Water Demand Model for the City Water Demand Model for the City of Makkah, Saudi Arabiaof Makkah, Saudi Arabia

AbdelHamid Ajbar, Emad Ali

Chemical Engineering Department, King Saud University, Riyadh, Saudi Arabia

Motivations for Water Motivations for Water Demand ModelDemand Model

Saudi ArabiaArid country: little rain and no surface water

Depends heavily on costly desalination plants

Growing populations and economic activity

Weak infrastructure management (Leak and Large Unaccounted-for-water)

Potable water resourcesPotable water resources

Desalination => 80%

◦Thermal desalination plant − 70%◦RO desalination – 30%

Underground water => 20%

City CharacteristicsCity Characteristics

• Makkah city, with a population of around 1.5 millions.

• Makkah is a focal point for local and international religious tourism

• The annual population’s growth estimated at 3% puts considerable strains on available water resources.

• The residential per capita water consumption in the city is estimated in 2010 to be 250 l/day

Econometric Water Demand Econometric Water Demand modelmodelThe standard functional population model for

estimating total water use:

Q = NqQ is the total annual water use, N the population

number and q is the water use per capita.

The water use (q) is assumed to depend on a number of explanatory variables (Xi).

2121 XXq

Explanatory VariablesExplanatory Variables

The selection of the explanatory variables is conditioned by the availability of historical data and also by the anticipated importance of the variable:

– The household median income (I )– The household size (i.e. persons per house) (Hs) – The maximum monthly temperature (T). – The monthly visitor flux (V)

Model ShortcomingsModel Shortcomings

Household and Income data are yearly based

Temperature varies monthlyVisitors vary monthly

4321 VTHIq

Further ChallengeFurther Challenge

Visitors distribution is well defined on the basis of lunar months

Average monthly temperature is consistent with the solar months

Suggested RemedySuggested Remedy

Use NN model – allow for multi-rate variables

Q = f(N,I,H,V,T)Model prediction should be based

on lunar monthsEstimate the lunar monthly

temperature using correlation between lunar and solar systems

Touristic CharacteristicsTouristic Characteristics

Combined Tourist FluxCombined Tourist Flux

1 2 3 4 5 6 7 8 9 10 11 120

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Lunar Month

Tot

al v

isit

ors

(x10

00,0

00)

Monthly Temperature Monthly Temperature DistributionDistribution

1 2 3 4 5 6 7 8 9 10 11 1226

28

30

32

34

36

38

40

42

44

Month

Tem

pera

ture

(o C

)

Average Gregorian

2003

2004

2005

2006

Neural Network Neural Network StructureStructure

input layer hidden layer output layer

I

N

H

)(f

)(f

)(f

)(f

)(f

)(f

)(f

T1

T12

V1

V12

D1

D2

D11

D12

w1

w2

w11

w12

NN model TrainingNN model Training

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 510.24

0.26

0.28

0.3

0.32

0.34

0.36

0.38

Lunar Month

Wat

er P

rodu

ctio

n (1

06 m3 /D

)Training dataModel prediction

NN Model validationNN Model validation

0 3 6 9 12 15 18 21 24 270.26

0.28

0.3

0.32

0.34

0.36

0.38

0.4

0.42

0.44

Lunar Month

Wat

er P

rodu

ctio

n (1

06 m3 /D

)Validation data

Model prediction

Thank YouThank You