Conference PaperPDF Available

Abstract and Figures

Residential water use has become a critical issue of the built environment due to drought and increasing water retail price in many regions around the world. However, there is limited research done to understand water use behavior in residential buildings. This paper presents data analytics and results from monitoring data of daily water use in 50 single-family homes in Texas, USA. Based on data analysis, residents’ regular water use patterns are investigated. The results help generate awareness of water use behavior and support further studies in clustering water use behavior patterns and developing water use models for simulation.
Content may be subject to copyright.
Water Use Behavior in Single-Family Homes: A Case Study in Texas
Peng Xue1,2, Tianzhen Hong3, Bing Dong4
1Beijing Key Laboratory of Green Built Environment and Energy Efficient Technology, Beijing
University of Technology, China
2Beijing Key Laboratory of Green Built Environment and Energy Efficient Technology, China
3Building Technology and Urban Systems Division, Lawrence Berkeley National Laboratory, USA
4Department of Mechanical Engineering, University of Texas at San Antonio, USA
Corresponding Author: Tel.: +86 18813030116; E-Mail:
Residential water use has become a critical issue of the
built environment due to drought and increasing water
retail price in many regions around the world. However,
there is limited research done to understand water use
behavior in residential buildings. This paper presents data
analytics and results from monitoring data of daily water
use in 50 single-family homes in Texas, USA. Based on
data analysis, residents regular water use patterns are
investigated. The results help generate awareness of water
use behavior and support further studies in clustering
water use behavior patterns and developing water use
models for simulation.
Most countries around the globe are experiencing a water
crisis today. One-third of the global population lives
without access to a toilet. A number of people equal to
twice the population of the United States live without
access to safe water (WHO and UNICEF 2015). Drought
conditions in the United States, including threatening
drought in California over the last four years, are causing
a re-examination of the value of water. Several western
states in the United States are surviving the most severe
drought conditions in history, with normal, seven-day
average stream flows at “extreme hydrologic drought”
and “severe hydrologic drought” levels (USGS 2015).
The latest 5-year report of the U.S. Geological Survey
(Maupin et al. 2010) indicates that total domestic water
use, including self-supplied withdrawals and public-
supply deliveries, was at 103,709 million liters a day in
2010, with California and Texas ranked the first two in
the total water withdrawals among 50 states. As a nation
overall, average domestic daily water use (DWU) per
capita is reported as 333 liters (88 gallons), which
includes potable and non-potable water and includes both
indoor and outdoor use. The average per capita use for
total domestic water use decreases 10% in last five years
and it still represents potential for water conservation
(Maupin et al. 2010). Significant energy and associated
cost savings are also possible with the reduction in water
demand (Malinowski et al. 2015).
Studies over the last decade found that domestic water use
is related to many factors. In an Arizona study, Balling et
al. (2008) claimed that 70% of household monthly
variance in water use could be explained by atmospheric
conditions in the state. In a Korean study, Praskievicz and
Chang (2009) confirmed that the weather condition plays
a key role in water use in that country. Water pricing
policy was shown to have influence on single-family
residential water use (Polebitski et al. 2010). A study by
Wentz et al. (2014) showed that the age of residents was
not a significant factor affecting domestic water use,
while another study showed that the number of teenagers
was a key variable of indoor water use (Aquacraft 2015).
Rosenberg and Madani (2014), in their editorial,
suggested that there is a need to think how water interacts
with energy. Household water and energy use are
heterogeneous and skewed with large variations among
households, but individual appliance shows great energy-
water linkage (Abdallah and Rosenberg 2012).
Water use characteristics can only be observed and
recorded by a person with relatively long intervals before
the installation of data loggers. The output could be just
the descriptive results such as the DWU per capita
(Bullock et al. 1980) and the hourly water use per
household (Papakostas et al. 1995). During the mid 1990s,
researchers in Boulder, Colorado, started using data
logging technique in data collection. (DeOreo and Mayer
1994, DeOreo et al. 1996). With this technique, a
computerized sensing device is attached to the water
meter and measures flow into the house at 10-second
intervals. This makes it possible to obtain and analyze
good resolution of water use data from a larger sample.
