0 20.0 19.7 55.6 58.8 31.5 24.1 29.7

37

10.0 19.1 20.0 16.8 18.9 17.4

38

10.0 19.8 19.8 19.8 20.0 19.8 29.1 52.1

39

10.0

9.2 10.3

8.6 12.8 11.6 43.5 36.3 34.0 13.5 31.7

40

30.0

6.2

2.5

41

10.0 12.5 13.1 19.6 11.5 18.9

42

10.0

0.9

8.5

4.5

1.0

1.1 32.4 11.6

4.9

5.9 26.7

43

10.0

2.6

0.7

0.9

3.4

7.8 32.1 31.4

5.1

4.1 12.8

44

10.0

3.0

5.3

5.1

7.7

3.9 37.0 12.1 22.5 19.4 46.1

45

35.0 19.7 19.2 19.4 18.9 19.3 44.7 42.2 42.1 47.9 57.1

46

35.0 14.6 15.8 17.6 17.9 15.2 37.4 45.7 53.3 46.8 51.9

47

40.0 19.7 19.3 19.7 18.8 19.9 47.0 35.0 48.7 35.1 35.1

48

10.0 19.0 19.7 15.6 18.5 16.3 41.8 40.5 27.3 24.2 56.3

49

10.0 18.3 19.5

50

10.0 11.0

51

10.0

52

90.0 19.6 19.3 19.2 19.7 19.2 10.6 16.7

4.0 16.6

6.0

7.4

7.8

7.3 37.9

6.0 14.2

9.7

2.5 25.3 52.8

9.0 30.1 35.1 31.5

8.4 27.7 21.7

0.6

1.5

7.6 54.6

4.4

3.9

9.5 22.3 43.9

0.2

1.2

1.7

2.3 10.9 57.7 46.9

9.3

9.5 35.9 21.9 21.9

7.4 49.9

6.0 11.4 18.4 12.3 17.2 46.9 52.6 20.6 47.9

8.8 16.1 13.6 13.2 10.1 30.6 42.9 44.0 35.3 37.1

53

0.1

6.3 10.3

54

0.1 14.3

8.4

3.1

4.9

8.1 46.1 15.2

9.7 40.6 50.8 57.2 21.9 32.1

4.4 15.1 12.8 28.7 36.6 48.7 45.8 16.0

55

40.0 18.1 19.1 18.8 18.8 19.6 57.8 53.3 54.8 58.3 58.6

56

10.0 14.0

57

25.0

58

5.9 11.7 10.3

6.6

6.8

7.7 10.3 11.6 10.1

8.9

0.7 42.0 23.9 24.4 16.8

1.0

6.4

4.1

7.6 50.2 31.7 51.0 51.9 42.1

59

0.1

6.4 12.8 10.7 13.5

6.1 21.3 44.3 51.9 26.8 17.3

60

1.0

8.5

61

1.0

6.9 14.6 13.5

5.2

5.4

4.3

7.1

3.4

0.5

0.7

9.7 12.1 50.4 40.8 18.4 33.0 23.6

6.0

6.8

3.0 15.1 17.8

6.8 24.6 24.4

2.9 13.8

62

20.0 10.6 18.1

9.3

3.7

63

20.0

4.5

9.6 12.6 30.1 48.6 39.8 34.7

64
65

7.9 13.8

1.0

8.7 30.4 41.2

8.5

1.5 19.5 19.6 19.4 19.2 19.7 59.8 59.5 59.2 60.0 59.7
30.0

6.8

3.5

4.5

5.2

4.0

0.0

36

0.0

0.0

0.0

0.0

66

1.0 18.7 18.1 19.5 16.5 15.3 59.8 59.2 59.1 59.8 59.4

67

1.0

68

1.0 19.4 19.1 19.7 19.5 19.8 59.8 59.7 59.8 60.0 60.0

69

1.2

1.4

0.0

1.2

1.1

1.6

0.3

0.5

