2.

1
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

CHAPTER 2

Water demand

2.1 TERMINOLOGY

Water conveyance in a water supply system depends on the rates of production,

delivery, consumption and leakage (Figure 2.1).

Figure 2.1. Flows in water supply systems.

Water production Water production (Qwp) takes place at water treatment facilities. It normally
has a constant rate that depends on the purification capacity of the treatment
installation. The treated water ends up in a clear water reservoir from where it is
supplied to the system (Reservoir A in Figure 2.1).
Water delivery Water delivery (Qwd) starts from the clear water reservoir of the treatment
plant. Supplied directly to the distribution network, the generated flow will match
certain demand patterns. When the distribution area is located far away from the
treatment plant, the water is likely to be transported to another reservoir (B in Figure
2.1) that is usually constructed at the beginning of the distribution network. In
principle, this delivery is done at the same constant flow rate that is equal to the water
production.
Water Water consumption (Qwc) is the quantity directly utilised by the consumers.
consumption This generates variable flows in the distribution network caused by many factors:
users’ needs, climate, source capacity, etc.
Water leakage Water leakage (Qwl) is the amount of water physically lost from the system.
The generated flow rate is in this case more or less constant and depends on overall
conditions in the system.
2.2 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Water demand In theory, the term water demand (Qd) coincides with water consumption. In practice
however, the demand is often monitored at supply points where the measurements
include leakage, as well as the quantities used to refill the balancing tanks that may
exist in the system. In order to avoid false conclusions, a clear distinction between the
measurements at various points of the system should always be made. It is commonly
agreed that Qd = Qwc + Qwl. Furthermore, when supply is calculated without having an
interim water storage, i.e. water goes directly to the distribution network: Qwd = Qd,
otherwise: Qwd = Qwp.

Water demand is commonly expressed in cubic meters per hour (m3/h) or per second
(m3/s), litres per second (l/s), mega litres per day (Ml/d) or litres per capita per day
(l/c/d or lpcpd). Typical Imperial units are cubic feet per second (ft3/s), gallon per
minute (gpm) or mega gallon per day (mgd)1). The mean value derived from annual
Average demand demand records represents the average demand. Divided by the number of consumers,
Specific demand the average demand becomes the specific demand (unit consumption per capita).
Apart from neglecting leakage, the demand figures can often be misinterpreted
due to lack of information regarding the consumption of various categories. Table 2.1
shows the difference in the level of specific demand depending on what is, or is not,
included in the figure. The last two groups in the table coincide with commercial and
domestic water use, respectively.

Table 2.1. Water demand in The Netherlands in 2001 (Source: VEWIN).

Annual (106 m3) Qd (l/c/d)*

Total water delivered by water companies 1247 214

Drinking water delivered by water companies 1177 202

Drinking water paid for by consumers 1119 192

Consumers below 10,000 m3/y per connection (metered) 940 161

Consumers below 300 m3/y per connection (metered) 714 122

* Based on total population of approx. 16 million

Accurate forecasting of water demand is crucial whilst analysing the hydraulic

performance of water distribution systems. Numerous factors affecting the demand are
determined from the answers to three basic questions:
1. For which purpose is the water used? The demand is affected by a number of
consumption categories: domestic, industrial, tourism, etc.
2. Who is the user? Water use within the same category may vary due to different
cultures, education, age, climate, religion, technological process, etc.
3. How valuable is the water? The water may be used under circumstances that
restrict the demand: scarce source (quantity/quality), bad access (no direct
connection, fetching from a distance), low income of consumers, etc.

Answers to the above questions reflect on the quantities and moments when the water
will be used, resulting in a variety of demand patterns. Analysing or predicting these
patterns is not always an easy task. Uncritical adoption of other experiences where the

1) A unit conversion table is given in Appendix 7. See also spreadsheet lesson 8-1: ‘Flow Conversion’.
 Water demand 2.3
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

field information is lacking is the wrong approach; each case is independent and the
conclusions drawn are only valid for local conditions.

Variations in water demand are particularly visible in developing countries where

prosperity is predominantly concentrated in a few major, usually overcrowded, cities
with peripheral areas often having restricted access to drinking water. These parts of
the system will be supplied from public standpipes, individual wells or tankers, which
cause substantial differences in consumption levels within the same distribution area.
Figure 2.2 shows average specific consumption for a number of large cities in Asia.

Figure 2.2. Specific consumption in Asian cities (Source: McIntosh, 2003).

Comparative figures in Africa are generally lower, resulting from the range of
problems that cause intermediate supply, namely long distances, electricity failures,
pipe bursts, polluted ground water in deep wells, etc.
A water demand survey was conducted for the region around Lake Victoria,
covering parts of Uganda, Tanzania and Kenya. The demand where there is a piped
supply (the water is tapped at home) was compared with the demand in un-piped
systems (no house connection is available). The results are shown in Table 2.2.

Table 2.2. Specific demand around Lake Victoria in Africa (Source: IIED, 2000).
Piped (l/c/d) Un-piped (l/c/d)

Average for the entire region 45 22

Average for urban areas (small towns) 65 26

Average for rural areas 59 8

Part of the region in Uganda 44 19

Part of the region in Tanzania 60 24

Part of the region in Kenya 57 21

2.4 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Unaccounted-for An unavoidable component of water demand is unaccounted-for water (UFW), the

water water that is supplied ‘free of charge’. In quite a lot of transport and distribution
systems in developing countries this is the most significant ‘consumer’ of water,
accounting sometimes for over 50% of the total water delivery.
Causes of UFW differ from case to case. Most often it is a leakage that appears
due to improper maintenance of the network. Other non-physical losses are related to
the water that is supplied and has reached the taps, but is not registered or paid for
(under-reading of water meters, illegal connections, washing streets, flushing pipes,
etc.)