End-use analysis includes disaggregate water use into
end-use components, such as bathing, washing clothes
and dishes, and flushing toilets, etc. In a well-known
study, Residential End Uses of Water (REUWS),
published in 1999 by the Water Research Foundation and
the American Water Works Association, researchers
showed that the average DWU of 262.3 liters per capita
per day (lpcd) in single-family homes goes into eight end-
use components: toilets, faucets, leaks, clothes washers,
dishwashers, showers, baths, and other (Mayer et al.
1999). Other studies (DeOreo et al. 2011) show similar
findings, which are essential for establishing benchmarks
(Mayer 2009) and developing water devices. Other
research shows that introducing engineered water
efficiency devices could reduce indoor water use by 35%
to 50% (Inman and Jeffrey 2006).
Occupant behavior-related water use in residential
buildings is a critical issue for water conservation and
water use prediction. Occupant behavior is complex and
stochastic, causing a high DWU variability both among
residences and within the same residence (Lutz 2012).
Corral-Verdugo et al. (2003) found that general beliefs
could influence specific water beliefs, and in turn could
affect water consumption. Willis et al. (2010) investigated
the effect of visual display monitors on residents’ shower
behavior; results confirmed a significant effectiveness
with 27% reduction in a shower water use event.
Consumer behavior may also be negatively affected by
water-saving devices. Inman and Jeffrey (2006) found
that residents took longer showers and consumed more
water after installation of water-saving devices, due to the
belief that their water-saving devices would save water
(rebound effect).
To predict water use, a demand model needs to be
developed. Chu et al. (2009) proposed a framework of
residential water demand with correlations among
variables, but much work remained to be explored. Chang
et al. (2010) also developed a water demand framework
that incorporates existing factors with urban development
policies. Analytical, hybrid, and regression models have
been tested for characterizing the households’ water
saving by retrofits (Suero et al. 2012). New demand
models based on empirical data have been tested to
predict better results (Aquacraft 2015). However, most
existing studies on water use behavior models are
observed from the perspective of use time of water-
consuming devices and lack in-depth behavioral analysis.
While energy-related occupant behavior has been studied
extensively for residential and commercial buildings
(Hong et al. 2015; Yan et al. 2015), water use behavior is
under-researched. Aiming to provide insights into
household water use behavior, this paper presents
analytical results from monitoring data of DWU in 50
single-family homes in Texas, USA, as well as
exploration of possible reasons behind household water
use behaviors.
This study uses data collected through a project by Pecan
Street, Inc. (, which is
the world's largest source of disaggregated customer
energy and water use data. The data are stored in 25 tables
in a SQL database, which consists of weather data, water
use, audits, annual surveys, energy consumption and other
information (e.g., gas use). The Pecan Street database
includes 1338 houses, 1105 of which are still active. The
database started collecting data January 1, 2011, and
continues up to the time of this study (September 26,
2015). Energy data is recorded in 1-minute time intervals,
while water use is recorded as daily sum before May 10,
2013, and by minute from then on.
In this study, household information comes from the
survey tables. DWU value is calculated from the water
usage table, which shows a household’s total water use
within a specific time interval (by day or minute). Energy
consumption data are from the hourly energy-use table,
which contains 67 columns showing energy
consumptions of different appliances in a house. Water
use data in the database is sparse and not always
continuous. After data processing (excluding the ones
without water use data), 50 single-family houses are
selected for this study.
Data processing
The big data of houses were first downloaded from the
database as comma-separated-value (CSV) files. The
main purposes were to calculate the DWU and daily
energy use from the cumulative data (by hour and minute)
for each house, and to convert units of the measured data.
All pre-processed data were further processed in the
following steps.
The second step was to clean all the translated data
obtained from the previous step, which includes
summarizing all household data into one sheet with
outdoor air temperature in chronological order and
removing data (cumulative raw data) with gaps of more
than one day.
After the translating and cleaning steps, 11852 logging
data points for 60 houses were collected in one Excel
sheet. Some zero values of DWU were also included,
which reflected that no residents were home and
consumed no water on those days. As the zero values may
have a significant influence on calculating the average
and DWU values, small values such as 0 and 1 liter/day
were excluded in the study of water use behavior. After
applying the above criteria, 10 of the households with
valid data had data for less than a month. These 10
households were excluded from the originally selected 60.
