ArticlePDF Available

Analysis of water savings: A case study during the 2004/05 water restrictions in Cape Town

Authors:

Abstract and Figures

In October 2004 the City of Cape Town implemented water restrictions in line with the city's holistic approach to water demand management. To better understand the savings and improve the effectiveness of future restrictions, a detailed demand analysis was conducted during this period. In this paper the authors report on various aspects pertaining to the savings achieved. Emphasis is placed on the residential sector, but savings from other sectors are also presented. This is the first reported work in South Africa on the analysis of metered water demand and savings obtained in different consumer categories during water restrictions. Monthly metered consumption prior to and during water restrictions was recorded and analysed. The analysis shows that the water restrictions resulted in notable water savings in all administrative areas and for all land use, stand size and income categories included in the analysis. The residential sector is found to have a significant contribution to the total water saving. The paper explains the nature of these savings and also addresses the pitfalls and successes of the project.
Content may be subject to copyright.
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 0716
Analysis of water
savings: a case study
during the 2004/05 water
restrictions in Cape Town
H E Jacobs, L Geustyn, K Fair, J Daniels and J A du Plessis
TECHNICAL PAPER
JOURNAL OF THE SOUTH AFRICAN
INSTITUTION OF CIVIL ENGINEERING
Vol 49 No 3,
2007, Pages 16–26, Paper 637
HEIN Z JACOB S is a seni or lec ture r at the U niver sit y
of Stell enbos ch. He joined t he universi ty as p ost-
doct oral fellow in A pril 2 006 . The fo cus are as of his
work over the pa st de cade in clude u rban w ater
demand and wa ter system maste r planning. Af ter
comple ting mechanical an d civil e ngine erin g degr ees at
the R AU (now Univer sit y of Joha nnesburg), he jo ined
Mer z & McLel lan (S A) Consultin g Engin eers. He th en join ed Kwezi V 3 and
spent seven ye ars in Worcest er and thereafte r four yea rs at G LS Con sulting
Engin eers in Stell enbos ch. Be fore joining the universi ty he s pent a ye ar wit h
the Ca nadia n consul ting fi rm QDS in B rit ish Col umbia as a spec ialis t in wat er
master planning.
Contact details:
Depar tment of Civil Engineer ing, U niversit y of Stellenbosch
Priv ate Bag X1, Matiela nd, 7602
T +27-21-808- 4059, hejacobs @sun. ac. za
LEON G EUST YN is a co-foun der (1989) and dire ctor
of GLS a s well a s soft war e developmen t companies
IMQS a nd GLS Engineeri ng Sof twar e. He o btai ned
BEng ( Civil ), MEng and PhD Eng de gree s from the
Unive rsi ty of Ste llenbosch. O ver th e past 17 years
Leon ha s been t he project leader fo r the com pleti on
of wate r and sewer sys tem analysis and pla nning fo r
sever al mun icipal ities rang ing fro m small t owns to la rge ci ties in s outh ern
Afri ca. He has bee n involved in the applicatio n of managemen t infor matio n
systems at most of th ese mun icipalitie s.
Contact details:
Geus tyn Loubse r Streicher Civil Consult ing Eng ineer s (GLS)
PO Box 814, Stel lenbo sch, 7599
T +27-21-880-0388, Leon@GLS.co.za
KERRY FAIR is an e ngineer at GL S and sof tware
developer for G LS Enginee ring S oft ware . Since she
joine d GLS in 199 9 she has been responsible fo r the
desig n and codi ng of sof twa re pro ducts, in cluding
Swif t (demand modelling ) and Sewsan (s ewer
model ling) . She is exper ienced in sewe r and wa ter
master planning. K err y obt ained a B Eng (Civil ) from
the Univer sity o f Stellenbosch and an M Sc (Ci vil Eng ) from t he Uni vers ity of
Nata l. She s tar ted her car eer wi th BK S, whe re she g ained experience in wate r
resou rce pla nning a nd river mode lling .
Contact details:
Geus tyn Loubse r Streicher Civil Consult ing Eng ineer s (GLS)
PO Box 814, Stel lenbo sch, 7599
T +27-21-880-0388, Kerry@GLS.co.za
JUL IAN DAN IEL S is Man ager : Water Demand
Manag ement for the Ci ty of C ape Town. Su bsequent
to his appoint ment in 2 005 , Julia n has be en
instrumental in numerous innovative and successful
projects being implemented. These projects include
nalisation of demand strategy; reuse of treated
effl uent; i ntegrated water l eaks across the cit y;
pressure r educ tion initia tives ; and tarif fs and rest ric tions .
Contact details:
City o f Cape Town
T +27-21-590-1601, julian.daniels@capetown.gov.za
KOBUS D U PLES SIS is a l ectu rer in H ydrol ogy an d
Envir onmental Engineering at the Univer sit y of
Stellenbosch. He ha s 20 year s’ exp erie nce in th e water
eld. B efore joinin g the Univer sity o f Stellenbosch he
ser ved as direc tor in the wate r divis ion of th e West
Coas t Distric t Muni cipality an d also worked fo r
DWAF and th e City of Cape Town . He obt ained B Eng
(Civ il) and M Eng ( Water Re source Manag ement) degrees fro m the Un iver sit y
of Stellenbosch.
Contact details:
Depar tment of Civil Engineer ing, U niversit y of Stellenbosch
Priv ate Bag X1, Matiela nd, 7602
T +27-21-808- 4358, jad up @s un.ac.z a
Keywords: resi dential water consu mptio n, wat er res tri ctio ns, wa ter
saving, water demand management
In October 2004 the City of Cape Town implemented water restrictions in line with
the city’s holistic approach to water demand management. To better understand the
savings and improve the effectiveness of future restrictions, a detailed demand analysis
was conducted during this period. In this paper the authors report on various aspects
pertaining to the savings achieved. Emphasis is placed on the residential sector, but
savings from other sectors are also presented. This is the first reported work in South
Africa on the analysis of metered water demand and savings obtained in different
consumer categories during water restrictions. Monthly metered consumption prior to
and during water restrictions was recorded and analysed. The analysis shows that the
water restrictions resulted in notable water savings in all administrative areas and for
all land use, stand size and income categories included in the analysis. The residential
sector is found to have a significant contribution to the total water saving. The paper
explains the nature of these savings and also addresses the pitfalls and successes of
the project.
INTRODUCTION
Background
This detailed investigation of water con-
sumption of users in the City of Cape Town
(COCT) and water saving during restrictions
provides interesting and useful information
on a topic often riddled with guesswork and
estimates. The study is a comprehensive
analysis of the metered water consump-
tion in the COCT prior to and during water
restrictions implemented at the end of 2004
(COCT 2005). Comprehensive demand
analysis is achieved by the application of
recognised computer models.
The study area includes the six adminis-
tration areas in the COCT at the time of this
study, namely Blaauwberg, Cape Town City,
Helderberg, Oostenberg, South Peninsula and
Tygerberg, as depicted in figure 1 (page 17).
The results provide insight into the
water savings achieved, thus making this
paper particularly relevant for water manag-
ers and practitioners alike. Further research
in this field is also encouraged by the
results.
Objective and approach
The study objectives include implementation
of a comprehensive water demand model for
the entire study area in order to investigate
the water savings achieved. The objectives
are achieved by implementing the following
step-wise approach:
Extraction of individual water meter read-
ings from the COCT’s treasury system for
all customers and the two periods applica-
ble to this study, namely 1 October 2003
to 1 April 2004 and 1 October 2004 to
1 April 2005
Analysis of the above data, mainly by
means of the Swift software tool
Investigation of the water saving that was
realised by the water restrictions. This was
achieved by comparison of the individual
water meter readings of the summer peri-
ods 1 October 2004 to 1 April 2005 with
1 October 2003 to 1 April 2004
The October 2004 water restrictions
The COCT receives water mainly from the
Berg and Breede R iver catchments, and this
water is shared with a number of other users
like agriculture and smaller local authori-
ties. The management of these resources
is the responsibility of the Catchment
Management Agency (CMA) as per the
National Water Act. In the absence of a
CMA – as is the case for the Berg River sys-
tem – the Department of Water Affairs and
Forestry (DWAF) executes this function.
In this regard, a number of stakehold-
ers are involved with the DWAF in a forum
known as the Western Cape Water System
Consultative Forum (WCWSCF). The main
objective of the WCWSCF is to monitor
the levels of the different supply reservoirs
(dams) and to make decisions regarding the
required curtailment in demand and subse-
quent restriction levels. Decisions made by
the WCWSCF are aimed at ensuring a sus-
tainable supply to all water users in the Berg
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 07 17
River system. These possible drought mitiga-
tion measures are assessed by the WCWSCF
on an ongoing, monthly basis in the spirit
of the COCT’s water demand management
(WDM) programme.