2.2 CONSUMPTION CATEGORIES

2.2.1 Water use by various sectors

Water consumption is initially split into domestic and non-domestic components. The
bulk of non-domestic consumption relates to the water used for agriculture,
occasionally delivered from integral water supply systems, and for industry and other
commercial uses (shops, offices, schools, hospitals, etc.). The ratio between the
domestic and non-domestic consumption in The Netherlands in the period 1960-2000
is shown in Figure 2.32).

Figure 2.3. Domestic and non-domestic consumption in The Netherlands (Source: VEWIN).

In the majority of developing countries, agricultural- and domestic water consumption

is predominant compared to the commercial water use, as the example in Table 2.3
shows. However, this water is rarely supplied from an integral system.
In warm climates, the water used for irrigation is generally the major
component of total consumption; Figure 2.4 shows an example of some European
countries around the Mediterranean Sea: Spain, Italy and Greece. On the other hand,
highly industrialised countries use huge quantities of water, often of drinking quality,
for cooling; typical examples are Germany, France and Finland, which all use more

2) The domestic consumption in Figure 2.3 is derived from consumers metered below 300 m3/y per connection. The real
consumption is assumed to be slightly higher; the figure assessed by VEWIN for 2001 is 126 l/c/d compared to 134 l/c/d in
1995.
 Water demand 2.5
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

than 50% of the total consumption for this purpose. Striving for more efficient
irrigation methods, industrial processes using alternative sources and recycling water
have been and still are a concern in developed countries for the last few decades.

Table 2.3. Domestic vs. non-domestic consumption in some African states (Source: SADC, 1999).
Country Agriculture (%) Industry (%) Domestic (%)
Angola 76 10 14
Botswana 48 20 32
Lesotho 56 22 22
Malawi 86 3 10
Mozambique 89 2 9
South Africa 62 21 17
Zambia 77 7 16
Zimbabwe 79 7 14

Figure 2.4. Water use in Europe (Source: EEA, 1999).

2.2.2 Domestic consumption

Domestic water consumption is intended for toilet flushing, bathing & showering,
laundry, dishwashing and other less water intensive or less frequent purposes:
cooking, drinking, gardening, car washing, etc. The example in Figure 2.5 shows
rather wide variation in the average domestic consumption of some industrialised
countries. Nevertheless, in all the cases indicated 50-80% of the total consumption
appears to be utilised in bathrooms and toilets.
2.6 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.5. Domestic water use in Europe (Sources: EEA, BGW, VEWIN)

The habits of different population groups with respect to water use were studied in
The Netherlands (Achttienribbe, 1993). Four factors compared were age, income
level, household size and region of the country. The results are shown in Figure 2.6.
The figures prove that even with detailed statistics available, conclusions about
global trends may be difficult. In general, the consumption is lower in the northern
part of the country, which is a less populated, mostly agricultural region. Nonetheless,
interesting findings from the graphs are evident: the middle-aged group is the most
moderate water user, more frequent toilet use and less frequent shower use is
exercised by older groups, larger families are with a lower consumption per capita,
etc.

In cases where there is an individual connection to the system, the structure of

domestic consumption in water scarce areas may well look similar but the quantity of
water used for particular activities will be minimized. Apart from the change of habits,
this is also a consequence of low pressures in the system directly affecting the
quantities used for showering, gardening, car washing, etc. On top of this, the water
company may be forced to ration the supply by introducing regular interruptions. In
these situations consumers will normally react by constructing individual tanks. In
urban areas where supply with individual tanks takes place, the amounts of water
available commonly vary between 50-100 l/c/d.

2.2.3 Non-domestic consumption

Non-domestic or commercial water use occurs in industry, agriculture, institutions and

offices, tourism, etc. Each of these categories has its specific water requirements.
 Water demand 2.7
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.6. Structure of domestic consumption in The Netherlands (Source: Achttienribbe, 1993).
2.8 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Industry

Water in industry can be used for various purposes: as a part of the final product, for
the maintenance of manufacturing processes (cleaning, flushing, sterilisation,
conveying, cooling, etc) and for the personal needs (usually comparatively marginal).
The total quantities will largely depend on the type of industry and technological
process. They are commonly expressed in litres per unit of product or raw material.
Table 2.4 gives an indication for a number of industries; an extensive overview can be
found in HR Wallingford (2003).

Table 2.4. Industrial water consumption (Adapted from: HR Wallingford, 2003).

Industry Litres per unit product

Carbonated soft drinks* 1.5 – 5 per litre

Fruit juices* 3 – 15 per litre

Beer* 4 – 22 per litre

Wine 1 – 4 per litre

Fresh meat (red) 1.5 – 9 per kg

Canned vegetables/fruits 2 – 27 per kg

Bricks 15 – 30 per kg

Cement 4 per kg

Polyethylene 2.5 – 10 per kg

Paper 4** – 35 per kg

Textiles 100 – 300 per kg

Cars 2500 – 8000 per car

* Largely dependant on the packaging and cleaning of bottles
** Recycled paper

Agriculture

Water consumption in agriculture is mainly determined by irrigation and livestock

needs. In peri-urban or developed rural areas, this demand may also be supplied from
the local distribution system.
The amounts required for irrigation purposes depend on the plant sort, stage of
growth, type of irrigation, soil characteristics, climatic conditions, etc. These
quantities can be assessed either from records or by simple measurements. A number
of methods are available in literature to calculate the consumption based on
meteorological data (Blaney-Criddle, Penman, etc.). According to Brouwer and
Heibloem (1986), the consumption is unlikely to exceed a monthly mean of 15 mm
per day, which is equivalent to 150 m3/d per hectare. Approximate values per crop are
given in Table 2.5.
 Water demand 2.9
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Table 2.5. Seasonal crop water needs (Source: Brouwer and Heibloem, 1986).
Crop Season Consumption
(days/year) (mm/season)

Bananas 300-365 1200-2200

Beans 75-110 300-500

Cabbages 120-140 350-500

Citrus fruit 240-365 900-1200

Corn 80-180 500-800

Potatoes 105-145 500-700

Rice 90-150 450-750

Sunflowers 125-130 600-1000

Tomatoes 135-180 400-800

Wheat 120-150 450-650

Water required for livestock depends on the sort and age of the animal, as well as
climatic conditions. Size of the stock and type of production also play a role. For
example, the water consumption for milking cows is 120-150 l/d per animal, whilst
cows typically need only 25 l/d (Brandon, 1984).