In the end, water use data for the remaining 50 households
were used in the study.
After all data were pre-processed, a dataset of 10659 valid
DWU values from 50 houses was built. Combined with
the house information, the data were summarized by
different objectives and shown in Table 1.
Analysis procedure
The first step of the analysis procedure is to investigate
the time and frequency distribution of DWU of a typical
house. Further relation between DWU and outdoor air
temperature, day of the week, and season are also studied.
A sudden (anomaly) peak of DWU is found as a common
phenomenon in many homes, which will be discussed
with leakage, water intensive use and residents’ habit in
the Discussion section. The second step is to generate
generic results using normalized data from all 50 homes,
to find water use patterns between weekdays and
weekends, and to establish a baseline model of DWU for
single-family homes. By comparing the results among
different houses, the third step is to find related factors
affecting DWU, namely residents’ income, education, age
and daily activity. The information for all of the selected
50 households is shown in the appendix (except for the
exact house ID which was anonymized due to privacy
Statistical analysis methods
1) Spearman's rank correlation coefficient
Spearman's rank correlation coefficient is adopted to
describe the relationship between two variables by
assessing the monotonic function. A perfect value of +1
or −1 occurs when one variable is a perfect monotone
function of the other. The coefficient ρ could be computed
𝜌 = 1 − 6 𝑑𝑖
2/𝑛(𝑛2− 1) (1)
where di is the difference between ranks of two variables;
i is the case number; n is the total number of cases. This
correlation coefficient was applied to investigate the
relationship between the DWU per house and the age
groups of occupants in the house.
2) Frequency distribution
In this study, frequency distributions are displayed as
graphs that show the frequency of DWU in the whole
dataset. A frequency distribution shows a summarized
grouping of DWU values divided into mutually exclusive
intervals and the number of occurrences in an interval.
3) Median for baseline
Water use distribution should be studied with medians,
not averages, as the feature is not symmetrical (Lutz 2012).
In this study, median values of all logging data can be
obtained in three different ways. The first method is to
obtain the median values from all logging data directly,
the second method is to calculate them from all household
median DWU values, and the third method is to calculate
them from all household average DWU values.
The first method chooses the median value from all the
data but ignores the fact that the number of data points
from each household is not the same (as shown in
appendix). The second method is more appropriate, which
considers the differences between households and obtains
the median values of each house first. However, the
median value of a house can only be explained as the most
likely condition. The value itself ignores the high water
use condition and sudden peak, which should be
considered as the behavior of the residents. Therefore, the
third method is most appropriate to establish the baseline,
which calculates the average values of DWU for each
household first and then finds the median DWU for the
entire dataset of 50 households.
Result analysis
Statistical analysis of a single house
We started studying the residents’ water use behavior in a
single house. House No. 7 is selected with the most
logging data points538 days of valid datafrom the 50
monitored homes.
With the 538 valid data of DWU, the frequency
distribution is shown as Figure 1. The X-axis interval is
set as 40 or 75 liters per day (lpd) and Y-axis shows the
occurrence number of days.
Figure 1: Frequency and cumulative distributions of
DWU (House No. 7)
The frequency distribution of DWU shown in Figure 1 is
neither symmetrical nor normal distribution. The curve
has a long tail, it features a striking peak around 450 lpd,
and most of the data are equal or greater than 300 lpd.
However, there are still 189 days when the DWUs are
much more than the average of 730.28 liters per
household per day (lphd). A second peak appears around
2250 lpd, which indicates another behavior pattern of high
water use that needs further study. It is worth noting that
this is a typical feature of DWU frequency distribution:
almost all 50 homes show a distribution with two or three
DWU differs from day to day and has large variations.
Monitored data from House No. 7 are shown in Figure 2
with the X-axis of outdoor air temperature. The dataset
grouped by weekdays and weekends is shown in Figure
2a, while it is also grouped in seasons as shown in Figure
2b. The seasons are divided by solstices and equinoxes.
