Content uploaded by Peter J. Coombes
Author content
All content in this area was uploaded by Peter J. Coombes on Oct 23, 2014
Content may be subject to copyright.
Insights into Household Water Use Behaviours Throughout
South East Queensland During Drought
Peter Coombes
Independent Scientist, Urban Water Cycle Solutions, Newcastle, Australia
E-mail: thecoombes@bigpond.com
This paper describes the water use behaviour of households throughout South East Queensland that
was monitored using mini smart meters during the period February 2009 to May 2010. The monitoring
program was combined with audits to define the characteristics of each household and each
household also completed a water use diary. Households with higher incomes were more likely to
include water efficient appliances and rainwater tanks which explained the small increases in per-
capita and household water demands associated with households with higher incomes. Per-capita
water use was found to be strongly dependent on the reported frequency of showering. Households
without rainwater tanks had an average daily mains water demand of 514 litres and a per-capita daily
demand of 139 litres. Households using rainwater for outdoor use had an average daily mains water
demand of 383 litres and a per-capita daily demand of 117 litres – annual savings of 47.8 kL.
Households using rainwater for indoor and outdoor uses had an average daily mains water use of 268
litres and a per-capita demand of 63 litres – annual savings of 89.8 kL. The performance of
households using rainwater for indoor and outdoor uses exceeded the requirements of Queensland
Development Code MP 4.2.
1. INTRODUCTION
This independent study was originally commissioned by the Queensland Water Commission (QWC) in
response to a request from the Auditor General and carried out by the author and colleagues (formerly
Bonacci Water). It is a snapshot of the longitudinal impacts of water efficient appliances and rainwater
tanks on residential water use during and after the recent drought throughout South East Queensland
(SEQ). Metering results were combined with responses to questionnaires and water use diaries to
develop an understanding of water use behaviours throughout the SEQ region. Approximately 50
households in the SEQ region were randomly selected for the survey. Mini smart meters were installed
at each property and participants were asked to complete questionnaires about their households.
Each household also completed a water use diary over a period of 30 days that outlined key water
uses on each day. This paper provides an overview of the results of the unique monitoring project that
has directly observed water use at a range of households.
2. METHODS
Approximately 325 households were invited to participate in this study (via email, telephone and letter
drops). It was expected that a higher number of people received this invitation as those contacted
directly were encouraged to forward on the details and recruit colleagues, family and friends. A total of
76 households from all areas throughout SEQ responded as willing to participate in the study.
However, 31 of the households were excluded from the investigation due to water meters on their
properties that were not compatible with the smart meter technology and 3 households did not
ultimately complete the monitoring program. Questionnaires and water use diaries were issued to
participating households and water use was observed at 6 minute intervals at each household during
the period February 2009 to May 2010. The main objectives of the end use study were:
1. Collect data on residential water usage;
2. Identify water use patterns and trends;
3. Reveal the use of water saving measures and behaviours; and
4. Understand the impacts of variations in climate on water use behaviours
Careful planning was required to ensure a selection of appropriate households that also accounted for
the logistics of collecting and processing the data. Our study equally targeted four regions throughout
the SEQ area, Brisbane, Gold Coast, Toowoomba and Sunshine Coast, to ensure a mix of residential
properties across the regions. This process increased the logistical difficulty of installing smart meters
and subsequent data collection. Nevertheless, the study aimed to include a wide variety of locations
PREPRESS PROOF FILE CAUSAL PRODUCTIONS1
and demographics. Detailed planning of the water use study began in late 2008. A brief summary of
this process is provided as follows.
Contact was made with the residents of residential properties in the 4 study areas (Brisbane, Gold
Coast, Toowoomba and Sunshine Coast). Several different approaches were utilised to ensure
participation of a random and varied group of residential properties. Participants were randomly
selected from the White Pages telephone directory and were invited to participate via telephone, by
mail drops in selected areas, by emails to business and community networks, and via word of mouth
(participants in the study forwarding the study details to friends, family and colleagues). An initial
invitation letter was sent to all participants explaining what the survey was about and what would be
required of the participants.