Table 2.6. Animal water consumption (Brandon, 1984).

Animal Litres per day

Cows 25-150

Oxen, horses, etc. 15-40

Pigs 10-30

Sheep, goats 5-6

Turkeys (per 100) 65-70

Chickens (per 100) 25-30

Camels 2-3

Institutions

Commercial consumption in restaurants, shops, schools and other institutions can be

assessed as a total supply divided by the number of consumers (employees, pupils,
patients, etc.). Accurate figures should be available from local records at water supply
companies. Some indications of unit consumption are given in Table 2.7. These
assume individual connection with indoor water installations and waterborne
sanitation, and are only relevant during working days.
2.10 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Table 2.7. Water consumption in institutions (Adapted from: HR Wallingford, 2003).

Premises Consumption

Schools 25 – 75 l/d per pupil

Hospitals 350 – 500 l/d per bed

Laundries 8* – 60 litre per kg washing

Small businesses 25 l/d per employee

Retail shops/stores 100 – 135 l/d per employee

Offices 65 l/d per employee

* Recycled water used for rinsing

Tourism

Tourist and recreational activities may also have a considerable impact on water
demand. The quantities per person (or per bed) per day vary enormously depending on
the type and category of accommodation; in luxury hotels, for instance, this demand
can go up to 600 l/c/d. Table 2.8 shows average figures in Southwest England.

Table 2.8.Tourist water consumption in Southwest England (Brandon, 1984).

Accommodation Consumption
(l/c/d)

Camping sites 68

Unclassified hotels 113

Guest houses 130

1- & 2-star hotels 168

3-,4- & 5-star hotels 269

Miscellaneous groups

Water consumption that does not belong to any of the above-listed groups can be
classified as miscellaneous. These are the quantities used for fire fighting, public
purposes (washing streets, maintaining green areas, supply for fountains, etc.),
maintenance of water and sewage systems (cleansing, flushing mains) or other specific
uses (military facilities, sport complexes, zoos, etc.). Sufficient information on water
consumption in such cases should be available from local records. Sometimes this
demand is unpredictable and can only be estimated on an empirical or statistical basis.
For example, in the case of fire fighting, the water use is not recorded and
measurements are difficult because it is not known in advance when and where the
water will be needed. Provision for this purpose will be planned with respect to
potential risks, which is a matter discussion between the municipality (fire
department) and water company.
 Water demand 2.11
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

On average, these consumers do not contribute substantially in overall demand. Very

often they are neither metered nor accounted for and thus classified as UFW.

Problem 2-1

A water supply company has delivered an annual quantity of 80,000,000 m3 to a city of 1.2 million
inhabitants. Find out the specific demand in the distribution area. In addition, calculate the domestic
consumption per capita with leakage from the system estimated at 15 % of the total supply, and billed
non-domestic consumption of 20,000,000 m3/y.

Answer:

Gross specific demand can be determined as:

80,000,000 × 1000
Qavg = ≈ 183 l / c / d
1,200,000
365

The leakage of 15% of the total supply amounts to an annual loss of 12 million m3. Reducing the total
figure further for the registered non-domestic consumption yields the annual domestic consumption of
80 – 12 – 20 = 48 million m3, which is equal to a specific domestic consumption of approx.110 l/c/d.

Self-study: Workshop problems 1.1 & 1.2 (Appendix 1)

Spreadsheet lesson 8.1 (Appendix 5)

2.3 WATER DEMAND PATTERNS

Each consumption category can be considered not only from the perspective of its
average quantities but also with respect to the timetable of when the water is used.
Demand variations are commonly described by the peak factors. These are the
ratios between the demand at particular moments and the average demand for the
observed period (hour, day, week, year, etc.). For example, if the demand registered
during a particular hour was 150 m3 and for the whole day (24 hours) the total demand
was 3000 m3, the average hourly demand of 3000/24 = 125 m3 would be used to
determine the peak factor for the hour, which would be 150/125 = 1.2. Other ways of
peak demand representation are either as a percentage of the total demand within a
particular period (150 m3 for the above hour is equal to 5 % of the total daily demand
of 3000 m3), or simply as the unit volume per hour (150 m3/h).

Human activities have periodic characteristics and the same applies to water use.
Hence, the average water quantities from the previous paragraph are just indications of
total requirements. Equally relevant for the design of water supply systems are
consumption peaks that appear during one day, week or year. A combination of these
maximum and minimum demands defines the absolute range of flows that are to be
delivered by the water company.
Time-wise, we can distinguish the instantaneous, daily (diurnal), weekly and
annual (seasonal) pattern in various areas (home, building, district, town, etc.). The
larger the area is, the more diverse the demand pattern will be as it then represents a
combination of several consumption categories, including leakage.
2.12 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

2.3.1 Instantaneous demand

Simultaneous Instantaneous demand (in some literature simultaneous demand) is caused by a small
demand number of consumers during a short period of time: a few seconds or minutes.
Assessing this sort of demand is the starting point in building-up the demand pattern
of any distribution area. On top of that, the instantaneous demand is directly relevant
for network design in small residential areas (tertiary networks and house
installations). The demand patterns of such areas are much more unpredictable than
the demand patterns generated by larger number of consumers. The smaller the
number of consumers involved, the less predictable the demand pattern will be.