Figure 2: DWU (House No. 7) with outdoor air
temperature: a) by weekdays and weekends; b) by
The phenomena of summer peak and the dual peaks in
Figure 1 can be reflected as the two-layer feature in this
figure. Figure 2 shows that the relationship between
household DWU and outdoor air temperature is not linear.
However, the two-layer feature indicates that residents
keep basic requirements of water use and do not use much
water for irrigation when the outdoor air temperature
drops below 15. As seen from the higher layer in Figure
2a, the high water use behavior occurred in both weekdays
and weekends, indicating that the residents have a
constant 2250 lpd of water use once or twice a week.
Figure 2b shows that the water irrigation behavior has
strong seasonality, with the winter months having lower
values. The average DWU values from spring to winter
are 593.87 lphd, 948.03 lphd, 694.81 lphd, and 607.42
lphd, respectively.
There is also an isolated data point with a very high value
in Figure 2, which is more than twice the value of other
data points. This kind of anomaly peak happens in almost
half of the 50 homes, which may result from water leaks,
watering, or filling swimming pools. This anomaly is
considered further in the Discussion section.
Statistical analysis of 50 houses
The data show large variations in water use from day to
day and from home to home. It is important to normalize
the water use for a single-family home on the basis of
number of persons living in the home and the total floor
area of the house. Figure 3 shows several water use
metrics for the 50 homes, including DWU median per
household, average DWU per household, average DWU
per capita, and cumulative distribution function of DWU
per household. The results are sorted by the average DWU
per household.
Figure 3: Water use metrics for 50 houses
As seen in Figure 3, the average DWU differs
significantly from house to house; the largest two houses
reach 2250 lpd. This figure also shows that the top 26%
of households use 48% of total water. The overall average
DWU of all houses is 676.27 liters, and the median DWU
of each house is also shown in the figure, which is less
than the average DWU. The median DWU in 95% of the
houses is between 90 lphd and 650 lphd. This result
reflects the frequency distribution curve of DWU is not
symmetrical and the long tail is significant. House No. 36
has a median DWU of 613.17 lphd and an average DWU
of 2267.84 lphd. The figure also shows the DWU
normalized by the number of residents. The calculated
result indicates that nearly 25% of people use 51% of the
total water. The overall average DWU per capita is 272.81
liters. The DWU, normalized by capita and square meter,
is also provided in the figure, but it can be much higher
for small houses.
In conclusion, this study found that it most effective and
appropriate to study the chosen data normalized by capita.
These results also show that the Pareto Principle is in
operation in this water use study. If high water use
households (or people) improve (decrease consumption),
water will be significantly saved.
The next analysis had to do with household DWU during
the week versus on the weekend. The houses grouped by
the types of occupancy are shown in Figure 4, and the
average DWU of each house is separated with the average
DWUs for both weekdays and weekends.
Figure 4: Average DWU during weekdays and weekends
for 50 houses among three occupancy types
Figure 4 shows the average DWU values of all houses
represented by a solid line. Compared with this solid line,
it is clear whether residents use more water during
weekdays or not. Average DWU is closer to weekdays
DWU since weekdays have a higher weighting factor.
Results of this analysis show that 68% of houses consume
more water per day on weekends than weekdays.
However, some houses have higher average DWU on
weekdays. Considering the occupancy on weekdays, no
significant relation can be found. According to results of
first two groups in Figure 5, both groups have households
using more water on weekdays than weekends. It seems
that DWU is less affected by occupancy than by residents’
Looking at all valid DWU data points as a whole, the
frequency distribution of all 50 houses is shown as Figure
5. The X-axis interval is set as 40 lpd.
Figure 5: Frequency distribution of DWU among 50
From Figure 5, it can be seen that the distribution curve
only has one peak and a long tail. The average water use
per capita per day across all 8949 data points is 272.81
lpcd but the standard deviation can be as high as 521.48
lpcd. When studying the baseline, the median value is
often adopted as a fair rule. In this study, median value is
obtained from all 50 houses’ average DWU value (Figure
3); the result is 186.00 lpcd. Therefore, the baseline of the
DWU for these households can be set as 186.00 lpcd.
Cross comparison of all houses
After being normalized by the number of residents, the
DWU of each house can be studied in more detail. The
box plot of DWU per capita for 40 houses is shown in
Figure 6 as the other 10 houses have no information about
the number of residents. The results in Figure 6 are sorted
by the median DWU value.