During the winter season of 2004 (rainy
season), the catchment areas of the main sup-
ply systems to the Cape Town area received
a mere 56 % of the average historical rainfall
for this period. The combined storage of
all the reservoirs in the system is shown in
figure 2. It dropped to only 57 % and on 1
Februar y 2005 was about 20 % lower than
the storage at the same time in the previous
year, which in turn was 20 % lower than the
previous year. The WCWSCF recommended
that restrictions be imposed to achieve at
least a 20 % reduction in water demand com-
pared to the 2003 consumption figures. A
tool used for this purpose by the WCWSCF
is shown in figure 3 and presents the level of
curtailment required for a given storage vol-
ume at a given time (DWAF 2005).
The COCT responded to this require-
ment with an official notice calling for
Level 2 restrictions in accordance with their
bylaws, with effect from 1 Januar y 2005.
An awareness campaign was initiated three
months prior to this date in October 2004.
Level 2 restrictions have a target of 20 %
water saving and entail the following:
Limited watering of gardens, lawns and
public open spaces – watering is only
allowed between the hours of 18:00 and
10:00 and up to twice a week
The use of hosepipes for washing motor
vehicles, motor boats, paths and paved
areas is prohibited
Limited use of irrigation systems is
allowed – only a single hand-held hose
(fitted with a control nozzle) or buckets
may be used
Automatic f lushing urinals (AFUs) are
to be turned off in all public buildings
when vacated and are prohibited in new
buildings
Increased water and sanitation tariffs to
ensure cost recovery and discourage high
consumption patterns
The measures pertaining to Level 2 restric-
tions listed above were pre-designed, form
part of the holistic approach to WDM in the
COCT and did not form part of this investi-
gation or analysis. Such ‘drought measures’
call for immediate action with a focus on
customer awareness, increased tariffs and
effective enforcement of regulations by
policing.
Comments regarding the effectiveness
of the particular measures that were imple-
mented, or improvement of the measures
and WDM programme, are beyond the
scope of this investigation. However, it is
worth noting that some of the measures
implemented during Level 2 restrictions
have subsequently been incorporated in the
COCT’s new Water Bylaw. The new bylaw is
explicit in regards to such issues, for exam-
ple the AFU regulation outlined above has
been included in the bylaw.
METHODOLOGY
Analysis of treasury data
The COCT uses the SAP treasury system to
record all water meter and billing informa-
Figure 2 Storage history for the Berg River system ( DWAF, 2005)
Figure 1 Administration areas in the City of Cape Town (at time of this study)
Blaauwberg
Tyger ber g
Oostenberg
Cape Town
South Peni nsula
South Pe nins ula
Cape Town
Blaauwberg
Tyge rb er g
Oostenberg
Helderberg
Legend:
Helderberg
Gross storage
100%
80%
60%
40%
20%
0%
1 April 1996
1 April 1997
1 April 1998
1 April 1999
1 April 2000
1 April 2001
1 April 2002
1 April 2003
1 April 2004
1 April 2005
Inaccessible, poor qual ity water
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 0718
tion. Water meter readings obtained from
treasury systems have been analysed in previ-
ous research ( Jacobs et al 2004; Van Zyl et al
2007) and a similar approach was considered
appropriate for this study. In order to com-
pare pre- and post-restriction consumption, it
was necessary to extract data from the older
treasury systems’ data bases that used to be
in use in the different administration areas
in Cape Town prior to the implementation
of SAP. Description of the data extraction
process is beyond the scope of this investiga-
tion; suffice it to say that the monthly water
consumption information could successfully
be obtained in a format suitable for demand
analysis and direct compar ison for all users
recorded in the SAP system.
The Swift software program for demand
analysis was used in previous studies
(Jacobs et al 2004; Van Zyl & Geustyn
2007) and is a tool for extracting and ana-
lysing water demand data from municipal
treasury systems. As part of this investiga-
tion Swift was used and the data was also
linked to the COCT’s geographic informa-
tion system (GIS) cadastral data base. This
made it possible to spatially inspect the
results.
The demand modelling software was
set up to produce different data sets, one
for each of the administrative areas in the
COCT. This was advantageous for three
reasons:
The method of data storage in SAP and
the previous (older) treasury systems
is simplified by first extracting records
according to administrative area
Verification of data and results is possible
– at various stages in the process – by
comparing the sum of the six administra-
tive areas’ results to the total COCT data
set
The COCT is interested in results for each
administrative area from a management
point of view, while the complete result
set – split along different lines – is valu-
able for research purposes, as discussed
in this text
In all six administrative areas the complete
record period is 1 February 2003 to 30
April 2005 and covers at least one summer
season prior to and after the implementation
of water restrictions. The complete period
of record was split into different time series
sections, of which the following two pertain
to this study:
Period 1: 1 October 2003 to 1 April 2004
– the summer average daily water demand
(SADD) for the pre-restriction period,
used to report the summer seasonal
water use prior to the implementation of
restrictions
Period 2: 1 October 2004 to 1 April
2005 – the SADD for the period during
restrictions
Individual users’ readings from Period 1
and Period 2 are compared directly in this
study to obtain information about changes
in summer use brought about by restric-
tions. The summer period is particularly
relevant, because most components of the
Level 2 restrictions target outdoor use,
which in turn contributes significantly to
the seasonal peak. For this reason these two
six-month ‘summer’ periods form the focus
of this study.
The different periods of analysis and
the fact that individual meter records could
be identified in each of the five adminis-
trative areas enable substantial resolution
to be obtained for analysis of the savings
achieved. The four parameters used exten-
sively in this study are water meter readings,
land use codes, stand size and property
valuation. These values are recorded in the
SAP system and GIS of the COCT, for each
user. The data extracted from the SAP sys-
tem that survived screening based on record
quality (AADD > 0, SADD > 0) was consid-
ered to be relatively robust in terms of this
analysis and was used w ithout amendment.
The records surviving this phase – about
95 % of those initially extracted – were used
for analysis.
Records lacking any of these four param-
eter values (say, with no land use code) were
grouped together during analysis in appro-
priate categories. In addition to the initial
robust screening phase the built in function-
ality of the demand analysis software was
applied to identify and correct typical inci-
dences relating to water meter readings (for
example meter clock-overs or replacement).
Land use codes
The land use code for a record is recorded
in SAP, thus enabling the statistics of water
demand to be broken down into the cor-
responding land use categories. The land
use codes in the complete data set include
43 different codes in Blaauwberg, 49 in
Cape Town Central, 44 in Helderberg, 42 in
Oostenberg, 50 in South Peninsula, and 51
in Tygerberg. In order to compare the dif-
ferent data sets, 16 land use categories were
defined and used in the demand analysis by
mapping the different codes in each admin-
istrative area to these 16 codes. For presen-
tation of results in this paper the 16 codes
were grouped into the following four land
use categories that are particularly relevant
to this study:
Residential – the residential land use cat-
egory represents only single residential
dwellings and forms the crux of this study
Industrial, commercial and institutional
(ICI) – the land uses grouped in this cat-
egory include ‘industrial’, ‘business’, ‘com-
mercial’, ‘institutional’, ‘education’ and
‘government’
Unknown – a category created for all
records with no land use codes in SAP, or
codes ref lecting an unknown use
Other – a category representing all
remaining land uses recorded in the data
base, including for example ‘group hous-
ing’, ‘flats, ‘farmland’ (agricultural hold-
ings), ‘medical’, ‘other’, ‘parks’ and ‘sport’
However, in this paper the focus is on resi-
dential land use, since it is the most notable
and best defined category of those consid-
ered in the study and contributes most to
water savings in this case.
Scope, limitations and focus of this study
The analysis of water use as per this study
includes only those water consumers who
are included in SAP, thus being metered
consumers. The COCT data extract from
SAP includes all water meter records for
both paying and non-paying customers.
However, because of ethical and confiden-
tiality issues the information regarding cus-
tomers’ payment history was withheld dur-
ing the extract procedure. Water saving by
unmetered water users is thus not included
in this study, while saving by all metered
users – both paying and non-paying (with
no means to distinguish between the two)
– is included. Savings by consumers mak-
ing use of stand pipes and other means
of supply are often not included in the
treasury system (where included it would
Storage
700
600
500
400
300
200
100
0
Gross storage (%)
100 %
90 %
80 %
70 %
60 %
50 %
40 %
30 %
20 %
10 %
0 %
Date
1 Jul 05 1 Aug 05 1 Sep 05 1 Oct 05 1 Nov 05
10 % curtailment 20 % curtailment 30 % curtailment >30 % curtailment
Figure 3 Required curtailment levels used by the WCWSF ( DWAF, 2005)
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 07 19
resort under ‘other’ land use category in
this study). Although the use of SAP records
could be viewed as a limitation in analysing
total water supply volumes (and losses), it
provides the researchers with valuable focus
to better understand changes by metered
consumers at the end-user level.
The focus of this text is the scope of sav-
ings achieved and identification of groups
and consumer categories responsible for such
savings. For the purpose of this analysis,
such savings are considered to be the direct
result of the Level 2 restrictions implement-
ed on 1 January 2005 and the awareness
campaign initiated in October 2004, three
months prior to the formal implementation
date – in time to reduce the 2004/2005 sum-
mer demand. Apart from restrictions, vari-
ous factors are known to influence demand,
such as price (Arbués et al 2004; Van Zyl et
al 2003; Espey et al 1997), income (Van Zyl
et al 2007), geographic location (Jacobs &
Haarhoff 2004), weather parameters (Kulik
1993) and system pressure (Haarhoff et al
2002; Gebhardt 1975).