0.2

0.2

0.5

70

1.2

2.7

4.3

3.5

3.4

4.0

3.3

4.4

3.5

3.0

4.0

71

1.2

6.0 10.8

7.1

8.6 15.9 31.6 12.0 10.6 36.7 27.1

72

1.2

1.1

5.8

3.8

2.0

0.4

2.6

1.8

5.2

8.0

2.2

73

1.2

0.5

0.3

0.0

0.3

0.2

0.7

0.3

0.7

0.4

0.7

74

1.2

6.5 12.0 11.8 10.4

75

1.2 14.2 13.7

2.8 10.5

8.4 11.6

5.4 15.6

7.6 51.9 12.8 10.0 47.9 23.4

5.0 17.1 46.9 14.8 43.0 12.8

8.6 23.3 33.1 28.5 30.1 26.7

76

40.0 20.0 20.0 20.0 20.0 20.0 59.9 60.0 60.0 60.0 60.0

77

40.0 20.0 20.0 20.0 20.0 20.0 60.0 60.0 60.0 60.0 60.0

78

40.0 20.0 20.0 20.0 20.0 20.0 60.0 60.0 60.0 60.0 60.0

79

40.0 19.9 20.0 19.9 19.9 19.9 59.6 59.9 60.0 59.8 59.9

80

1.0

9.4

7.3

4.8 12.7 12.0 25.0 33.4 17.5 31.5 21.7

81

40.0 20.0 20.0 20.0 20.0 20.0 60.0 60.0 60.0 60.0 60.0

82

650.0 19.7 19.7 19.5 19.9 19.7 59.6 59.8 59.8 59.2 59.9

83

55.0 18.5 18.6 18.3 18.7 18.9 16.3 16.8 16.5 16.5 16.3

84

650.0 19.9 19.8 19.9 19.9 19.9 59.9 59.9 59.8 59.9 59.9

85
86

1.5 15.0 12.0 11.2 13.0

5.0

8.5 40.0 39.0 34.1 21.9

650.0 20.0 20.0 20.0 20.0 20.0 60.0 60.0 60.0 60.0 60.0

87

25.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

88

60.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

89

40.0 20.0 20.0 20.0 20.0 20.0 54.7 43.0 58.4 59.1 48.5

90

20.0 11.5

91

0.1 12.8

9.1

92

0.1

8.0 13.1

6.8 15.0 19.0 29.7 14.6 19.9 53.3

93

0.1 13.1

4.5

6.3 15.4 41.7 26.7 45.1 45.3 18.9

94

0.1 13.1 13.6 10.9

4.7

2.8 12.0 11.5

5.0 21.8 35.9 25.2 38.8 23.9

6.0 13.8 14.9 31.4 19.3 40.5 32.7 28.6

6.2

8.5

3.0 42.3 40.1 10.0 34.5 37.8

95

12.0 20.0 20.0 19.9 20.0 20.0 38.5

0.4 14.1 19.7 31.0

96

13.0 20.0 20.0 20.0 19.9 19.9

0.0 16.7

0.1 30.2

8.6 11.0 16.9

4.8 25.8

97

2.0

98

1.0 19.3 18.2 16.0 19.9 19.1 42.6 21.3 28.4 16.6 43.2

99

0.8

3.5

0.2

0.0

2.2

0.1

0.6

0.2

0.1

0.0

0.4

100

5.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

101
102

7.9

8.1

8.5

8.1

8.6

0.1

13.0 20.0 20.0 20.0 20.0 20.0 31.1 45.6 28.2 28.7 28.1
4.0 20.0 20.0 20.0 20.0 20.0