Figure 6: Box plot of DWU per capita for 40 houses
sorted by median values
As seen in Figure 6, the median values of DWU per capita
among the 40 houses are between 15 liters and 320 liters.
The median value of these household median DWU
values is 127.76 liters. Focusing on the highest value at
each house shows that 50% of households have median
DWU values higher than 1500 lpd, which means these
data may be experiencing anomaly peak. Some of the
interquartile ranges (cubic length) shown in the figure are
very big. This result reflects that the residents in some
householdsnamely houses 12, 25, 28, 35, and 36have
frequent high water use behaviors compared to their own
average DWU. The detailed DWU results on a long
interquartile range will be shown in the next section.
As discussed, the DWU of each householdeven the
value of DWU per capitadiffers significantly. This
research next looked at which factors could account for
higher water use in some households compared to others,
factors including higher personal income, better education,
teenagers at home, or washing behaviors. Water use was
also compared to energy use to see if energy use somehow
correlates to water use.
Income and education
To test assumptions that might explain the correlation
between income, education, and DWU, the next analysis
normalized DWU and income by the number of residents
in each house. The assumption was that residents with
higher income may have a higher standard of living and
consume more water. The personal income is calculated
from house information and grouped in seven levels, as
shown in Table 1. The relation between DWU per capita
and personal income is shown in Figure 7, with the levels
of education presented in different shapes.
Table 1: Personal income levels
capita ($)
Figure 7: Average DWU per capita with personal
income and education
The analysis represented in Figure 7 seems to show that
DWU per capita has a positive correlation with the
personal income levelresidents with higher incomes
seem to consume more water. Focusing on the personal
income level from 3 to 6, residents with college degrees
use more water than the postgraduates, on average.
Though the numbers of cases are not equal, people with
undergraduate college degrees consume the most in three
out of four income levels. Using these 40 cases as a guide,
it is reasonable to say that residents with more education
are likely to use less water that those residents with less
Age group
The next analysis examined the assumption that teenagers
use more water than other age groups. It is difficult to
separate DWU by age groups since a house may hold
people in several different age groups. Therefore,
Spearman correlation coefficient is adopted to study the
relation of DWU per household with age group. The key
group will be presented with significant coefficient,
which means that the corresponding group has more
weight to inform the house total DWU. The result of the
statistical analysis of 40 houses and 108 residents is
shown in Table 2.
Table 2: Spearman correlation coefficients of DWU per
house and age groups
DWU per
* Correlation significant at the 0.05 level (two-tailed).
As seen in Table 2, no significant value is presented. The
results make it clear that there is no significant correlation
between DWU per home and age groups; there can be no
assumption that any age group uses more water than
Energy use in appliances
Among the 50 houses, only four have both daily total
energy consumption data and DWU data at the same time.
House No. 19 has the longest monitored days among these
four houses and its energy use is also sub-metered into
three separate data streams, all assumed to have direct
relation with water use behavior: bathroom, clothes
washer, and dish washer. Figure 8 shows the daily energy
use of those three data streams and the DWU of House No.
Figure 8: DWU (House No. 19) and appliances energy
As seen from Figure 8, the bathroom shows the most
constant and consistent use, at the frequency of 92 days
out of a total of 116 monitoring days. While the clothes
washer and dish washer are operated in 43 and 37 days,
respectively, which are twice a week on average. Energy
consumption in the bathroom is much less than the energy
consumption of the clothes washer and dish washer, on
average. This may result from the fact that light bulbs
often have the power level of less than 100 W, while the
dish and clothes washers have the power level of more
than 2000 W. Given their different power, even if lights
are turned on in the bathroom, its overall consumption is
lower than that of dishwashers or washing machines. In
general, the DWU has ups and downs over the monitored
days; it seems higher DWU points have a corresponding
higher use of energy. To test the bivariate relationship
between DWU and daily energy use, linear regression is
adopted. Household daily total energy use is set to be the
variable at first and the result is shown in Figure 9a.