The Level 2 restrictions target various
aspects of consumer behaviour while fac-
tors such as those mentioned above simul-
taneously influence demand, leading to a
complex interaction. In the process water
demand is adjusted by users, part of which
could be ascribed to the water restrictions.
This study aims to provide insight into the
specific water savings brought about by the
combined effect leading to demand reduc-
tion during the specified restriction period.
Explanation of the complex relationship
between the explanatory variables inf luenc-
ing demand and the sav ings noted is beyond
the scope of this paper.
The impact of growth in water demand
with time, due to development, is accounted
for in this study by considering the unit
water demand of users, instead of the total
demand. Analysis of unit water demand is
possible since the number of water users is
known by consumer category for each period
of analysis. However, the total demand and
volume of savings over the period of analysis
is also included in this text to ensure a com-
prehensive presentation of results.
Stand size (A) and property value (PV)
are used in this study to investigate the sav-
ings achieved. The modern tendency towards
high-income group housing with individual
title deeds and individual water meter con-
nections complicates the smaller size catego-
ries. Small stands might have a high PV or
low PV. This is not often the case for large
properties. It is generally accepted that small
stands have low values, but this is not always
the case. This aspect could be included in
future investigations to better understand
the savings achieved by the smallest stand
size category. The number of low PV stands
in the smallest size category (typical of RDP-
type housing schemes) is considered for the
purpose of this work to notably outnumber
those stands with a high PV.
Water saving – a limitation
The term ‘water saving’ should be viewed
in the correct perspective. The word ‘save’
is defined as ‘to avoid the spending, waste,
or loss (of something)’ and ‘to treat with
care so as to preserve (Collins 2004). For
this reason the term ‘water saving’ is often
used in relation to water demand reduction.
However, without additional explanation
the term could be misinterpreted completely
because the SAP data is analysed instead of
‘actual water use’ – given analysis of munici-
pal water meter readings, as is the case here.
The water use records extracted from
SAP in this study are based on the corre-
sponding consumer meter readings and is an
indication – hopefully a relatively accurate
one – of how much water the consumer
actually uses. In addition to meter error, the
meter reading does not represent actual use.
This is illustrated by, for example, customers
bypassing a municipal water meter with an
illegal connection, thus creating the percep-
tion of a ‘saving’ from the municipal supply
because of the reduction in metered demand.
Boreholes are another example: when
water supply from the municipal system is
replaced by a borehole to meet a consumer’s
demand (or part thereof), the water is in tur n
abstracted from an alternative water source
(groundwater aquifer). From an environ-
mental viewpoint it could be argued that the
water is not ‘saved’, since the efficiency of
use does not improve in such cases.
Users with rainwater tanks or on-site
reuse technology may be ‘sav ing water’ by
making use of more effective means, also
leading to a ‘saving’ being recorded as a
result of reduced municipal water meter
reading – despite no change in the actual
water used by end-users on the property.
For the purpose of this study the change
recorded by the municipal water meter at
a particular property with a metered water
connection, as recorded in SAP, is consid-
ered to represent a change in the consumers’
water use. Two subsequent summer seasons
are evaluated for this purpose. Any reduc-
tion in the consumers’ water use is thus con-
sidered to constitute a water saving.
OVERVIEW OF RESULTS
Sample size
The relatively large sample size used for
analysis is reflected by the data shown in
table 1. The total number of records for
Periods 1 and 2 are 526 341 and 552 034
Table 1 Number of water users in Period 1 and Period 2
Administration
area
Residential ICI Other Unknown TOTAL
Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 1 Period 2
Actual number of water users
Blaauwberg 35 631 39 364 1 549 1 725 977 1 203 202 258 38 359 42 550
Cape Town 143 666 146 598 6 073 6 288 5 520 5 975 16 870 17 065 172 129 175 926
Helderberg 28 433 30 437 1 312 1 432 2 992 3 340 879 1 051 33 616 36 260
Oostenberg 57 316 59 555 1 524 1 607 723 803 4 570 4 844 64 133 66 809
South Peninsula 61 951 65 657 1 842 2 794 2 099 2 909 2 515 4 339 68 407 75 699
Tygerberg 136 475 140 473 4 179 4 556 2 093 2 511 6 950 7 250 14 9 6 97 15 4 790
COCT (all areas) 463 472 482 084 16 479 18 402 14 404 16 741 31 986 34 807 526 341 552 034
Expressed as a percentage of the total number of water users (%)
Blaauwberg 92,9 92,5 4,0 4,1 2,5 2,8 0,5 0,6 100, 0 100 ,0
Cape Town 83,5 83,3 3,5 3,6 3,2 3,4 9,8 9,7 100, 0 100 ,0
Helderberg 84,6 83,9 3,9 3,9 8,9 9,2 2,6 2,9 100, 0 100,0
Oostenberg 89,4 89,1 2,4 2,4 1,1 1,2 7,1 7,3 100,0 100,0
South Peninsula 90,6 86,7 2,7 3,7 3,1 3,8 3,7 5,7 10 0,0 10 0,0
Tygerb erg 91,2 90,8 2,8 2,9 1,4 1,6 4,6 4,7 100, 0 100,0
COCT (all areas) 88,1 87,3 3,1 3,3 2,7 3,0 6,1 6,3 100,0 100,0
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 0720
respectively. A relatively small fraction of
about 6 % of the records have an ‘unknown’
land use allocation in the data base and is
not a concern in view of this work. All such
records were grouped as ‘unknown’ during
the analysis as a unique set in order to pre-
vent them from influencing the results for
the ‘known’ land use categories.
Owing to spatial development the
number of records increases from Period 1
to Period 2 for all land use categories in all
six administration areas. With consideration
for the number of users, the residential land
use is dominant and contributes 88,1 %
and 87,3 % to the total for the COCT in
Periods 1 and 2 respectively. In Period 2 the
fraction varies between the administration
areas from 83,3 % in the Cape Town City
area to 92,5 % in Blaauwberg.
This study encompassed the analysis
of 463 472 residential users’ water demand
prior to restrictions (Period 1) and 482 084
during restrictions (Period 2). However, only
those data points present in both periods
– that is, 463 472 users – could be used
to investigate savings. The size of the data
set makes it the largest reported analysis of
water savings at individual properties that
could be traced in the literature review of
local and international sources.
Total water demand
A summary of the actual total water use is
presented in table 2, showing the consump-
tion as an average daily value (in kl/d). The
following interesting characteristics are noted
when inspecting the results for total water use:
The total demand for all users in all
land use categories combined reduced
by 97 627 kl/d (±14 % decrease) from
Period 1 to Period 2, despite the number
of users increasing over the corresponding
period by 25 693 (±5 % increase)
The residential consumer category con-
tributes about 55 % to the total demand
The total demand for the residential con-
sumer category reduced by 59 450 kl/d
(±15 % decrease) from Period 1 to
Period 2, despite the number of users
increasing over the corresponding period
by 18 612 (±4 % increase)
For the COCT as a whole, the water used
by ICI, ‘other’ and ‘unknown’ land use
categories also decreased from Period 1 to
Period 2, with a corresponding increase in
the number of users
The residential fraction of the total vol-
ume of water used contributes respec-
tively 55,1 % and 54,2 % to the total
water use in the COCT in Periods 1 and
2. These fractions are significantly smaller
than the fractions of 88,1 % and 87,3 %
stated earlier for the number of users.