The following hypothetical example illustrates the relation between the peak demands
and the number of consumers.
A list of typical domestic water activities with provisional unit quantities
utilised during a particular period of time is shown in Table 2.9. Parameter Qins in the
table represents the average flow obtained by dividing the total quantity with the
duration of the activity, converted into litres per hour.

Table 2.9. Example of domestic unit water consumption.

Activity Total quantity Duration Qins
(litres) (minutes) (l/h)

A - Toilet flushing 8 1 480

B - Showering 50 6 500

C – Hand washing 2 1/2 240

D – Face & teeth 3 1 180

E - Laundry 60 6 600

F - Cooking 15 5 180

G - Dishes 40 6 400

H - Drinking 1/4 1/20 300

I – Other 5 2 150

For example, ‘Activity A-Toilet flushing’ is in fact refilling of the toilet cistern. In this
case there is a volume of 8 l, within say one minute after the toilet has been flushed. In
theory, to be able to fulfil this requirement, the pipe that supplies the cistern should
allow the flow of 8 × 60 = 480 l/h within one minute. This flow is thus needed within
Instantaneous a relatively short period of time and is therefore called the instantaneous flow.
flow
Although the exact moment of water use is normally unpredictable, it is well known
that there are some periods of the day when it happens more frequently. For most
people this is in the morning after they wakeup, in the afternoon when they return
from work or school, or in the evening before they go to sleep.
Considering a single housing unit, it is not reasonable to assume a situation in
 Water demand 2.13
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

which all water-related activities from the above table are executed simultaneously.
For example, in the morning, a combination of activities A, B, D & H might be
possible. If this is the assumed maximum demand during the day, the maximum
instantaneous flow equals the sum of the flows for these four activities. Hence, the
pipe that provides water for the house has to be sufficiently large to convey the flow
of:

480 + 500 + 180 + 300 = 1460 l/h

With an assumed specific consumption of 120 l/c/d and, say, four people living
Instantaneous together, the instantaneous peak factor will be:
peak factor
1460
pf ins = = 73
120 × 4
24

Thus, there was at least one short moment within 24 hours when the instantaneous
flow to the house was 73 times higher than the average flow of the day.
Applying the same logic for an apartment building, one can assume that all
tenants use the water there in a similar way and at a similar moment, but never in
exactly the same way and at exactly the same moment. Again, the maximum demand
of the building occurs in the morning. This could consist of, for example, toilet
flushing in say three apartments, hand washing in two, teeth brushing in six, doing the
laundry in two and drinking water in one. The maximum instantaneous flow out of
such a consumption scenario case would be:

3A + 3B + 2C + 6D + 2E + 1H = 6000 l/h

which is the capacity that has to be provided by the pipe that supplies the building.
Assuming the same specific demand of 120 l/c/d and for possibly 40 occupants, the
instantaneous peak factor is:

6000
pf ins = = 30
120 × 40
24

Any further increase in the number of consumers will cause the further lowering of the
instantaneous peak factor, up to a level where this factor becomes independent from
the growth in the number of consumers. As a consequence, some large diameter pipes
that have to convey water for possibly 100,000 consumers would probably be
designed based on a rather low instantaneous peak factor, which in this example could
be 1.4.
Simultaneity A simultaneity diagram can be obtained by plotting the instantaneous peak
diagram factors against the corresponding number of consumers. The three points from the
above example, interpolated exponentially, will yield the graph shown in Figure 2.7.
2.14 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.7. Simultaneity diagram (example).

In practice, the simultaneity diagrams are determined from a field study for each
particular area (town, region or country). Sometimes, a good approximation is
achieved by applying mathematical formulae; the equation: pfins ≈126×e(-0.9×logN)
where N represents the number of consumers, describes the curve in Figure 2.7.
Furthermore, the simultaneous curves can be diversified based on various standards of
living i.e. type of accommodation, as Figure 2.8 shows.

Figure 2.8. Simultaneity diagram of various categories of accommodation

In most cases, the demand patterns of more than a few thousand people are fairly
predictable. This eventually leads to the conclusion that the water demand of larger
group of consumers will, in principle, be evenly spread over a period of time that is
longer than a few seconds or minutes. This is illustrated in the 24-hour demand
diagram shown in Figure 2.9 for the northern part of Amsterdam. In this example
there were nearly 130,000 consumers, and the measurements were executed at 1-
minute intervals.
 Water demand 2.15
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.9. Demand pattern in Amsterdam (Source: Municipal Water Company Amsterdam, 2002).

One-hour durations are commonly accepted for practical purposes and the
instantaneous peak factors within this period of time will be represented by a single
Hourly peak value called the hourly (or diurnal) peak factor, as shown in Figure 2.10.
factor

Figure 2.10. Instantaneous demand from Figure 2.9 averaged by the hourly peak factors.

There are however extraordinary situations when the instantaneous demand may
substantially influence the demand pattern, even in the case of large numbers of
consumers.
Figures 2.11 & 2.12 show the demand pattern (in m3/min) during the TV
broadcasting of two football matches when the Dutch national team played against
Saudi Arabia and Belgium at the 1994 World Cup in the United States of America.
The demand was observed in a distribution area of approximately 135,000 people.
2.16 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.11. Night-time demand during football game (Water Company ‘N-W Brabant’, NL, 1994).

Figure 2.12. Evening demand during football game (Water Company ‘N-W Brabant’, NL, 1994).