Figure 9: DWU (House No. 19) with the energy use: a)
daily total energy use; b) daily energy use of clothes
From Figure 9a, a positive correlation is seen between
DWU and daily energy use, though not a strong
correlation, with an R2 value of 0.403. The disparity may
come from the incomplete statistics and residents’
different behaviors between water use and energy use.
Therefore, the energy uses of bathroom, clothes washer,
and dish washer are set to be variables separately as they
have direct influence on water use. These results show
that the daily energy use of the clothes washer has a better
positive correlation with DWU, with R2 value of 0.646
(Figure 9b). Meanwhile, the DWU per household could
not be predicted by the energy use of the bathroom as little
lighting energy is used in their bathrooms.
However, not all families use dish washers or other
appliances. Restricted by the sample size, this result only
proves that energy use can indicate the condition of water
use in residential buildings qualitatively. The most
important is saving water saves energy.
In this case study, 95% of house median DWU values are
between 90 and 650 liters per household per day, and the
baseline DWU can be set as 186.00 lpcd for these houses.
Due to the limitation of the house information and no sub-
metering of water use data, the current result could just
show that 25% of residents consumed 51% of the total
water. High water use households (or people) have
potentials to significantly reduce water consumption.
These results help generate awareness of water use
behavior in homes and support further studies in
clustering water use behavior patterns and developing
water use models for simulation.
In contrast with previous research, our study found that
DWU is less affected by occupancy than by residents’
habits, and the resident’s habits of water use is much
different from those of energy use.
This work is supported by the Assistant Secretary for
Energy Efficiency and Renewable Energy of the U.S.
Department of Energy under contract number DE-AC02-
05CH11231. The source data were provided by Pecan
Street, Inc., headquartered in Austin, TX. The authors
thank this non-profit research institute for allowing us
access to their subscriber water usage database.
Abdallah AM and Rosenberg DE. Heterogeneous
residential water and energy linkages and
implications for conservation and management.
Journal of Water Resources Planning and
Management, 2012, 140(3): 288-297.
Aquacraft Inc. Application of end use study data for
development of residential demand models. 2015.
Retrieved from:
Balling RC, Gober P and Jones N. Sensitivity of
residential water consumption to variations in climate:
an intraurban analysis of Phoenix, Arizona. Water
Resources Research. 2008; 44.
Bullock DC, Peebles RW and Smith HH. Water usage
patterns in the U.S. Virgin Islands. Water Resources
Research Center, Caribbean Research Institute,
College of the Virgin Islands, 1980.
Chang H, Parandvash GH and Shandas V. Spatial
variations of single-family residential water
consumption in Portland, Oregon. Urban geography.
2010; 31: 953-972.
Chu J, Wang C, Chen J and Wang H. Agent-based
residential water use behavior simulation and policy
implications: A case-study in Beijing City. Water
Resour Manage. 2009.; 23: 3267-95.
Corral-Verdugo Vc, Bechtel RB and Fraijo-Sing B.
Environmental beliefs and water conservation: An
empirical study. Journal of Environmental
Psychology. 2003; 23: 247-257.
DeOreo WB, Heaney JP, Mayer PW. Flow trace analysis
to assess water use. American Water Works
Association. Journal, 1996, 88(1): 79.
DeOreo WB and Mayer PW. Project report: A process
approach for measuring residential water use and
assessing conservation effectiveness. City of Boulder
Office of Water Conservation, Boulder, Colorado.
DeOreo WB, Mayer PW, Martien L, et al. California
single-family water use efficiency study. Aquacraft
Water Engineering and Management, Boulder,
Colorado, USA. 2011. Retrieved from:
Hong T, Taylor-Lange SC, D’Oca S, Yan D, Corgnati S.
Advances in research and applications of energy-
related occupant behavior in buildings. Energy and
Buildings, 2015; 116: 694-702.
Inman D and Jeffrey P. A review of residential water
conservation tool performance and influences on
implementation effectiveness. Urban Water Journal.
2006; 3: 127-143.
Lutz J. Hot water draw patterns in single-family houses:
findings from field studies. Lawrence Berkeley
National Laboratory, 2012: LBNL Paper LBNL-
4830E. Retrieved from:
Malinowski PA, Stillwell AS, Wu JS and Schwarz PM.