Collect and compile contact details of residential properties willing to participate in the study. Create
detailed questionnaires and water use diaries that were distributed to participating residential
properties. This was a crucial part of the study that details indoor and outdoor appliances and water
use behaviours critical to the overall analysis.
Questionnaires and water use diaries were sent to participants. Participants were also advised that the
installation of the mini smart meters would commence in early February 2009. Self-addressed and
stamped envelopes were also included to ensure a timely response by participants. A schedule to
install meters at all participating properties was undertaken during early February 2009. The mini
smart meters were installed during a three week period. When residents were onsite during the
installation process, researchers explained the operation of mini smart meters and provided contact
details for any queries or issues. When residents were not onsite during the installation process,
researchers left introductory packages to explain the study and the operation of mini smart meters.
Contact details were provided for any queries or to resolve issues.
At some locations mini smart meters could not be installed due to incompatible water meters. Letters
were sent to households advising that their water meter was unsuitable and that their property could
not be included in the study. Researchers visited residential properties during the period February
2009 to May 2010 to download data from the mini smart meters. During the period February 2009 to
May 2010, all data were collected and analysed from mini smart meters and from the returned
questionnaires and water use diaries. Statistical analysis of data from each site was used to determine
the major variables affecting residential water use. During the period March to May 2009 residential
properties were contacted via telephone and email reminding them to complete the questionnaires
and water use diaries.
Mini smart meters and data loggers were fitted to water meters at all households to continuously
record water use. Water use data was stored on the data logger and downloaded throughout the
project. The downloaded data was then validated for timing, total water use and quality of data by
manual observations of water use from household water meters.
3. RESULTS AND DISCUSSION
The responses to the questionnaires revealed a diversity of household water use characteristics and
water related infrastructure. The only characteristics found to be common to all of the surveyed
households was that none of the households were connected to a centralised recycled water system
and an absence of single flush toilets. Only one household had a single flush toilet installed in addition
to other dual flush toilets on the property. Two of the households in the study were found to have
significant mains water leaks during the monitoring period. One of households was found to consume
approximately 2,100 litres/day and the other approximately 1,700 litres/day of mains water. These
leaks represent a significant volume of water in comparison to average household water use. Daily
water use in these properties was more than 5 times larger than the average household water use of
the study. The average per-capita water use for households in the study was 98 L/day and the
average household size was 3.75 persons. The relationship between average daily water use in
households and the size of households is presented in Figure 1.
Figure 1 reveals a non-linear relationship between average daily household water use and household
size. These results indicate that per-capita water-use diminishes with increases in household size
which is consistent with current research (such as Cui et al., 2008). The distribution of household sizes
in this study is shown in Figure 2 and compared to the distributions of household sizes in the major
areas of the region sourced from the Australian Bureau of Statistics (ABS, 2007) shown in Figure 3.
2
Figure 1: Household water-use versus household size.
Figure 2: Distribution of households in the
study
Figure 3: Distribution of households in SEQ
Figure 2 shows that the distribution of household sizes in the study was dominated by households with
2, 3 and 4 residents. Comparison with the distributions of household sizes in SEQ shown in Figure 3
reveals that the cohort of households in the study has significantly fewer one person households and a
greater proportion of 4, 5 and 6 person households. This difference is explained by the study’s focus
on detached housing due to metering constraints that has resulted in a limited presence of units and
apartments in the cohort that are expected to be mostly 1 and 2 person households whilst detached
housing will consist of a greater distribution of household sizes. The return of questionnaires and
water use diaries allowed detailed verification of the continuous water use data from the mini smart
meters to determine the characteristics of the end uses of water in households (water use for showers,
toilet flushes, washing machines and so on). Data from the mini smart meters was calibrated to
individual flow events at various household fixtures as described by flow rates, timing and magnitude
of water use events. The description of appliances and water use behaviour from questionnaires was
vital to this process. A summary of household characteristics is shown in Table 1.