The excitement of the viewers is clearly confirmed through the increased water use
during the break and at the end of the game, despite the fact that the first match was
played in the middle of the night (with different time zones between The Netherlands
and USA). Both graphs point almost precisely to the start of the TV broadcast that
happened at 01:50 and 18:50, respectively. The water demand dropped soon after the
start of the game until the half time when the first peak occurs; it is not difficult to
 Water demand 2.17
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

guess for what purpose the water was used! The upper curves in both figures show the
demand under normal conditions, one week before the game at the same period of the
day.
This phenomenon is not only typical in The Netherlands; it will be met
virtually everywhere where football is sufficiently popular. Its consequence is a
temporary drop of pressure in the system while in the most extreme situations a pump
failure might occur. Nevertheless, these demand peaks are rarely considered as design
parameters and adjusting operational settings of the pumps can easily solve this
problem.

Problem 2-2

In a residential area of 10,000 inhabitants, the specific water demand is estimated at 100 l/c/d (leakage
included). During a football game shown on the local TV station, the water meter in the area registered
the maximum flow of 24 l/s, which was 60 % above the regular use for that period of the day. What was
the instantaneous peak factor in that case? What would be the regular peak factor on a day without a
televised football broadcast?

Answers:

In order to calculate the peak factors, the average demand in the area has to be brought to the same units
as the peak flows. Thus, the average flow becomes:

10,000 × 100
Qavg = ≈ 12 l / s
24
3600

The regular peak flow at a particular point of the day is 60 % lower than the one registered during the
football game, which is 24 / 1.6 = 15 l/s. Finally, the corresponding peak factors will be 24 / 12 = 2
during the football game, and 15 / 12 = 1.25 in normal supply situations.

Self-study: Workshop problems 1.3 to 1.5 (Appendix 1)

2.3.2 Diurnal patterns

For sufficiently large group of consumers, the instantaneous demand pattern for 24-
Diurnal demand hour period converts into a diurnal (daily) demand diagram. Diurnal diagrams are
diagram important for the design of primary and secondary networks, and in particular their
reservoirs and pumping stations. Being the shortest cycle of water use, a one-day
period implies a synchronised operation of the system components with similar supply
conditions occurring every 24 hours.
The demand patterns are usually registered by monitoring flows at delivery
points (treatment plants) or points in the network (pressure boosting stations,
reservoirs, control points with either permanent or temporary measuring equipment).
With properly organised measurements the patterns can also be observed at the
consumers’ premises. Firstly, such an approach allows the separation of various
consumption categories and secondly, the leakage in the distribution system will be
excluded, resulting in a genuine consumption pattern.
A few examples of diagrams for different daily demand categories are given in
Figures 2.13 to 2.16.
2.18 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.13. Urban demand pattern (Source: Gabrić, 1996 & Trifunović, 1993).

Figure 2.14. Industrial demand pattern - example from Bosnia and Herzegovina (Obradović, 1991).

Figure 2.15. Tourist demand pattern - example from Croatia (Obradović, 1991).
 Water demand 2.19
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.16. Commercial/institutional demand pattern - example from USA (Obradović, 1991).

A flat daily demand pattern reflects the combination of impacts from the following
factors:
- large distribution area,
- high industrial demand,
- high leakage level,
- scarce supply (individual storage).

Commonly, the structure of the demand pattern in urban areas looks as shown in
Figure 2.17: the domestic category will have the most visible variation of consumption
throughout the day, industry and institutions will usually work in daily shifts, and the
remaining categories, including leakage, are practically constant.
By separating the categories, the graph will look like Figure 2.18, with peak
factors calculated for the domestic consumption only, then for the total consumption
(excluding leakage), and finally for the total demand (consumption plus leakage). In
this case, it appears that contributions from the industrial consumption and leakage
flatten the patterns.

Figure 2.17. Typical structure of diurnal demand in urban areas.

2.20 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.18. Peak factor diagrams of various categories from Figure 2.17.

2.3.3 Periodic variations

The peak factors from diurnal diagrams are derived on the basis of average
consumption during 24 hours. This average is subject to two additional cycles: weekly
and annual.
Weekly demand Weekly demand pattern is influenced by average consumption on working and
pattern non-working days. Public holidays, sport events, etc. play a role in this case as well.
One example of the demand variations during a week is shown in Figure 2.19. The
difference between the two curves in this diagram reflects the successful
implementation of the leak detection programme.

Figure 2.19. Weekly demand variations - Alvington, UK (Dovey and Rogers, 1993).
 Water demand 2.21
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Consumption in urban areas of Western Europe is normally lower over weekends. On.
Saturdays and Sundays people rest, which may differ in other parts of the world. For
instance, Friday is a non-working day in Islamic countries and domestic consumption
usually increases then.

Annual variations in water use are predominantly linked to the change of seasons
Seasonal and are therefore also called seasonal variations.
variations

Figure 2.20. Seasonal demand variation in a sea resort (Obradović, 1991).

The unit consumption per capita normally grows during hot seasons but the increase in
total demand may also result from a temporarily increased number of consumers,
which is typical for holiday resorts. Figure 2.20 shows the annual pattern in Istria,
Croatia on the Adriatic coast; the peaks of the tourist season, during July and August,
are also the peaks in water use.

Figure 2.21. Weekly and monthly peak factor diagrams.

Just as with diurnal patterns, typical weekly and annual patterns can also be expressed
through peak factor diagrams. Figure 2.21 shows an example in which the peak daily
2.22 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

demand appears typically on Mondays and is 14% above the average, while the
minimum on Sundays is 14% below the average daily demand for the week. The
second curve shows the difference in demand between summer and winter months,
fluctuating within a margin of 10%.