This result is a common phenomenon
and is ascribed to water users in the ICI
category contributing little to the number
of users, but significantly to the volume of
water used
Unit water demand
The Period 1 and Period 2 summer unit
water demand (SUWD) is shown in table 3
for all land use categories and all adminis-
tration areas. The following interesting char-
acteristics are noted when inspecting the
results for unit water use:
The unit demand for residential users is
the lowest of all land use categories, with
ICI, unknow n and other type users hav-
ing substantially higher unit demands in
most cases
There is notable variation in the residential
unit demand between different administra-
tion areas, with values for Period 1 ranging
Table 2 Summer average daily water demand (SADD) for Period 1 and Period 2
Administration
area
Residential ICI Other Unknown TOTAL
Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 1 Period 2
SADD (kl/d)
Blaauwberg 33 160 27 868 4 218 3 897 21 324 22 359 502 431 59 203 54 554
Cape Town 109 511 99 244 13 435 12 453 75 497 70 849 11 589 10 089 210 033 192 635
Helderberg 27 034 20 407 2 156 2 044 21 424 14 035 988 880 51 602 37 367
Oostenberg 45 326 38 207 3 327 2 992 12 078 15 016 3 296 2 940 64 026 59 155
South Peninsula 58 696 47 083 3 772 4 039 20 427 24 600 2 906 2 807 85 801 78 529
Tygerberg 110 848 92 317 9 976 8 993 100 705 71 901 5 350 4 464 226 879 177 676
COCT (all areas) 384 576 325 126 36 884 34 418 251 454 218 760 24 629 21 612 697 543 599 916
SADD expressed as a percentage of the total (%)
Blaauwberg 56,0 51,1 7,1 7,1 36,0 41,0 0,8 0,8 100,0 10 0,0
Cape Town 52,1 51,5 6,4 6,5 35,9 36,8 5,5 5,2 10 0, 0 100,0
Helderberg 52,4 54,6 4,2 5,5 41,5 37,6 1,9 2,4 100,0 100,0
Oostenberg 70,8 6 4,6 5,2 5,1 18,9 25,4 5,1 5,0 100,0 10 0,0
South Pe ni nsu la 68 ,4 6 0,0 4,4 5,1 23,8 31,3 3,4 3,6 100,0 10 0,0
Tygerberg 48,9 52, 0 4,4 5,1 44,4 40,5 2 ,4 2,5 10 0,0 10 0,0
COCT (all areas) 55,1 54,2 5,3 5,7 36,0 36,5 3,5 3,6 100,0 100,0
Table 3: Summer unit water demand ( SUWD) for Period 1 and Period 2
Administration
area
Residential ICI Other Unknown TOTAL
Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 1 Period 2 Period 1 Period 2
SUWD (l/stand·d)
Blaauwberg 931 708 2 723 2 259 21 826 18 586 2 484 1 671 1 543 1 282
Cape Town 762 677 2 212 1 980 13 677 11 858 687 591 1 220 1 095
Helderberg 951 670 1 643 1 428 7 160 4 202 1 124 838 1 535 1 031
Oostenberg 791 642 2 183 1 862 16 705 18 700 721 607 998 885
South Peninsula 947 717 2 048 1 446 9 732 8 456 1 155 647 1 254 1 037
Tygerberg 812 657 2 387 1 974 48 115 28 634 770 616 1 516 1 148
COCT (all areas) 830 674 2 238 1 870 17 457 13 067 770 621 1 325 1 087
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 07 21
from 762 l/stand·d in Cape Town City area
to 951 l/stand·d in Helderberg
The unit demand for users in the category
for ‘other’ land use shows the greatest var-
iation from one area to the next, probably
due to the poorly defined nature of users
grouped in this category
Water savings
The change in unit water demand from
Period 1 to Period 2 is calculated to obtain an
indication of the water saving brought about
by the restrictions. The results presented in
Table 4 show that the unit water demand
decreases substantially from Period 1 to
Period 2 for all land use categories in all six
administration areas, bar the ‘other’ land use
category in Oostenberg. An overall unit water
demand saving of 18 % was achieved, with
savings recorded throughout the entire spec-
trum of water users and administration areas.
The average reduction in residential unit
water demand from Period 1 to Period 2 in
the COCT is 155 l/stand (suggesting a sav-
ing of 19 %). This varies between 85 l/stand·d
in Cape Town City area (saving of ±11 %) to
280 l/stand·d in Helderberg (saving of ±30
%). This relatively large variation in sav-
ing between different administration areas
– represented by arbitrary boundaries with
no impact on water use – is best explained by
extending the analysis to include parameters
known to describe water use. Two character-
istics that proved the most sensitive were:
Stand size – known to explain water
demand ( Jacobs et al 2004), stand size is
available from the data set for each prop-
erty and is well suited for data analysis
Household income – this is another known
explanatory variable for demand and is
often replaced by a suitable surrogate,
namely stand value. The stand value, based
on property valuations recorded in the data
base, is readily available for further analy-
sis and has been used for this purpose in
previous research (Van Zyl et al 2007)
RESIDENTIAL WATER SAVINGS
BY STAND SIZE
Figure 4 shows a frequency histogram of the
residential SUWD for the following stand
size categories and for Periods 1 and 2:
A < 1 m2, in other words, no or incorrect
stand size is recorded in the data base
A 500 m2, excluding cases where
A < 1 m2
500 m2 < A 1000 m2
1000 m2 < A 1500 m2
1500 m2 < A 2000 m2
A > 2 000 m2
The unit water demand increases with
increased stand size category for both
Period 1 and Period 2. Also, when compar-
ing the unit water demand for each stand
size category in Periods 1 and 2, it is evident
that the unit water demand remains rela-
Table 4 Unit water demand saving
Administration area Residential ICI Other Unknown Total
SUWD reduction from Per iod 1 to Period 2 (l/stand·d)
Blaauwberg 223 464 3 240 813 261
Cape Town 85 232 1 819 96 125
Helderberg 280 216 2 958 286 505
Oostenberg 149 321 -1 996 114 113
South Peninsula 230 602 1 275 508 217
Tyg er b er g 155 413 19 4 81 1 54 368
COCT (all areas) 155 368 4 390 149 239
Saving achieved in comparison to initial unit water demand (%)
Blaauwberg 23,9 17,0 14,8 32,7 16,9
Cape Town 11,2 10,5 13,3 13,9 10,3
Helderberg 29,5 13,1 41,3 25,5 32,9
Oostenberg 18,9 14,7 -11,9 15,8 11,3
South Pe ni ns ula 24,3 29,4 13,1 44,0 17,3
Tygerberg 19,1 17,3 40,5 20,0 24,3
COCT (all area s) 18,7 16,4 25,1 19,4 18,0
SUWD (l/stand.d)
3 000
2 500
2 000
1 500
1 000
500
0
Period 1 (before restrictions) Period 2 (during restrictions)
633 635
1 126
1 681
2 167
2 612
488 586
816
1 046
1 278
1 647
1 500 m
2
< A 2 000 m
2
500 m
2
< A 1 000 m
2
A 1 m
2
1 000 m
2
< A 1 500 m
2
A 500 m
2
A > 2 000 m
2
Figure 4 Residential SUWD by stand size category for Period 1 and Period 2
Table 5 Residential unit water demand saving
Reduction from Period 1 to Period 2
Period 1 Per iod 2 Saving
(l/stand·d) (l/stand·d) (l/stand·d) (%)
Stand size categor y
A < 1 m
2
(note A) 630 490 140 22,2
A 500 m
2
640 590 50 7,8
500 m2 < A 1 000 m
2
1 130 820 310 27,4
1 000 m
2
< A 1 500 m
2
1 680 1 050 630 37,5
1 500 m
2
< A 2 000 m
2
2 170 1 280 890 41,0
A > 2 000 m
2
2 610 1 650 960 36,8
Property value category
R 0
(note A) 1 300 830 470 36,2
R 1 – R 200 000 820 680 140 17,1
R 200 000 – R 400 000 1 100 770 330 30,0
R 400 000 – R 600 000 1 390 910 480 34,5
R 600 000 – R 800 000 1 630 1 050 580 35,6
R 800 000 – R 1 000 000 1 870 1 190 680 36,4
R 1 000 000+ 2 440 1 620 820 33,6
Note
(A) These r ecords w ith no va lues in t he SAP s ystem ar e grouped together in one categor y and removed from the analysis.
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 0722
tively constant in the smaller size categories
while the larger size categories represent a
substantially reduced residential SUWD.
The water saving is summarised in
table 5, showing that the percentage saving
increases from 7,8 % for the smallest size
category to 36,8 % for the largest size cat-
egory. This increase in the saving achieved
with increased stand size is shown in fig-
ure 5. The histogram of residential SUWD
for Period 1 and Period 2 suggests a linear
trend. However, a linear fit to the data is not
attempted, since detailed statistical analysis is
beyond the scope of this study. This implies
that savings achieved in the COCT are lim-
ited to the periods under discussion and are
not intended for extrapolation to other areas.
It is also interesting from the viewpoint
of water saving to consider the total volume
of water used by all users combined in each
category. Figure 6 shows the frequency his-
togram of the total residential water demand
for each stand size category in Periods 1
and 2. The total demand decreases with
increased stand size category. For example,
the SADD by all users in the category for
A 500 m2 in Period 1 is 203 953 kl/d
(193 534 kl/d in Period 2). This is almost
20 times more than the total volume of
8 608 kl/d in Period 1 in the category for
1500 m2 < A 2000 m2.
With reference to figure 4, this is
in contrast to the increase noted in the
unit demand by stand size for the same
time periods. The largest volume of water
(203 953 kl/d in Period 1) is used by the
large number of users in the smallest
stand size category and the least water is
used by the small number of users in the
largest stand size category. This apparent
anomaly makes sense when considering
the two extremes: there are 318 676 users
in the size category for A 500 m2 and
only 5 569 users in the largest category,
A > 2 000 m2. Thus, there are about 57
times more users in the smallest size cat-
egory than in the largest category. The
unit demand for the largest size category is
2 612
l
/stand·d (Period 1), which is only four
times higher than the unit demand of 635
l
/
stand·d for the A 500 m2 category in the
same period. The consumers in the small
size categories are as important, if not more
so, than the larger properties in view of the
total saving achieved. Also, water losses on
private properties (often termed plumbing
leaks) are often considered to be more prev-
alent among properties in the smaller stand
size categories.