0.0

37

6.1 12.1

2.4

6.5

103

40.0 20.0 20.0 20.0 20.0 20.0 60.0 59.9 59.7 59.9 59.7

104

5.0 20.0 20.0 20.0 20.0 20.0 60.0 59.9 60.0 59.9 60.0

105

10.0 20.0 20.0 19.9 19.9 19.8 59.7 59.6 59.7 59.4 59.9

106

1.0 18.8 19.5 19.9 19.6 19.5 60.0 60.0 59.9 60.0 59.7

107

0.5

108

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

10.0 20.0 20.0 19.9 20.0 20.0 59.9 59.8 58.1 59.0 59.8

109

0.5 20.0 20.0 20.0 20.0 20.0

110

1.0 20.0 20.0 20.0 20.0 20.0 20.2 17.2

111

4.0 19.9 19.9 19.9 19.9 20.0 50.7 30.0 38.0 56.0 59.5

112

4.0 20.0 20.0 20.0 20.0 20.0

0.0

0.1

113

1.0 20.0 20.0 20.0 20.0 20.0 60.0 60.0 59.9 59.8

5.8

114

1.0 19.9 20.0 19.8 19.9 19.8 59.8 59.9 59.0 59.7 59.9

115

3.0

0.0

2.0

4.0

0.7

3.7

0.4

0.5

2.2

3.0

1.7

116

0.5

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

117

0.3

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

118

0.1 13.1 13.3 13.0 12.8 12.9

9.7

9.5

8.6

8.1

6.7

119

4.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

120

4.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

121

0.1

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

122

0.5 17.1 17.7 16.3 17.9 19.3 59.5 59.8 59.6 57.4 59.7

123

0.1

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

124

1.5

0.0

0.0

0.0

0.0

0.0

0.1

0.0

0.0

0.0

0.0

125

0.1

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

126

1.0 16.9 13.5 13.8

7.7

6.4 43.7 34.0 28.4 34.9 14.4

127

0.5

8.7 18.6 58.4 58.9 56.1 56.5 57.8

128

1.0 18.9 18.4 19.4 19.0 18.8

7.9 20.1 10.8 47.0

5.4

129

2.0

0.0

0.0

130

2.0 19.7 16.4 17.6 14.6 18.2 38.8 38.3 25.6 29.0 21.4

131

2.0 15.0 13.5 18.1

8.4 16.6 38.0 54.4 31.5 49.8 22.6

132

2.0 11.9

5.5 19.8

7.7 11.4 52.2 21.9 15.5 23.4 50.4

133

2.0 11.2

9.5 16.3 15.3 13.2 54.3 23.2 19.6 32.4 56.6

134

2.0

1.0

135

2.0 14.7 13.4 11.3 15.3 12.9 32.4 11.1

9.4 22.9 19.5

136

2.0

0.0

137
138
139

8.4 14.8 14.7

0.0

1.0

0.0

0.0

0.0

0.0

1.0

0.0

0.0

1.0

0.0

0.0

1.0

0.0

0.0

0.4

1.0

0.0

0.0

0.0

0.0

1.0

0.0

0.0

0.0

1.7

0.0 11.1

0.0

2.2

0.0

1.0

0.0

1.0

0.0

1.0

0.0

55.0 20.0 20.0 20.0 20.0 20.0 40.1 30.5 22.6 24.1 37.0
1.0 19.8 19.4 19.7 19.6 19.7