Table 1: Summary of household characteristics
Income
($1000
/hh/yr)
Houses Household
size
(people)
Per-capita
use
(L/day)
Household water
use (L/day)
Rainwater
tanks
Water
efficient
fixtures
Ave Min Max
0 - 30 3 3 109 328 173 637 2 2
30 - 60 5 2 94 187 8 959 3 2
60 - 90 5 5.6 86 481 293 565 4 2
90 - 120 6 3.4 116 394 225 801 2 3
120 + 13 4.9 113 441 99 906 10 10
Table 1 shows that larger household sizes and daily water use were associated with higher incomes
which resulted in a trend to larger per-capita water use for households with higher incomes. The use
of rainwater tanks was fairly uniform across all income groups and a greater adoption of water efficient
appliances was apparent for higher income households. The greater proportion of rainwater tanks and
water efficient appliances in households has assisted with reducing household and per-capita water
use in those households. Clearly a higher income may be related to a greater capacity to invest in
water saving measures. It is noted that the lowest and highest income groups appear to be under-
12345+
Household size (people)
0
200
400
600
800
1000
Water use (L/hh/day)
Median
25%-75%
Non-Outlier Range
12345+
Household size (people)
0
100
200
300
400
500
600
700
Water use (L/pp/day)
Median
25%-75 %
Non-Outlier Range
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
123456
Household size (people)
Proportion of households
Toowoomba
Sunshine Coast
Brisbane
Gold Coast
3
represented and over-represented, respectively, in the study. A significant proportion of households
reported that they supplemented mains water supplies with roof collected rainwater stored in a tank.
Approximately 77% of households that responded to the questionnaires and about 53% of the entire
cohort used rainwater from tanks for household water supply. The proportion of houses with rainwater
tanks is greater than the 32% reported by the QWC. The average results for the houses without
rainwater tanks, using rainwater for indoor and outdoor water uses, and for outdoor water uses only
are presented in Figure 4.
Figure 4: Summary of monitoring results for SEQ
Figure 4 demonstrates that households utilising rainwater for indoor and outdoor water uses, and for
outdoor uses only used significantly less mains water than houses without rainwater tanks. The use of
rainwater tanks has also reduced the seasonal variability of demands for mains water. This effect was
particularly evident for households using rainwater for indoor and outdoor water uses. In addition,
Figure 4 shows that the effectiveness of rainwater tanks for reducing mains water use was increased
during the last few months of the monitoring period when rainfall returned to more normal patterns.
Note that the monitoring period included high level water restrictions until April 2009, medium water
restrictions to December 2009 and then permanent water conservation measures thereafter. A
summary of the monitoring results for rainwater tanks are provided in Table 2.
Table 2: Summary of monitoring results from SEQ for rainwater tanks
Household
rainwater use
Number of
houses
Average monthly water use
(L/day) Reduction
(%)
Reduction
(kL/annum) Per-capita
use
(L/day)
Average Maximum Minimum
No Tanks 17 514 871 278 - - 139
Indoor 11 268 428 125 48 89.79 63
Outdoor 14 383 725 263 24 47.82 117
Table 2 highlights that 17 households were not connected to rainwater tanks whilst 11 households
used rainwater for indoor and outdoor uses and 14 households use rainwater for outdoor water uses.
The average reduction in demands for mains water for households utilising rainwater for indoor and
outdoor uses, and for outdoor uses was 89.8 and 47.8 kL/annum respectively. The performance of
households using rainwater for indoor and outdoor purposes was considerably in excess of the targets
in the Queensland Development Code (MP 4.2). The observed average per-capita water use from
households without rainwater tanks of 139 Litres/person/day is slightly higher than the per-capita
water use reported by the QWC for households during the drought of 131 Litres/person/day as
expected. It is significant that the use of rainwater tanks has reduced per-capita water use to 117 and
63 Litres/person/day for outdoor, or indoor and outdoor uses respectively. This result is less than the
per-capita water use of 200 Litres/person/day targeted by the SEQ water strategy. The impact of
household rainwater harvesting on seasonal demands for mains water is demonstrated in Table 3.