The following can be noted from the
stand size-based results:
The saving in unit water demand increas-
es approximately linearly with increased
stand size and varies between 50
l
/stand·d
and 960
l
/stand·d (table 5)
The category for A 500 m2 contributes
by far the most to the total volume of
water used in each administrative area
and for all time periods analysed, while
the two categories for A > 1 500 m2 con-
tribute least to the total demand (figure 6)
The highest residential SUWD is noted
for the largest stand size category, and the
smallest residential SUWD for the small-
est stand size category
There is a notable reduction in the resi-
dential SUWD for the large stand sizes
with water restrictions, but it is much less
notable in the smallest stand size category
The unit water demand increases approxi-
mately linearly with increased stand size
for both Periods 1 and 2, but the rate of
increase reduces in Period 2 (figure 5)
RESIDENTIAL WATER SAVINGS
BY PROPERTY VALUE
Consideration of the available data for prop-
erty value (PV) led to the final selection of
the following seven PV categories:
PV R 1 – a category to include all
records with a property value smaller
than R1, in other words those properties
with no value recorded in SAP
R1 < PV R200 000
R200 000 < PV R400 000
R400 000 < PV R600 000
R600 000 < PV R800 000
R800 000 < PV R1 000 000
PV > R1 000 000
The residential SUWD per valuation cat-
egory for Period 1 and Period 2 is shown in
figure 7. The unit water demand increases
with increased PV for both Period 1 and
Period 2. When comparing the unit water
demand for each PV category in Periods 1
and 2 the same finding is made as is the case
with stand size categories: the unit water
demand remains relatively constant from
Period 1 to Period 2 in the smaller PV cat-
egories, while the categories with larger PV
represent a substantially reduced unit water
demand in Period 2. The saving in unit water
demand is summarized in table 5, showing
that the percentage saving increases from
SUWD (
l
/stand.d)
3 000
2 500
2 000
1 500
1 000
500
0
Stand size categor y (m
2
)
500 501–1 000 1 001–1 500 1 501–2 000 > 2 000
Period 1 (Summer 2003/04) Period 2 (Summer 2003/04) Saving in SUWD
Figure 5 Residential SUWD and saving by stand size (refer to table 5)
SUWD (
l
/stand.d)
250 000
Period 1 (before restrictions) Period 2 (during restrictions)
5 372
1 500 m
2
< A 2 000 m
2
500 m
2
< A 1 000 m
2
A 1 m
2
1 000 m
2
< A 1 500 m
2
A 500 m
2
A > 2 000 m
2
200 000
50 000
100 000
150 000
0
203 953
115 853
36 254
8 608 14 53 6 6 0 81
193 534
87 035
23 435
5 309 9 732
Figure 6 Residential SADD by stand size category for Period 1 and Period 2
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 07 23
17,1 % for the smallest PV category to 33,6 %
for the largest PV category.
The increase in the saving achieved with
increased PV is similar to that for stand
size, as shown in figure 8. In accordance
with the result for stand size, the SUWD
values for both Period 1 and Period 2 are
approximately linear when plotted against
the PV categories, but this does not describe
a linear fit to the data for the same reasons
as stated before.
Figure 9 shows the frequency histo-
gram of the total volume of water used
in Periods 1 and 2 by all users com-
bined in each PV category. The total
demand decreases with increased PV
category. For example, the SADD by
all users in the smallest PV category,
R 1 PV < R200 000, in Period 1 is
242 800 kl/d (205 648 kl/d in Period 2),
while the total for the largest PV category
is 23 494 in Period 1 (16 235 kl/d in
Period 2), that is about 10 times less. This
decrease is in contrast to the increase
noted in the SUWD by PV for the same
time periods – a similar result to the one
obtained for the analysis by stand size.
The relatively large number of users with
low PVs are as important from the view-
point of total volume saved, if not more
so, than the significantly smaller number
of users in the larger PV categories.
The following can be noted from the PV-
based results:
The saving in residential SUWD varies
between 140
l
/stand·d and 820
l
/stand·d
(ta ble 5)
The two categories with the lowest PV
(R 1 PV < R200 000 and R200 000
PV < R400 000) contribute by far the
most to the total volume of water used in
each administrative area and for all time
periods analysed, while the three highest
value categories are responsible for the
smallest volumes of use (figure 9)
The residential SUWD and saving increas-
es approximately linearly with increased
stand size for both Periods 1 and 2, but
the rate of increase reduces in Period 2.
In other words, people living on proper-
ties with a higher PV use more water per
household than those on properties with
lower PVs (figure 8)
DISCUSSION
Water demand for the residential
consumer category
Although comparison of SADD and AADD
values may appear inappropriate, the SADD
from this study for the six-month summer
period could be superimposed on AADD
values. This would give an indication as to
whether the SUWD is in fact higher than
the theoretical A ADD, based on guideline
curves – as would be expected. The result
is shown in figure 10, where it is compared
to two recently published A ADD guide-
lines: the winter rainfall region proposed by
Jacobs et al (2004) and the coastal data set
by Van Zyl et al (2007). It is interesting to
note that the six-monthly summer ADD cor-
responds roughly to the guideline curves for
AADD values.
SUWD (
l
/stand.d)
3 000
2 500
2 000
1 500
1 000
500
0
Period 1 (before restrictions) Per iod 2 (during restrictions)
R600 000–R800 000
R400 000–R600 000R1–R200 000 R800 000–R1 000 000
R200 000–R400 000R0
1 301
821
1 102
1 390
1 301
1 870
2 436
832
679 773
905
1 050
1 190
1 622
R1 000 000+
Figure 7 Residential SUWD by value category for Period 1 and Period 2
SUWD (
l
/stand.d)
2 500
0
2 000
1 500
1 000
500
< 0,2 0, 2– 0,4 0,4–0,6 0,6–0,8 0,8–1,0 > 1,0
PV category (R-million)
Period 1 (Summer 2003/04) Period 2 (Summer 2004/05) Saving in SUWD
Figure 8 Residential SUWD and saving by property value (Refer to Table 5)
Figure 9 Residential SADD by value category for Period 1 and Period 2
300 000
250 000
200 000
150 000
100 000
50 000
0
SUWD (
l
/stand.d)
Period 1 (before restrictions) Period 2 (during restrictions)
R600 000–R800 000
R400 000–R600 000
R1–R200 000 R800 000–R1 000 000
R200 000–R400 000R0
1 824
242 80 0
117 001
18 164
R1 000 000+
47 372
10 093
23 494
5 561
205 648
84 346
31 763
12 031 6 585 16 235
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 0724
The SADD during restrictions is lower
than before restrictions – as expected – and
corresponds to the lower envelope, despite it
not being an annual average value. This sug-
gests that use of the AADD guideline curves
proposed by Jacobs et al (2004) would have
led to over-estimates for demand during the
Level 2 water restrictions in Cape Town.
This is despite the results by Jacobs et al
(2004) being much less conservative than
previous guidelines (CSIR 2000).
Water saving for the residential
consumer category
The results suggest that properties in
the lower size and PV categories (and
all administration areas independently,
although this is not shown in the graphs)
do not adjust water consumption consider-
ably with water restrictions, but those in
higher categories do. Garden water demand
is known to be relatively price elastic, while
indoor demand is considered to be inelastic.
In addition, mainly garden water use is tar-
geted by the Level 2 restr ictions – an end-
use more common to the properties with
a higher value or larger stand size, thus
leading to a more significant unit saving per
property in the larger categories than the
smaller ones.
For residential land use the saving
achieved in each administration area for the
smallest and largest stand size categories is:
Blaauwberg: 11,8 % reduction (A 500 m2)
and 40,2% (A > 2 000 m2)
Cape Town: 4,6 % reduction (A 500 m2)
and 31,7% (A > 2 000 m2)
Helderberg: 12,5 % reduction (A 500 m2)
and 44,0% (A > 2 000 m2)
Oostenberg: 12,2 % reduction (A 500 m2)
and 35,2% (A > 2 000 m2)
South Peninsula: 13,8 % reduction
(A 500 m2) and 35,3 % (A > 2 000 m2)
Tygerberg: 5,9 % reduction (A 500 m2)
and 40,3% (A > 2 000 m2)
The saving for the combined COCT region
increases with stand size category and w ith
PV category, as shown in table 5.
In both instances the percentage saving
achieved rises steeply in the lower two catego-
ries to an apparent asymptotic value of about
40 % in the large categories. This percentage
saving in SUWD is in broad agreement with
savings reported elsewhere. A 46 % saving
was achieved during drought conditions in
California (Loaiciga 1997) and a saving of
up to 50 % is reported by Hunt et al (1998)
for gardens that are extensively xeriscaped.
Extensive xeriscaping of gardens in Cape
Town might hold promise for additional sav-
ings in the larger stand size category (and PV
category) where garden irrigation is common.
In South Africa a saving of 36,8 % was
achieved with a retrofit project in Sebokeng,
while a saving of 18,5 % was reported for
946 flats in central Johannesburg after a
plumbing repair project (McKenzie et al
2002). The latter example does not include
properties with gardens, thus restricting the
potential savings achieved in that case to a
lower value.