Generalising such trends leads to the conclusion that the absolute peak consumption
during one year occurs on a day of the week, and in the month when the consumption
Maximum is statistically the highest. This day is commonly called the maximum consumption
consumption day day. In the above example, the maximum consumption day would be a Monday
somewhere in June, with its consumption being 1.14 × 1.05 = 1.197 times higher than
the average daily consumption for the year. In practice however, the maximum
consumption day in one distribution area will be determined from the daily demand
records of the water company. This is simply the day when the total registered demand
was the highest in a particular year.

Finally, the daily, weekly and annual cycles are never repeated in exactly the same
way. However, for design purposes a sufficient accuracy is achieved if it is assumed
that all water needs are satisfied in a similar schedule during one day, week or year.
Regarding the seasonal variations, the example in Figure 2.22 confirms this; the
annual patterns in the graph are more or less the same while the average demand
grows each year as a result of population growth.

Figure 2.22. Annual demand patterns in Ramallah (Thaher, 1998).

Problem 2-3

A water supply company delivered an annual quantity of 10,000,000 m3, assuming an average leakage
of 20 %. On the maximum consumption day, the registered delivery was as follows:

Hour 1 2 3 4 5 6 7 8 9 10 11 12

m3 989 945 902 727 844 1164 1571 1600 1775 1964 2066 2110

Hour 13 14 15 16 17 18 19 20 21 22 23 24

m3 1600 1309 1091 945 1062 1455 1745 2139 2110 2037 1746 1018
 Water demand 2.23
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Determine:
a) diurnal peak factors for the area,
b) the maximum seasonal variation factor,
c) diurnal consumption factors,

Answers:

a) From the above table, the average consumption on the maximum consumption day was 1454.75
m3/h leading to the following hourly peak factors:

Hour 1 2 3 4 5 6 7 8 9 10 11 12

pfh 0.680 0.650 0.620 0.500 0.580 0.800 1.080 1.100 1.220 1.350 1.420 1.450

Hour 13 14 15 16 17 18 19 20 21 22 23 24

pfh 1.100 0.900 0.750 0.650 0.730 1.000 1.200 1.470 1.450 1.400 1.200 0.700

b) The average consumption, based on the annual figure, is 10,000,000/365/24 = 1141.55 m3/h. The
seasonal variation factor is therefore 1454.75/1141.55 = 1.274.
c) The average leakage of 20% assumes an hourly flow of approx. 228 m3/h, which is included in the
above hourly flows as water loss. The peak factors for consumption will therefore be recalculated
without this figure, as the following table shows:

Hour 1 2 3 4 5 6 7 8 9 10 11 12

m3 761 717 674 499 616 936 1343 1372 1547 1736 1838 1882

pfh 0.620 0.584 0.549 0.407 0.502 0.763 1.095 1.118 1.261 1.415 1.498 1.534

Hour 13 14 15 16 17 18 19 20 21 22 23 24

m3 1372 1081 863 717 834 1227 1517 1911 1882 1809 1518 790

pfh 1.118 0.881 0.703 0.584 0.680 1.000 1.237 1.558 1.534 1.475 1.237 0.644

The diagram of the hourly peak factors for the two situations will look as follows:

Self-study: Workshop problems 1.6 & 1.8 (Appendix 1)

Spreadsheet lessons 8.2 to 8.4 (Appendix 5)
2.24 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

2.4 DEMAND CALCULATION

Knowing the daily patterns and periodical variations, the demand flow can be
calculated from the following formula:

Qwc ,avg × pf o
Qd = (2.1)
 l 
1 −  × fc
 100 

The definition of the parameters is as follows:

Qd = Water demand of a certain area at a certain moment.
Qwc,avg = Average water consumption in the area.
pfo = Overall peak factor; this is a combination of the peak factor values
from the daily, weekly and annual diagrams: pfo = pfh × pfd × pfm. The
daily and monthly peak factors are normally integrated into one
(seasonal) peak factor: pfs = pfd × pfm.
l = Leakage expressed as a percentage of the water production.
fc = Unit conversion factor.

The main advantage of Equation 2.1 is its simplicity although some inaccuracy will be
necessarily introduced. Using this formula, the volume of leakage increases with
higher consumption i.e. the peak factor value, despite the fixed leakage percentage.
For example, if Qwc,avg = 1 (regardless of the flow units), pfo = 1 and the leakage
percentage is 50%, then as a result Qd = 2. Thus, half of the supply is consumed and
the other half is leaked.
If pfo = 2, Qd = 4. Again, this is ‘fifty-fifty’ but this time the volume of leakage
has grown from 1 to 2, which implies its dependence on the consumption level. This is
not true as the leakage level is usually constant throughout the day, with a slight
increase over night when the pressures in the network are generally higher (already
shown in Figure 2.17). Hence, the leakage level is pressure dependent rather than
consumption dependant.
Nonetheless, the above inaccuracy effectively adds safety to the design. Where
this is deemed unnecessary, an alternative approach is suggested, especially for
distribution areas with high leakage percentages:

Qd = (Qwc ,avg × pf o + Qwl )

1
(2.2)
fc

where:

l
Qwl = × Qwp (2.3)
100

In the case of pfo = 1, demand equals production and assuming the same units for all
parameters (fc = 1):

l
Qwp = Qwc ,avg + × Qwp (2.4)
100
 Water demand 2.25
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

This can be re-written as:

Qwc ,avg
Q wp = (2.5)
 l 
1 − 
 100 

By plugging Equation 2.5 into 2.3 and then to 2.2, the formula for water demand
calculation evolves into its final form:

Qwc ,avg  l 
Qd = ×  pf o +  (2.6)
fc  100 − l 

Where reliable information resulting from individual metering of consumers is not

available, the average water consumption, Qwc,avg, can be approximated in several
ways:

Qwc ,avg = ncq (2.7)