0.6

7.6

0.1 40.4

4.5

55.0 20.0 20.0 20.0 20.0 20.0 57.0 47.7 57.0 53.8 48.6

38

140

1.0 17.6

8.5

5.1

4.8

7.6

1.0 10.9

0.2

0.0

5.2

141

5.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

142

0.5

7.8 17.4 12.3 14.6 14.0 15.9

143

5.0 19.0 14.2 16.1 17.0 18.7 44.0 59.9 58.3 59.4 59.5

144

0.5

145

1.0 10.0 10.9 13.7 12.8 15.0 58.7 56.3 59.4 51.5 58.6

146

0.5

6.1 10.2

8.7 10.8 14.4 35.4 43.0 13.8 29.8 20.0

147

0.5

9.7

7.6

6.7 12.6 11.8 38.4 40.8 20.8 34.9

7.1

148

2.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

149

1.0 18.6 19.9 18.7 19.8 18.5

4.5

3.6

0.2

6.6

0.7

150

1.0

6.9

2.5

8.0

4.6

6.0 36.0 30.9 44.8 17.0 24.3

151

1.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

152

1.0 16.4 14.8 14.4 18.1 14.4 39.5 50.0 21.6 35.1

1.9

153

1.0 10.0

3.3

154

1.0 18.4 15.0

155

9.8 18.7

9.6

7.7 11.0 11.7

0.0

0.0

4.9 16.1 25.2 17.4

6.0 43.6 28.2 10.5 15.5

0.0

0.0

0.0

2.5 13.6 14.3 13.4 52.3 11.2 37.7

9.4 17.2 10.3 47.8 32.4 34.9 34.6 11.0

120.0 19.2 19.7 19.4 20.0 19.8

9.4 42.4

0.1

156

1.0

5.8 10.0

157

3.0

2.4

158

3.0 20.0 17.7 16.4 15.6 12.0 13.1

1.2 11.2 18.5 13.9

159

3.0

0.0

0.1

5.5 14.3 15.9

2.6 14.2

2.9 11.1

0.0

0.1

8.1 15.2

0.0

0.0

6.2 44.8 29.9

8.5 37.4

0.0 32.4

0.4

0.0

0.0

0.0

0.0

2.9

0.0

160

10.0 11.4 11.2

9.1 14.8

161

40.0

7.1 14.5

6.0 11.9 14.7

8.5

1.4 18.5

1.7

2.9

162

1.5

2.7

0.2

5.0

2.0

1.1

0.3

1.2

1.8

0.4

163

1.5

1.3

1.8

2.6

2.4

5.0

1.9 13.0

5.0 14.4

1.1

6.4 13.0

7.3

7.8 29.9 30.1 42.6 53.5 41.2

0.6

0.2

0.2

164
165

25.0 12.4
1.0

0.3

0.3

7.4 55.1 51.8 57.7 50.4 59.8

7.3

2.8

0.3

2.8

0.8

3.5

166

10.0 18.8 19.4 18.8 19.1 18.0 10.9 38.4 21.7 30.5 58.6

167

10.0

168

40.0 19.5 18.9 18.8 19.8 19.0 24.8 55.5 31.5

169

40.0 19.8 19.9 19.9 19.9 19.7 37.0 33.8 20.6 33.5 16.5

170

10.0

171

60.0 19.5 17.3 17.0 15.7 17.3 46.9 54.7 52.5 54.0 13.2

172

10.0 17.4 18.8 16.8 16.3 18.4 23.5 23.4 48.4 46.1 37.5

173

10.0 19.1 19.1 17.9 18.8 17.5 56.9 56.0 46.1 22.9 48.3

174

10.0 15.6 15.8 18.1 13.5 16.2

175

60.0 16.7 17.5 15.9 19.2 15.6 49.9 44.1 51.9 53.9 43.6

176

60.0 10.9 10.8

5.8 15.7

2.5

9.6

7.2

1.2 10.4 45.5 23.6 34.4 32.1 41.4

7.0 13.0

9.1

1.1 35.9

4.1 39.3 37.8 26.3 28.1 11.9

8.1

5.9 23.2

7.7 15.2 53.4 51.5

39

7.9 17.5

6.6 21.5

2.5

177

60.0 13.0 12.7 17.4 15.8

8.8 54.8 47.2 40.5 53.1 52.5

178

60.0

6.3

0.2

5.8

2.2

0.9

7.0

5.9

2.3 20.1

3.3

179

60.0

6.4

6.3

5.1

3.2

2.6

1.7 22.9

6.4 32.2

8.3

180

1.5

0.0

0.1

0.0

0.0

0.1 44.9 43.1 43.6 17.7 20.9

181

1.5

2.3

7.1 13.6

182

1.5 12.0 10.3

6.4 13.0 10.8 36.0 34.5 40.7 38.7

8.1

183

1.5

6.5 10.8

7.8

184

1.5 14.6

7.9 14.5 12.7

185

1.5

9.7

3.4 15.5 14.7 12.7 29.5 23.8 48.9 48.9 31.0

186

1.5 11.5

3.7 16.2 14.0 10.7 53.2 24.8 33.4 50.3 25.4

187

1.5

6.0

1.6

188

1.5

5.5

8.3 15.5 15.2 10.0 32.9 27.9 14.8 19.5 39.9

189

1.5

7.4

7.8 10.5

190

1.5

3.5 12.2

191

7.4

4.4

5.0

9.3 13.9 33.6 32.9 28.1 29.8 15.8