Nature of residential water savings
The Level 2 water restrictions target mainly
outdoor residential water use. The impact is
clear from this research, where a significant
reduction in SUWD is noted in the large
stand size and large property value catego-
ries where outdoor use for garden irrigation
is common.
However, it is important not to underes-
timate the significant contribution to water
savings by the numerous residences in the
small stand size and low value categories,
despite a relatively small percentage saving
by customers in this group. The achieved
saving of about 10 % in SUWD for custom-
ers in the small stand size and stand value
categories, where indoor demand is consid-
ered to be dominant, suggests that a saving
in indoor demand was also achieved with
water restrictions in this group specifically
(indoor demand savings could also have
been achieved for the other categories, but
this is clouded by the addition of notable
outdoor use volumes).
Work by Van Zyl et al (2003) shows that
the short-term price elasticity for indoor
demand in townships is –0,30. To achieve
the observed saving of 7,8 % in this category
in COCT, a price increase of 26 % would
have had to be observed by a customer in
this volume bracket – based purely on the-
ory and the published price elasticity value.
This is of the right order of magnitude:
the actual water account for a 15 kl/month
use at the time of implementing the Level
2 restrictions would have increased from
R25,80 to R32,37 – an increase of 25 %. In
the same manner the actual price increase
faced by a 10 kl/month and 20 kl/month
user under Level 2 restrictions would have
been 8 % and 33 % respectively. The saving
brought about by these low-use customers
could thus be ascribed mainly to the price
increase brought about by the restrictions
implemented.
Comparison of results to the total
volume of water supplied
The monthly total volume of water supplied
to the COCT from all water sources com-
bined is presented in table 6. The monthly
SUWD (
l
/stand.d)
3 000
2 500
2 000
1 500
1 000
500
0
2 0001 9001 8001 7001 6001 5001 4001 3001 2001 1001 000900800700600500400300200100
0
AADD, Upper (2004)
COCT, SUW D – Period 1 COCT, SUW D – Period 2 AADD, Van Zyl et al (2007)
AADD, Lower (2004) AADD, Jacobs et al (2004)
Stand size (m
2
)
Figure 10 Residential SUWD versus stand size based guidelines for AADD
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 07 25
record presented in the table covers the
complete study period. The six-monthly
summer average for the total supply vol-
ume to the COCT reduced by 15,8 % from
Period 1 to Period 2. This reduction in the
total volume supplied is slightly more than
the 14,0 % reported earlier (refer to table 2)
for the total reduction from Period 1 to
Period 2 during analysis of the individual
users’ water meter readings as per the SAP
records.
Additional research
The findings from this study present sub-
stantial results in a field of critical impor-
tance to water managers and engineers.
Water restrictions and water saving should
receive renewed attention, with particular
focus on specific results based on measured
data. This study was limited in a few ways,
as discussed before, most of which could be
addressed by means of future research shed-
ding light on the following topics:
Extension of the work beyond the limita-
tion of metered, billed users included in
the SAP system to also include all water
use (and losses)
Knowledge of the water source at each
residence, including the use of boreholes,
rain water and on-site grey water reuse.
These alternative methods of supply are
often intended to meet garden watering
demand
Investigation into the scope of illegal
connections, including temporary con-
nections, prov ided by consumers during
restrictions. Such connections would
create the false impression of savings
and have the additional negative effect of
increased non-revenue water volumes
Evaluation of plumbing leaks on private
properties, particularly in the small stand
size- and low value categories
Detail financial analysis regarding the
impact of tariff structures and pricing
during restrictions
Comprehensive statistical analysis of the
available data to extrapolate the savings
achieved in the COCT to other regions
Investigation into the permanent versus
temporar y nature of the savings
Investigation into water use and saving as
a function of PV value for properties with
relatively small areas
CONCLUSION
A substantial reduction in demand was
achieved when comparing the summer
period prior to restrictions (1 October 2003
to 1 April 2004) to the summer period
after the implementation of water restric-
tions (1 October 2004 to 1 April 2005).
The overall reduction in SUWD achieved in
the COCT is 14 % and is substantial in all
categories, with a clear trend of increased
percentage saving with increased stand size
and also with increased PV.
The larger stands were found to achieve
a much larger unit water demand reduction
than the smaller stands (the reduction per
stand is more, but there are relatively few
users within categories defining low-density
and high-income areas). The maximum per-
centage reduction in SUWD from Period 1
to Period 2 is achieved in the second largest
stand category and second largest value cat-
egory. In both cases the saving is about 40 %.
For the high-density areas the SUW D
reduction is only 8 %, but because of the
much larger number of users in the smaller
categories, this results in a larger total vol-
ume saving for the high-density and low-
income areas.
ACKNOWLEDGEMENTS
The authors would like to t hank:
The City of C ape Town and a ll sta ff who were
involved with the project
Jaco de Bruyn for valuable input s to thi s study and
John Frame who was involved during the inception
phase
Commun ity Eng ineer ing Services and A fricon
Consu lting Engineers who were r espon sible for con-
ducting t he COCT consumpt ion project
Ninha m Shand C onsulting Engineers and DWAF for
inputs regarding the Berg River system and nature of
the water restrictions
Finally, the authors would like to commend all the indi-
vidual water users in the COCT who contributed to the
significa nt sav ings achieved.
REFERENCES
Arbués, F, Barberán, R and Vil lanúa, I 2004. Pr ice impact
on residential water dem and: a dy namic panel data
approach. Water Resources Research, 40(11).
COCT (City of C ape Town) 2005. City of Cape Town water
consumption study – progres s report 1, a report by
Commun ity Eng ineer ing Services and A fricon to the
COCT.
Collins 2004. Student’s dictionary. HarperCollins.
CSIR (Counc il for Scie ntific and Industrial Research) 20 00.
Guidelines for human settle ment planning and design. A
report compiled under the patronage of the Depar tment
of Housing by the CSI R Building and Constr uction
Tec hn olog y.
DWAF (Depar tment of Water Affa irs) 2005. Western Cape
Water Supply System monthly monitor ing repor t. A
monthly r eport compi led by Ninham Sha nd Consulting
Engineers for DWAF and the WCWSCF.
Espey, M, Es pey, J and Shaw, W D 1997. Price elasticity
of residential demand for water: a meta-analysis. Wa ter
Resources Research, 33(6):1369–1374.
Gebhardt, D S 1975. The effects of pressure on domest ic
water supply including observations on the effect of lim-
ited garden-watering restrictions during a period of high
demand. Municipal Engineer, 6(6):41–47.
Haarhoff, J, Geustyn, L C and Van Zyl, J E 2002.
Compilation of a residential water us e model for Rand
Water with elast icity for pr ice, erf area, inc ome and
pressure, pilot st udy. A report compi led by the R AU and
GLS Civil Consult ing Eng ineer s for Rand Water.
Hunt, J B, McD evitt, W and Hunt, G 199 8. Water efficiency
manual for commercial, industrial and institutional facili-
ties. A joint publication of the Div ision of Pollution
Prevention and Environ menta l Assista nce and Division
of Water Resources of the North C arolina Depar tment of
Table 6 Total water supply to COCT during study period
Reading date Month Period Total COCT water
supply (kl/month) Sub-total (kl)
1 November 2003 October
Period 1
30 677 703
179 888 609
1 December 2003 November 30 563 598
1 Januar y 2004 December 32 470 015
1 Februar y 2004 January 31 749 010
1 March 2004 February 30 609 480
1 April 2004 March 23 818 803
1 May 2004 Apr il
24 797 357
1 June 2004 May 21 908 381
1 July 2004 June 21 825 276
1 August 20 04 July 21 754 961
1 September 2004 August 23 609 633
1 October 2004 September 23 742 054
1 November 2004 October
Period 2
26 069 970
151 424 631
1 December 2004 November 27 722 759
1 Januar y 2005 December 26 677 153
1 Februar y 2005 January 23 904 865
1 March 2005 February 25 282 823
1 April 2005 March 21 767 061
Average Period 1 29 981 435 kl/month (summer)
Average Period 2 25 237 439 kl/month (summer)
Reduction (Period 1 to Period 2) 4 743 996 kl/month (summer)
Percentage reduction 15,8 %
Journ al of the S outh Af rican I nstit ution o f Civil E nginee ring • Volume 49 Numbe r 3 Septem ber 20 0726
Environment and Natural Resources, and the Land-of-
Sky Reg ional C ouncil – W RATT Prog ram.
Jacobs, H E and Haarhof f, J 2004. Appl ication of a resi-
dential end-us e model for estimating cold- a nd hot
water dem and, wastewater flow and sali nity. Wat er S A,
30(3):305–316.
Jacobs, H E, G eustyn, L C, L oubser, B F and Van der
Merwe, B 20 04. Est imating residential water dem and in
southern Africa. Journal of the South African Institution of
Civil Engineering, 46 (4):2–13.
Kulik, V 1993. Simple weather-water demand model.
International Journal of Water Resources De velopment,
9(3):293–304.