0
200
400
600
800
1,000
1,200
1,400
1,600
0
100
200
300
400
500
600
700
800
13579111315
Rainfall (mm/month)
Water use (L/day)
Months since January 2009
No RWT RWT: indoor RWT: garden Overall average Rain
4
Table 3: Impact of rainwater harvesting on seasonal mains water use throughout SEQ
Household
rainwater use
Number
of
houses
Average monthly water use (L/day) Reduction
(%)
Household
size
(people)
Per-capita
use (L/day)
Average Maximum Minimum
No tanks 17 514 871 278 - 3.7 139
Outdoor 14 383 725 263 24 3.3 117
Toilet & outdoor 4 296 425 192 42 3.5 84
Laundry & outdoor 2 225 271 130 56 5.5 41
Toilet, laundry &
outdoor 5 254 376 121 51 5.2 49
Tables 2 and 3 reveal that the use of rainwater tanks has reduced average, maximum and minimum
water uses throughout the monitoring period. This demonstrates the resilience of rainwater tanks for
reduction in mains water demands throughout SEQ. In addition, Table 3 shows that the use of
rainwater for laundry and outdoor uses, and toilet, laundry and outdoor uses produces the greatest
reduction in demands for mains water. The observations of the impacts of different capacities of
rainwater tanks in the survey are presented in Table 4.
Table 4: Mains water savings versus capacity of rainwater tanks
Tank size
(kL) Number
Average monthly water use
(L/day)
Average Maximum Minimum
No tank 17 514 871 278
0 - 2 6 248 341 100
2 - 5 8 376 663 198
5 + 11 303 462 153
Table 4 shows that households with capacities of rainwater tanks less than 2 kL generated the lowest
demands for mains water. Households with rainwater tanks larger than 5 kL provided the lower
demands for mains water than tanks with capacities between 2 and 5 kL, and higher mains water
demands than houses with rainwater tanks smaller than 2 kL. This result highlights that the
performance of rainwater harvesting is dependent on water use from tank and roof area rather than
the size of the rainwater tank. Observations of the impacts of different roof areas connected to
rainwater tanks in the survey are presented in Table 5.
Table 5: Impact of roof area connected to rainwater tanks
Connected roof
area (m2) Number
Average monthly water use
(L/day)
Average Maximum Minimum
No tank 17 514 871 278
< 50 3 388 513 281
50 - 100 6 304 439 79
100 - 150 5 338 492 192
150 - 200 5 292 574 114
>200 6 329 607 143
Table 5 shows that rainwater tanks connected to all roof areas produce significant reductions in
demands for mains water. However, rainwater tanks connected to roof areas in the range of 50 m2 to
100 m2 and 150 m2 to 200 m2 produced lowest demands for mains water. This outcome highlights
that the performance of rainwater harvesting is also primarily dependent on water demands from the
rainwater tank and that relatively small roof areas can produce significant savings in mains water.
Information about peak daily and hourly water use is used to design water distribution infrastructure.
The average per-capita peak daily and hourly water uses recording during the monitoring period is
provided in Table 6. Table 6 shows that the use of rainwater tanks to supply indoor and outdoor water
demands provides substantial reductions in peak hourly and daily water demands. This result
indicates that the use of rainwater tanks will reduce impacts on or requirement for local and regional
infrastructure including water distribution systems, pumping stations, water treatment plants and
5
pressure reservoirs. It is noteworthy that limiting the use of rainwater to outdoor uses will not produce
benefits for local distribution infrastructure.