Qwc ,avg = dAcq (2.8)

Qwc ,avg = Acq a (2.9)

Qwc ,avg = nu qu (2.10)

where:
n = Number of inhabitants in the distribution area.
c = Coverage of the area. It can happen that some of the inhabitants are not
connected to the system, or some parts of the area are not inhabited.
This factor, which has a value of between 0 and 1, converts the number
of inhabitants into the number of consumers.
q = Specific consumption (l/c/d).
d = Population density (number of inhabitants per unit surface area).
A = Surface area of the distribution area.
qa = Consumption registered per unit surface area.
nu = Production capacity. It represents a number of units (kg, l, pieces, etc.)
produced within a certain period.
qu = Water consumption per unit product.
Unit consumptions, q, qa, & qu, are elaborated in Paragraph 2.1. The data for n, c, d, A
& nu are usually available from statistics or set by planning: local, urban, regional, etc.

As already mentioned, the demand in large urban areas is often composed of several
consumption categories. More accuracy in the calculation of demand for water is
therefore achieved if the distribution area is split into a number of sub-areas or
districts, with standardised categories of water users and a range of consumptions
based on local experience. The average consumption per district can then be
calculated from Equation 2.9, which has been modified:
2.26 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

n
Qwc ,avg = A∑ (q a ,i pi ci ) (2.11)
i =1

where:
A = Surface area of the district.
n = Number of consumption categories within the district.
qa,i = Unit consumption per surface area of category i.
pi = Percentage of the district territory occupied by category i.
ci = Coverage within the district territory occupied by category i.
With a known population density in each district, the result can be converted into
specific demand (per capita).

Regarding the pfo values, the following are typical combinations:

1) pfh = 1, pfs = 1; Qd represents the average consumption per day. This demand is
the absolute average, usually obtained from annual demand records and converted
into required flow units.
2) pfh = 1, pfs = max; Qd represents the average demand during the maximum
consumption day.
3) pfh = max, pfs = max; Qd is the demand at the maximum consumption hour on the
maximum consumption day.
4) pfh = min, pfs = min; Qd is the demand at the minimum consumption hour on the
minimum consumption day.
The entire range of demands that appear in one distribution system during one year is
specified by the demands under 3) and 4), which are shown in Figure 2.23. These peak
demands are relevant as parameters for the design of all system components: pipes,
pumps and storage.

Figure 2.23. Hypothetical annual range of flows in a distribution system.

 Water demand 2.27
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Problem 2-4

A water supply company in a town with a total population of approx. 275,000 conducted a water
demand survey resulting in the following categories of water users:

Category of water users qa

(m3/d/ha)

A- Residential area, apartments 90

B- Residential area, individual houses 55

C- Shopping areas 125

D- Offices 80

E- Schools, Colleges 100

F- Hospitals 160

G- Hotels 150

H- Public green areas 15

The city is divided into 8 districts, each with a known population, contribution to demand from each of
the categories, and estimated coverage by the distribution system, as shown in the table below.

Districts A B C D E F G H
86,251 p1(%) 37 23 10 0 4 0 0 26
A1=250 ha c1(%) 100 100 100 0 100 0 0 40
74,261 p2(%) 20 5 28 11 12 0 5 19
A2=185 ha c2(%) 100 100 95 100 100 0 100 80
18,542 p3(%) 10 18 3 0 0 42 0 27
A3=57 ha c3(%) 100 100 100 0 0 100 0 35
42,149 p4(%) 25 28 20 2 15 0 0 10
A4=88 ha c4(%) 100 100 95 100 100 0 0 36
22,156 p5(%) 50 0 11 0 10 0 0 29
A5=54 ha c5(%) 100 0 100 0 100 0 0 65
9958 p6(%) 24 11 13 15 13 8 0 16
A6=29 ha c6(%) 100 100 100 100 100 100 0 35
8517 p7(%) 22 28 8 19 6 0 10 7
A7=17 ha c7(%) 100 100 100 100 100 0 100 50
12,560 p8(%) 0 0 0 0 55 20 15 10
A8=16 ha c8(%) 0 0 0 0 85 100 100 45

Determine the total average demand of the city.

Answer:

Based on Formula 2.11, a sample calculation of the demand for District 1 will be as follows:

5
Q1,avg = A∑ (q a ,i pi ci ) = 250 × (90 × 0.37 + 55 × 0.23 + 125 × 0.1 + 100 × 0.04 + 15 × 0.26 × 0.40) = 666.77 m 3 / h
i =1

The remainder of the results are shown in the table below. The specific demand has been calculated
based on the registered population in each district.
2.28 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Qavg (m3/h) Population Qavg (l/c/d)

District 1 666.67 86,251 186

District 2 651.97 74,261 211

District 3 216.76 18,542 281

District 4 288.90 42,149 165

District 5 161.05 22,156 174

District 6 99.74 9958 240

District 7 58.03 8517 164

District 8 67.95 12,560 130

TOTAL 2211.16 274,394 193

Self-study: Spreadsheet lesson 8.5 & 8.6 (Appendix 5)

2.5 DEMAND FORECASTING

Water demand usually grows unpredictably as it depends on many parameters that

have their own unpredictable trends. Figure 2.24 illustrates how the rate of increase in
consumption may differ even in countries from the same region and with a similar
level of economic development.

Figure 2.24. Domestic consumption increase in some European countries (Source: EEA, 2001).

The experience from Germany proves again how uncertain the forecast can be. Figure
2.25 shows the development of domestic consumption in the period 1970 – 2000. The
forecast made by experts in the Seventies and the Eighties was that the demand in the
 Water demand 2.29
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

year 2000 would grow to as much as 220 l/c/d, while in reality it fell to approximately
140 l/c/d.