Loaiciga, H A and Renehan, S 1997. Municipal water use
and water r ates dr iven by severe drought: a case study.
Journal of the Amer ican Water Resources Asso ciation,
33(6):1313 –1326.
McKenzie, R S, Wegeli n, W A and Meyer, N 2002. Leakage
reduction projects undertaken by Rand Water. A
report compiled by WRP Water Resource Planning and
Conservation for Rand Water W DM Pilot Project s.
Van Zyl, H J, Van Zyl, J E, Geusty n, L C, Illemobade, A a nd
Buckle, S J 20 07. Benchmarking of water consumption
in selected South African cit ies. WRC Report K5/1525.
Van Zyl, J E and Geusty n, L C 2007. Development of
a nation al water consumpt ion archive. WRC Report
1605 /1/ 07.
Van Zyl, J E, Haarhoff, J and Hus selm an, M L 2003.
Potenti al application of end-use demand modelling in
South Africa. Journal of the South African Institution of
Civil Engineering, 45(2):9–19.
LIST OF ABBREVIATIONS AND ACRONYMS
A Property area or stand siz e (m
2
)
AADD average annua l daily w ater dema nd (kl/d)
AFU Automat ic flu shing u rin al
CMA Catchment management agency
COCT City of Cape Town
d day
DWAF Depart ment of Water A ffairs and Forestry
GIS Geographic information system
ICI Industrial, commerc ial and i nstitutiona l (land
use category)
k
l
kilolitres (thousand litres)
k
l
/stand·d volume of use (k
l
) per sta nd, or property, per
day
l
/stand·d volume of use (
l
) per sta nd, or property, per
day
PV Property value (according to municipal valua-
tions)
RDP Re construction a nd development programme
SAP A treasury system used by the COCT
SADD Summer average da ily water demand ( kl/d),
1 October to 31 Marc h
SUWD Summer unit water demand (kl/stand·d),
1 October to 31 Marc h
WCWSCF Western Cap e Water System C onsultative
Forum
WDM Water demand management
... When attempting to predict human intentions and behaviour, there exists probabilistic factors, externally and internally, which lead to stochastic results. The various dispositions of households due to geographic, economic, social, political and cultural factors is difficult to scale, and their tendency to conserve water based on these factors requires a dynamic approach [24]. ...
... Du Plessis has many years of experience in urban water systems, including serving as director in the water division of the West Coast District Municipality, performing project work with the Department of Water Affairs and Sanitation, as well as consulting in water-related work with the Cape Town government. He co-authored the article entitled Analysis of water savings: A case study during the 2004/05 water restrictions in Cape Town [24], which helped to verify and validate the developed model. [30], which provided great insight into the awarenessintention gap and the intention-action gap, as well as the factors affecting pro-environmental tendencies. ...
... A technical paper, written by Jacobs, Du Plessis et al. [24], investigates the reduction in water demand during the 2004/2005 drought in the entire city of Cape Town. The water demand management policy implemented by the Cape Town government included level 2 restrictions, increased tariffs, and awareness campaigns. ...
Thesis
During the last century, global population tripled and water consumption increased sixfold. Water scarcity is indeed a global and local phenomenon. In order to address this and the problems that result from it, investigations into water demand management strategies and the conservation tendencies of the population are imperative. Water demand management pertains to controlling end-use behaviours and encouraging efficient water use through the implementation of various strategies. Existing analysis and prediction models in the field of urban water demand are considered, although these studies di�er in methodology and focus. This project investigate the tendency of a residential population within the city of Cape Town to conserve water in response to certain demand management strategies. Accompanying this study is an agent-based simulation model which was developed within the AnyLogic software environment. It may be used as an investigative tool to explore the effect of different water demand management strategies on the tendencies of households, within different wards of Cape Town, to conserve water. The model allows for the population within a ward of interest to be defined by certain attributes, such as income level and access to information. The user may prompt certain demand management strategies during the simulation run and explore the change in environmental awareness and conservation tendency of the user-defined population visually. The model has the ability to portray a given scenario with reasonable accuracy, when equipped with valid input parameter values. In order to ensure the reasonableness and reliability of the agent-based model, it was subject to a number of verification and validation techniques. Insight into the real-world system was provided by subject matter experts, whom were consulted and concluded that the model was indeed sound. Furthermore, a sensitivity analysis was performed to instill �confidence in the model's robustness. A notable outcome of the model was the overwhelming influence of a household's inherent attitudes and values regarding water conservation when responding to demand management strategies.
... When attempting to predict human intentions and behaviour, both external and internal probabilistic factors exist that lead to stochastic results. The various dispositions of households owing to geographic, economic, social, political, and cultural factors are difficult to scale, and their tendency to conserve water based on these factors requires a dynamic approach [5]. ...
... A household is typically made up of homogeneous individuals with similar attributes who display similar behavioural patterns, such as water use and conservation tendencies. A person's characteristics dictate their behaviour, which in turn influences their water demand [5]. Understanding consumer behaviour is considered an effective starting point for managing energy conservation and, by extension, water conservation. ...
Article
Full-text available
Water scarcity is both a global and a local phenomenon. In order to address this, investigations into water demand management strategies and the conservation tendencies of a population are imperative. This paper investigates the tendencies of a residential population in the City of Cape Town to conserve water in response to certain demand management strategies, using an agent-based simulation approach. The model allows for the population in a ward of interest to be defined by certain attributes, such as income level and access to information. The user may trigger certain demand management strategies during the simulation run, and visually explore the changes in environmental awareness and conservation tendency of the user-defined population. The model has the ability to be used as an investigative tool to explore the effect of different water demand management strategies on the tendencies of households in different wards of Cape Town to conserve water.
... In addition, the same factors increase domestic outdoor water use (garden irrigation and the topping up of swimming pools). Prior to the introduction of water restrictions from 2016, domestic outdoor consumption was a sizable component of Cape Town's water demand, especially for large high-income properties (Jacobs et al. 2007). The influence of outdoor consumption is evident in the seasonality of Cape Town's water demand before the introduction of water restrictions (see 2014 water demand shown in Figure 6.4) and which closely follows annual weather patterns as shown in Figure 6.5. ...
... In addition, the same factors increase domestic outdoor water use (garden irrigation and the topping up of swimming pools). Prior to the introduction of water restrictions from 2016, domestic outdoor consumption was a sizable component of Cape Town's water demand, especially for large high-income properties (Jacobs et al. 2007). The influence of outdoor consumption is evident in the seasonality of Cape Town's water demand before the introduction of water restrictions (see 2014 water demand shown in Figure 6.4) and which closely follows annual weather patterns as shown in Figure 6.5. ...
... The combined effects of expected temperatures increases, higher evapotranspiration rates and an expected overall reduction in effective rainfall are likely to lead to increases in water consumption in the region, especially agricultural irrigation, which accounts for 34% of water use in the WCWSS (DWS, 2014) and domestic outdoor consumption (garden irrigation and the topping up of swimming pools) . Jacobs et al. (2007) show that domestic outdoor consumption is a significant component of Cape Town water demand, especially for large high-income properties. The influence of outdoor consumption on total Cape Town water demand can been seen in the seasonality of Cape Town's water demand (see Figure 4) which closely follows annual weather patterns, as shown in Figure 5. Increased agricultural irrigation is likely to lead to increased competition for regional raw water resources. ...
Preprint
Full-text available
This chapter describes how the growth in water demand has been successfully managed in Cape Town since the introduction of a comprehensive programme of water demand management (WDM) from around 2000.
... The combined effects of expected temperatures increases, higher evapotranspiration rates and an expected overall reduction in effective rainfall are likely to lead to increases in water consumption in the region, especially agricultural irrigation, which accounts for 34% of water use in the WCWSS (DWS, 2014) and domestic outdoor consumption (garden irrigation and the topping up of swimming pools) . Jacobs et al. (2007) show that domestic outdoor consumption is a significant component of Cape Town water demand, especially for large high-income properties. The influence of outdoor consumption on total Cape Town water demand can been seen in the seasonality of Cape Town's water demand (see Figure 4) which closely follows annual weather patterns, as shown in Figure 5. Increased agricultural irrigation is likely to lead to increased competition for regional raw water resources. ...
Preprint
Full-text available
This postscript provides an update on the City of Cape Town Water Demand Management programme in the context of the severe drought conditions starting 2015 and covers the role it played in mitigating the impacts of the drought.
... The links between a HWS, intermittent supply, relatively low system pressure and associated health risks need to be modelled and better understood. Potable water use at home typically reduces during stringent water restrictions (Jacobs et al., 2007) and also potentially under intermittent supply conditions, but community health risks increase due to reduced source water quality and reduced frequency of washing, as well as water sharing among family members (Fan et al., 2014). ...