Table 6: Per-capita peak water uses at households
Category Peak water use
(L/pp/minute)
Hour Day
No tanks 1.64 0.36
Tanks - garden 1.74 0.19
Tanks - indoor 1.07 0.17
The distribution of these results is demonstrated for peak hourly and daily mains water demands in
Figures 5 and 6 respectively.
Figure 5: Per-capita peak hourly use
Figure 6: Per-capita peak daily use
Figure 5 shows that the median values of per-capita peak hourly water use at households with and
without rainwater tanks are similar. However, use of rainwater tanks for only garden watering shows a
trend to increases peak hourly water use and the use of tanks for indoor uses show a significant trend
to decreases in peak hourly water use. Figure 6 reveals the use of rainwater tanks for indoor water
uses significantly reduces peak daily water demands. Use of rainwater tanks for indoor uses will
reduce impacts on water distribution, or requirement for, treatment and storage infrastructure. The
reported frequency of toilet flushing, showering, clothes washing and washing dishes is presented in
Table 7.
Table 7: Frequency of appliance use in households
Criteria
Frequency (events per week)
0 - 5 5 - 10 10 - 15 15 - 20 20+
Toilet (flushes/person) 1 6 7 4 8
Shower (Showers/person) 6 18 1 - -
Clothes Washer (Loads) 12 10 4 - -
Dish Washer (Loads) 10 8 5 1 3
Table 7 shows considerable variation in the frequency of appliance use in households. The frequency
of toilet flushing has the greatest range of responses which may represent a higher impact of
household behaviour on this end use – if residents are away from home at work or other activities the
frequency of flushing will be less whilst those at home with (say) children will generate a higher
frequency of toilet flushing. The frequencies of showering and clothes washing displayed significantly
less variance with the majority of households experiencing up to 10 showering events per person and
up to 10 clothes washing events each week. The washing of dishes showed a greater range of
frequencies which displays household use of dish washers rather than washing dishes by any method.
The frequency of shower use is compared to household size and household water use in Figures 7
and 8 respectively.
Figure 7 shows that the frequency of per-capita shower use decreases for household sizes greater
than two people. Figure 8 reveals that household water use increases with increased frequency of
shower use. The frequency of toilet flushing is compared to household size and household water use
in Figures 9 and 10 respectively.
No Tanks
Tanks - garden
Tanks - indoor
Category
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Peak hourly water
use (L/min)
Median
25%-75%
Non-Outlier Range
No Tank
Tank - garden
Tank - indoor
Category
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Peak daily water
use (L/minute)
Median
25%-75%
Non-Outlier Range
6
Figure 7: Frequency of shower use versus
household size
Figure 8: Household water use versus
frequency of shower use
Figure 9: Frequency of toilet flushing
versus household size
Figure 10: Household water use versus
frequency of toilet flushing
Figure 9 shows that the frequency of per-capita toilet flushing decreases with increases in household
size. Figure 3.10 reveals that household water use increases with more frequent toilet flushing up to
20 flushes per person per week. Household water use declines with increases in frequency of toilet
flushing greater than 20 flushes per person per week. The frequency of washing dishes is compared
to household size and household water use in Figures 11 and 12 respectively. Figure 11 shows that
the frequency of washing dishes increases with household size up to a three person households.
Figure 12 reveals that household water use is relatively independent of the frequency of washing
dishes. Household water use is greater from frequencies of dish washing greater than 15 loads.