Figure 2.25. Domestic consumption increase in Germany (Source: BGW).

Awareness for the environment in the last decade, combined with low population
growth, caused a drop in domestic water use in many countries of Western Europe. In
addition, lots of home appliances (i.e. shower heads, taps, washing machines,
dishwashers, etc.) have been replaced with more advanced models, able to achieve the
same effect with less water (see Figure 2.26). Finally, industry has been moving
towards alternative water-saving production technologies, positively influencing the
overall water demand. This is unfortunately not the case in many developing
countries, where the population growth is much higher, consumers’ attitude towards
water conservation is comparatively low, and outdated technologies and equipment
are still widely used.

Figure 2.26 Water consumption of washing appliances in Europe (Source: EEA, 2001).
2.30 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Apart from monitoring technological developments, several other assessments must be

taken into account while estimating future demand:
- historical demand growth trends,
- projection-based on per capita consumption and population growth trends for
domestic category,
- forecast-based on assessment of growth trends of other main consumer
categories (industry, tourism, etc.).

When combined, all these projections can yield several possible scenarios of
consumption growth. While thinking, for instance, about population growth, which is
for many developing countries still the major factor in an increase in water demand,
useful conclusions can be drawn if the composition of the existing population, fertility
and mortality rates, and particularly the rate of migration, can be assessed. That the
population and demand growth match reasonably well in general is shown by the
example in Figure 2.27.

Figure 2.27. Population and demand growth in The Netherlands (Source: VROM).

Two models are commonly used for demand forecast: linear and exponential.

Linear model
 a 
Qi + n = Qi 1 + n  (2.12)
 100 

Exponential model
n
 a 
Qi + n = Qi 1 +  (2.13)
 100 
 Water demand 2.31
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

In the above equations:

Qi = Water demand at year i.
Qi+n = Forecast water demand after n years.
n = Design period.
a = Average annual population growth during the design period (%).

Which of the models will be more suitable will depend on the conclusions from the
above-mentioned analyses. These are to be reviewed periodically, as trends can
change within a matter of years.
Figure 2.28 shows the annual demand in The Netherlands in the period 1955 -
1995. In the first part of this period, up until 1970, the exponential model with an
average annual growth of 5.5 % (a in Equation 2.13 = 5.5) matches the real demand
very closely. However, keeping it unchanged for the entire period would show
demand almost three times higher than was actually registered in 1995.

Figure 2.28. Consumption growth in The Netherlands according to the exponential model.

Problem 2-5

In a residential area of 250,000 inhabitants, the specific water demand is estimated at 150 l/c/d, which
includes leakage. Calculate the demand in 20 years’ time if the assumed annual demand growth is 2.5
%. Compare the results by applying the linear and exponential models.

Answers:

The present demand in the city is equal to:

250,000 × 150 × 365

Qavg = = 91,250 m 3 / y
1000

Applying the linear model, the demand after 20 years will grow to:
2.32 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

 2.5 
Q21 = 91,2501 + 20  = 136,875 m / y
3

 100 

which is an increase of 50 % compared to the present demand. In the case of the exponential model:

20
 2 .5 
Q21 = 91,2501 +  ≈ 149,524 m 3 / y
 100 

which is an increase of approximately 64 % compared to the present demand.

Self-study: Workshop problems 1.9 & 1.10 (Appendix 1)

2.6 DEMAND FREQUENCY DISTRIBUTION

A water supply system is generally designed to provide the demand at guaranteed

minimum pressures, for 24 hours a day and 365 days per year. Nevertheless, if the
pressure threshold is set high, such a level of service may require exorbitant
investment that is actually non-affordable for the water company and consumers. It is
therefore useful to analyse how often the maximum peak demands occur during the
year. The following example explains the principle.

Figure 2.29.Example of a typical diurnal demand pattern.

Knowing both a typical diurnal peak factor diagram (such as the one shown in Figure
2.29) and the range of seasonal peak factors allows for integration of the two. The
hourly peak factors corrected by the seasonal peak factors will result in the annual
range of the hourly peak hours (Figure 2.30). These are absolute values that refer to
the average hour of the average consumption day (Figure 2.31). Consequently, each
hour of the year (total 24×365) will have a unique peak factor value assigned to it.
 Water demand 2.33
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.30. Example of the annual range of the peak factors.

Figure 2.31. Example of the annual range of hourly demands.

Applying this logic, the diagram with the frequency distribution of all hourly peak
factors can be plotted, as the example in Figure 2.32 shows. Converting this diagram
into a cumulative frequency distribution curve helps to determine the number of hours
in the year when the peak factors exceed the corresponding value. This is illustrated in
Figure 2.33, which for instance shows that the peak factors above 2.0 only appear
during some 500 hours or approximately 5% of the year. In theory, excluding this
fraction from the design considerations would eventually create savings based on a
20% reduction of the system capacity. The consequence of such a choice would be the
occasional drop of pressure below the threshold, which the consumers might consider
acceptable during a limited period of time.
2.34 Introduction to water transport & distribution
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Figure 2.32. Frequency distribution of the diurnal peak factors.

Figure 2.33. Cumulative frequency distribution of the diurnal peak factors.

In practice, the decision about the design peak factor results from the comparison
between the costs and benefits. Indeed, it seems rather inefficient to lay pipes that will
be used for 80% of their capacity for less than 5% of the total time. However, where
there is a considerable scope for energy savings from daily use by lowering the energy
losses on a wider scale, such a choice may look reasonable. Moreover, careful
assessment of the network reliability could justify the laying of pipes with reserve
capacity that could be utilised during irregular supply situations. Finally, some spare
capacity is also useful for practical reasons since it can postpone the construction of
phased extensions to expand the system.

The relation between demand for water and hydraulic losses is thoroughly discussed in
the following chapter.