Article
Full-text available
This paper addresses on-site supplementary household water sources with a focus on groundwater abstraction, rainwater harvesting and greywater reuse as available non-potable water sources to residential consumers. An end-use model is presented and used to assess the theoretical impact of household water sources on potable water demand in formal residential areas. Reliable potable municipal supply to urban consumers via the water distribution system is typically linked to relatively low uptake of household water sources. However, stringent water restrictions in some large South African cities that prohibit outdoor use, and reports of intermittent water supply, have led to increased uptake of household sources in South Africa. This paper describes the legal position regarding such sources in South Africa, and describes an end-use model to assess the theoretical impact on water demand in formal residential areas. The model provides valuable strategic direction and indicates a significant theoretical reduction in potable municipal water demand of between 55% and 69% for relatively large properties when household sources are maximally utilised (when compared to exclusive unrestricted municipal use as a baseline). This load reduction on piped reticulation systems could be an advantage in order to augment municipal supply, but water service planning and demand management are complicated by the introduction, and possible future decommissioning, of any household water source. The extent of both positive and negative impacts of household water sources requires further research. © 2017, South African Water Research Commission. All rights reserved.
Article
Full-text available
This research investigated relationships between the most notable characteristics of end-use events, namely, event duration, volume, and intensity, in order to categorize water use as being indoor or outdoor. Three classification models were developed, calibrated, and compared using more than 200,000 household end-use events that were recorded independently in Australia and South Africa. The three methods were also compared to a practice-based limit classification scheme. The classification model presented in this paper correctly apportions ∼81% of the indoor end-use event volumes and ∼98% of the outdoor end-use event volumes, thus reinforcing the value of basic smart water meter data sets as a source of useful information for water demand management.
Article
Full-text available
As urban droughts make headlines across the globe, it is increasingly relevant to critically evaluate the long‐term sustainability of both water supply and demand in the world's cities. This is the case even in water‐rich regions, where upward swings in water demands during periods of hot, dry weather can aggravate already strained water supplies and increase cities' vulnerability to water shortage. Summer spikes in water demand have motivated several cities to impose permanent restrictions on outdoor water uses; however, little is yet known about their effectiveness. This paper examines daily water production data from 15 Canadian cities to (1) quantify how overall and seasonal demands are evolving over time across humid and semiarid settings and (2) determine whether permanent water use restrictions have been effective in curbing summer water demands both seasonally and during specific hot and dry periods. Results show that while per‐capita water demand is declining in all cities studied, the seasonal distribution of that demand has remained largely stable in all but a few cases. While average demands in the summer months remain largely unaffected by the imposition of permanent restrictions, cities that enforce stringent limits on outdoor water use have seen a reduction in the variability of daily demands and a decline in peak demands following their implementation. During short‐term periods of exceptionally hot and dry weather when vulnerability to water shortage is most acute, cities with strict restrictions also see smaller surges in demand than those with weaker or no restrictions in place.
Article
Full-text available
Garden irrigation is a significant and variable household water end‑use, while groundwater abstraction may be a notable supplementary water source available in some serviced residential areas. Residential groundwater is abstracted by means of garden boreholes or well points and – in the study area – abstracted groundwater is typically used for garden irrigation. The volume irrigated per event is a function of event duration, frequency of application and flow rate, which in turn are dependent on numerous factors that vary by source – including water availability, pressure and price. The temperature variation of groundwater abstraction pipes at residential properties was recorded and analysed as part of this study in order to estimate values for three model inputs, namely, pumping event duration, irrigation frequency, and flow rate. This research incorporates a basic end‑use model for garden irrigation, with inputs derived from the case study in Cape Town, South Africa. The model was subsequently used to stochastically evaluate garden irrigation. Over an 11-d period, 68 garden irrigation events were identified in the sample group of 10 residential properties. The average garden irrigation event duration was 2 h 16 min and the average daily garden irrigation event volume was 1.39 m3.
Article
Full-text available
The structure and data requirements of an end-use model for residential water demand and return flow are presented in a companion paper. This paper focuses on the practical application of the model. The model is first applied to confirm a few commonly observed water demand patterns: Seasonal variation in demand, the positive correlation between average annual daily water demand and stand size, and the increase in water demand, hot water demand and wastewater flow with an increase in household size. The convergence between the predicted model results and independently observed values by others encourages practical use of the model. Secondly, the effects of some specific water demand management measures are evaluated by adjusting selected model parameters. The measures include xeriscaping, the installation of dual-flush toilets and low-flow showerheads, pool ownership and pool cover use. The model provides a rapid means to obtain first estimates of the likely effects of different water demand management measures.
Article
Full-text available
End-use water demand modelling is used to generate water demand projections by modelling various end uses, for example showers, toilets and washing machines. End-use models can be used to estimate water demand changes due to various scenarios, such as price increases, housing densification and conservation programmes. This study reports on the potential application of end-use modelling in South Africa, based on a pilot study that was done for Rand Water. The model includes elasticities of water demand with respect to variations in water price, household income, stand size and pressure. The study highlights many of the difficulties and limitations, as well as the potential applications of end-use modelling as a water deman predictor. A special effort is made to explain the meaning and application of elasticity in end-use modelling. Various data sources were used to determine elasticities for the variables used, and to identify minimum and maximum elasticity values. The implications of the elasticities are illustrated using a sensitivity analysis and case study.
Article
Full-text available
Meta-analysis is used to determine if there are factors that systematically affect price elasticity estimates in studies of residential water demand in the United States. An econometric model is estimated, using price elasticity estimates from previous studies as the dependent variable. Explanatory variables include functional form, cross-sectional versus time series, water price specification, rate structure, location, season, and estimation technique. Inclusion of income, rainfall, and evapotranspiration are all found to influence the estimate of the price elasticity. Population density, household size, and temperature do not significantly influence the estimate of the price elasticity. Pricing structure and season are also found to significantly influence the estimate of the price elasticity.
Article
It is understandable that an easy method to obtain estimates of residential water demand is often used. These estimates are also extended to calculate peak demand and sewer flow, and impact an authority's water and sewer infrastructure budget and finally its expenditure. Guideline curves are presented in this paper that can be used to estimate annual average residential water demand based on stand size. The measured water consumption and stand size of more than 600 000 single residential stands were obtained. Treasury databases for Cape Town, Ekurhuleni, George, Midrand, Randfontein and Tshwane were analysed in detail and the results compared to similar work in Windhoek. The large number of records made it possible to conduct statistical analyses and to investigate the distribution of data for stand size intervals of 100 m 2. The water demand of similar sized stands in townships and suburbs could be compared. A strong relationship exists between the average annual water demand and stand size. The authors note that a model based on stand size has limited application only when better methods are not available.
Article
To employ plain simulation instead of complicated correlation consumption models one has to utilize modern database facilities. Reciprocally, theoretical rules have to be built into the database as a criterion for the data import into this database and for debugging of previously imported data. Practical use of the model is very forthright; no special program is required. Its applications are: analysis of pricing policies impact, water demand planning, simulation of water demand for new cities and suburbs, etc.
Article
In this paper, we formulate and estimate a model of residential water demand with the aim of evaluating the potential of pricing policies as a mechanism for managing residential water. The proposed econometric model offers a new perspective on urban water demand analysis by combining microlevel data with a dynamic panel data estimation procedure. The empirical application suggests that residential users are more responsive to a lagged average price specification. Another result of the estimated model is that price is a moderately effective tool in reducing residential water demand within the present range of prices, with the estimated values for income elasticity and ``elasticity of consumption with respect to family size'' reinforcing this conclusion.
Article
This paper synthesizes and interprets data pertaining to the evolution of average water revenue, water use, and the average cost of water supply in the City of Santa Barbara, California, from 1986 to 1996, a period which included one of the most devastating droughts in California this century. The 1987–1992 drought hit the study area particularly hard. The City of Santa Barbara was dependent exclusively on local sources for its water supply. That made it vulnerable as the regional climate is prone to extreme variability and recurrent droughts. The 1986–1992 drought provided a rare opportunity to assess the sensitivity of municipal water use to pricing, conservation, and other water management measures under extreme drought conditions. Our analysis indicates that the average cost of water rose more than three-fold in real terms from 1986 to 1996, while the gap between the average cost of supply and the average revenue per unit of water (= 100 cubic feet) rose in real terms from $0.14 in 1986 to $ 0.75 in 1996. The rise of $3.08 in the average cost of supplying one unit of water between 1986 and 1996 measures the cost of hedging drought risk in the study area. Water use dropped 46 percent at the height of the drought relative to pro-drought water use, and remains at 61 percent of the pre-drought level. The data derived from the 1987–1992 California drought are unique and valuable insofar as shedding light on drought/water demand adaptive interactions. The experience garnered on drought management during that unique period points to the possibilities available for future water management in the Arid West where dwindling water supplies and burgeoning populations are facts that we must deal with.
The effects of pressure on domestic water supply including observations on the effect of limited garden-watering restrictions during a period of high demand
  • D Gebhardt
Gebhardt, D S 1975. The effects of pressure on domestic water supply including observations on the effect of limited garden-watering restrictions during a period of high demand. Municipal Engineer, 6(6):41-47.
Development of a national water consumption archive
  • J Van Zyl
  • L Geustyn
Van Zyl, J E and Geustyn, L C 2007. Development of a national water consumption archive. WRC Report 1605/1/07.