8.5

9.8 42.2 42.6 17.1 40.8

7.5 32.2 34.6 22.2 13.7 27.6

6.5 38.5 18.1 30.9 16.0 15.0

5.5

5.9 28.6 28.2

4.2

6.6

5.9 46.0 12.7 21.4 26.9 30.8

1.5

7.9 11.2 13.6

4.4

3.8 38.9 35.5 35.6 46.1 28.2

192

1.5

3.5 13.1

4.0 10.9 18.3 28.3 42.0 14.1 59.2

193

1.5 14.3 14.9 13.9

194

1.5 10.1

9.6

8.1 18.1 17.8 13.7 36.9

195

1.5

4.8

7.4

196

1.0 15.8

8.6 11.2 17.2 17.0 28.9

197

1.0 14.4

8.9 13.8 12.3 10.9 26.3 28.9 43.3 19.0 54.8

198

1.0

2.0 11.0

7.6

9.9 38.1 45.9 16.9 14.6 43.1

199

1.0 18.7 18.3 19.2

5.8

8.1 20.3 13.8 16.2

200

1.0

5.2

5.6

201

3.0 19.6 19.2 18.3 19.2 19.1 31.8 44.4 57.7 58.2 59.5

202

1.0

203

10.0

3.6 14.4 10.5

3.1

6.1

6.5

8.5 12.6 27.9

6.9

204

10.0

4.4 12.9

9.4

4.9

9.2

6.1

2.8

8.0

5.1

3.1

205

1.0

0.0

0.0

0.0

0.0

0.0

0.2

0.0

0.0

0.0

0.0

206

1.0

6.7 14.9

9.4

8.8

8.5 12.7 19.6 20.2 37.8 31.4

207

15.0

4.9

4.9

5.2

5.1

4.8

7.7

8.7

5.3

4.8

5.1

208

1.0

0.0

0.0

0.0

0.0

0.0

0.0

0.2

0.0

0.0

0.0

209

80.0

4.2 16.5

7.4

5.8

2.6 35.1 29.4 37.0 33.5 26.4

210

0.1

1.0

1.0

1.0

1.0

211

1.0 14.8 17.2 11.8 19.3 16.6 49.7 58.2 56.4 57.7 57.4

212

1.0

8.2

6.9 16.0 15.9 17.9 34.7 18.9 30.3 58.2 55.7

213

1.0

8.1

6.3

9.7

7.5

1.2

9.7

6.9 15.9

8.6 12.4

1.0

7.2

6.7

9.2 11.7

5.3 46.1 28.0

9.1 38.0 54.9 25.3 28.1 44.6

9.2 41.5 30.7

8.2 27.5 49.2 17.6 17.1 28.7

4.1 19.2 29.4

2.2

2.1

2.9

4.5 32.4 37.7 48.2 12.3

9.2 23.2 34.4 34.4 30.4 26.3

4.8 13.5 13.8

1.0

0.5

40

1.0

0.2

1.0

1.0

5.1 30.8

1.0

4.9

214

1.0

6.7

5.5 17.2 19.1 11.3 14.8

215

1.5 18.6 15.8 13.2 17.3 10.6 48.6 50.7

3.3 29.6 52.0

216

4.0

0.0

0.1

0.0

0.1

0.2

0.0

0.0

0.0

0.0

0.0

217

4.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.1

0.0

0.0

218

0.1

0.4

0.3

0.1

0.0

0.0

0.1

0.0

0.3

0.0

0.1

219

1.0

0.2

0.3

0.4

0.3

0.0

0.1

0.4

0.0

0.2

0.0

220

1.0

8.7

6.3 10.5 12.6 15.6 13.3 16.2 49.9 20.0 58.6

221

5.5

2.5

0.6

0.6

1.8

222

1.0 13.3

8.9

8.4

4.3 11.2 14.5 22.9 42.5 19.8 21.7

223

1.0 10.6 11.1 12.3

2.9 12.2 36.1 33.1 48.3 22.3 48.5

224

1.0 11.4

6.3

2.0 35.9 41.6 30.2 35.6 37.9

225

1.0 11.8 16.9

5.4

7.5

5.8 36.4

226

1.0 10.0

3.5

3.0 17.4 18.7

227

1.0 12.6

8.7 17.4

228

1.0 12.6

0.2

229

1.0 12.7 16.1 18.6

9.6 11.1 19.2 59.7 11.1 55.0

230

1.0

8.5 10.3 11.0

8.8

7.1 29.8 45.4 50.9 39.5 27.7

231

1.0

5.0

8.9 13.6

9.9

5.7 47.4 26.3 17.0 22.5

232

1.0

7.5

2.4

3.6 19.6 19.7 39.2

0.0 59.8

1.7 59.9

233

1.0

1.0

1.0

1.0

1.0

1.0

234

1.0 18.0

2.9

0.0 20.0 20.0 35.6

0.0 60.0

0.0 60.0

235

1.0

7.8

4.6

2.4 19.2 19.3 39.6

0.3 59.6

0.2 59.6

236

1.0

8.4

0.0

0.0

0.3

0.2

0.0

0.2

237

5.0

1.0

5.6 18.6

8.7 17.1 57.2 20.9 51.1

0.0

9.7

9.4 10.6

8.2

0.6

0.9

0.3

0.3

0.4

0.6

0.5

0.9

8.9 39.5 36.9 18.8

0.2 59.4

0.9 59.4

9.9 49.5 16.3 25.8 24.5 29.1

4.3 19.2 18.7 36.3

1.0

7.6 34.3 43.1 41.2

1.0

0.2

1.0

3.3

41

0.3 59.3

1.0

2.5 59.6
9.5

2.8

1.0