Figure 11: Frequency of washing dishes
versus household size
Figure 12: Household water use versus
frequency of washing dishes
The frequency of washing dishes is compared to household size and household water use in Figures
13 and 14 respectively. Figure 13 shows that the frequency of washing clothes increases with
household size and that two person households have a higher frequency of clothes washing than 1, 3
12345+
Household size (people)
3
4
5
6
7
8
9
10
Showers per person
per week
Median
25%-75%
Non-Outlier Range
0 - 5 5 - 7 7+
Shower use per person per week
0
200
400
600
800
1000
Water use (L/hh/day)
Median
25%-75%
Non-Outlier Range
12345+
Household size (people)
0
5
10
15
20
25
30
35
40
Toilet flushes per
person per week
Median
25%-75%
Non-Outlier Range
0 - 10
10 - 15
15 - 20
20 - 25
25+
Toilet flushes per person
per week
0
200
400
600
800
1000
Water Use (L/hh
/day)
Median
25%-75%
Non-Outlier Range
12345+
Household size (people)
0
2
4
6
8
10
12
14
16
Washing dishes
(loads/week)
Median
25%-75%
Non-Outlier Range
0 - 5 5 - 10 10 - 15 15+
Washing dishes (Loads/week)
0
200
400
600
800
1000
1200
Water use (L/hh/Day)
Median 25%-75% Non-Outlier Range
7
and 4 person households. Figure 14 reveals that household water use increases with more frequent
clothes washing up to 7 loads per week and declines thereafter.
Figure 13: Frequency of clothes washing
versus household size
Figure 14: Household water use versus
frequency of clothes washing
4. CONCLUSIONS
The water use behaviour of households throughout South East Queensland was monitored using mini
smart meters during the period February 2009 to May 2010. Audits were also completed to accurately
define the characteristics of each household. Each household also completed a water use diary.
Metering results were combined with responses to questionnaires and water use diaries to develop an
understanding of water-use behaviours throughout the SEQ region. The size of the residential sample
was relatively small. However, the length and multiple processes involved in the monitoring allowed a
range of insights from the study. Residential water use was observed to be dependent on household
size. However, increases in per-capita residential water uses were seen to diminish with increased
household size. The relationship between per-capita water use and household size has a non-linear
form. A non-linear relationship was also observed between household income ($/annum) and
household or per-capita water use. Increases in household income were related to small increases in
per-capita water demands and increases in household size. Households with higher incomes were
more likely to have water efficient appliances and rainwater tanks. This result explained the small
increases in per-capita and household water demands associated with households with higher
incomes. Per-capita water use was found to be strongly dependent on the reported frequency of
showering.
Households without rainwater tanks had an average daily mains water demand of 514 litres and a per-
capita daily demand of 139 litres. Households using rainwater for outdoor use only had an average
daily mains water demand of 383 litres and a per-capita daily demand of 117 litres – annual savings of
47.8 kL. Households using rainwater for indoor and outdoor uses had an average daily mains water
use of 268 litres and a per-capita demand of 63 litres – annual savings of 89.8 kL. The performance of
households using rainwater for indoor and outdoor uses exceeded the requirements of Queensland
Development Code MP 4.2. Relatively small rainwater tanks (2 kL) and roof areas (50 m2 to 100 m2)
generated the largest reductions in mains water use. Use of rainwater for indoor uses reduced peak
daily and hourly mains water demands which is expected to diminish impacts on and requirement for
water distribution, pumping and treatment infrastructure.
5. ACKNOWLEDGEMENTS
This project benefitted from dedicated contributions of Josh Bairstow and Andrea McGrath, and from
the reviewers. The support of the QWC and Synergies Economics is gratefully acknowledged.
6. REFERENCES
Cui L., M. Thyer, P.J. Coombes, G. Kuczera, (2008). A Hidden State Markov Model for Identifying the
Long Term Dynamics of Indoor Household Water Uses. 31st Hydrology and Water Resources
Symposium. Adelaide. Australia.
ABS (2007). 2006 Community profile series. Cat.No. 2001.0. Commonwealth Government of Australia.
12345+
Household size (people)
0
10
20
30
40
50
Washing clothes
(loads/week)
Mean
Mean±SD
Mean±1.96*SD
0 - 5 5 - 7 7+
Clothes washing (loads/week)
0
200
400
600
800
1000
Water use (L/hh/day)
Median
25%-75%
Non-Outlier Range
8