Energy-water nexus: potential energy savings and
implications for sustainable integrated water
management in urban areas from rainwater
harvesting and gray-water reuse. Journal of Water
Resources Planning and Management. 2015; 141(12):
Maupin MA, Kenny JF, Hutson SS, Lovelace JK, Barber
NL and Linsey KS. Estimated use of water in the
United States in 2010: U.S. Geological Survey
Circular 2014.1405, 56 p.
Mayer PW. Water efficiency benchmarks for new single-
family homes. Aquacraft Water Engineering and
Management, Boulder, Colorado, USA. 2009.
Retrieved from:
Mayer PW, DeOreo WB, Opitz EM, et al. Residential end
uses of water. AWWA Research Foundation and
American Water Works Association Denver, CO,
Papakostas KT, Papageorgiou NE and Sotiropoulos BA.
Residential hot water use patterns in Greece. Solar
Energy. 1995; 54: 369-374.
Pecan Street, Inc. Pecan Street's Dataport.
Polebitski AS, Palmer RN and Waddell P. Evaluating
water demands under climate change and transitions
in the urban environment. Journal of Water
Resources Planning and Management. 2011; 137(3):
Praskievicz S and Chang H. Identifying the relationships
between urban water consumption and weather
variables in Seoul, Korea. Physical Geography. 2009;
30: 324-337.
Rosenberg DE, Madani K. Water resources systems
analysis: A bright past and a challenging but
promising future. Journal of Water Resources
Planning and Management, 2014, 140(4): 407-409.
Suero FJ, Mayer PW and Rosenberg DE. Estimating and
verifying United States households’ potential to
conserve water. Journal of Water Resources Planning
and Management, 2012, 138(3): 299-306.
U.S. Geological Survey. State drought information. U.S.
Department of the Interior, 2015. Retrieved from:
Wentz EA, Wills AJ, Kim WK, Myint SW, Gober P and
Balling Jr RC. Factors influencing water
consumption in multifamily housing in Tempe,
Arizona. The Professional Geographer. 2014.; 66:
WHO, UNICEF. Progress on sanitation and drinking
water: 2015 update and MDG assessment. Geneva:
World Health Organization, 2015.
Willis RM, Stewart RA, Panuwatwanich K, Jones S and
Kyriakides A. Alarming visual display monitors
affecting shower end use water and energy
conservation in Australian residential households.
Resources, Conservation and Recycling. 2010; 54:
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Occupant behavior is one of the major factors influencing building energy consumption and contributing to uncertainty in building energy use prediction and simulation. Currently the understanding of occupant behavior is insufficient both in building design, operation and retrofit, leading to incorrect simplifications in modeling and analysis. This paper introduced the most recent advances and current obstacles in modeling occupant behavior and quantifying its impact on building energy use. The major themes include advancements in data collection techniques, analytical and modeling methods, and simulation applications which provide insights into behavior energy savings potential and impact. There has been growing research and applications in this field, but significant challenges and opportunities still lie ahead.
Technical Report
Full-text available
The California Energy Commission (CEC) is updating the building energy efficiency standards (Title 24 regulations). Water heating is one of the largest energy end uses in residential buildings and therefore an important consideration in the building energy efficiency standards. Hot water draw patterns in residential buildings are used in the water heating energy calculations of the standards. To improve the calculations, we analyzed hot water draw patterns in single-family residences using data from a number of recent studies that monitored hot water use in single-family residences at time resolutions of 1 minute or less. This report presents the volume of hot water use and the number of draws per day as a function of the number of people and as a function of conditioned floor area. Hourly hot water use schedules are also presented. We compare the results from this study with the assumptions in the Title 24 calculations and with results from previous studies, and find that some of the assumptions should be modified.
Full-text available
As anthropogenic climate change threatens the reliability of urban water supplies, it is essential to build understanding of the relationships between weather and water consumption. We used daily and monthly data from 2002 to 2007 to conduct a statistical analysis of how seasonal water use in Seoul, South Korea is affected by weather variables. The Pearson, Kendall, and Spearman tests indicated that all weather variables were significantly correlated with per capita water use at most timescales, with mean, minimum, and maximum temperatures and daylight length positively correlated, and precipitation, wind speed, relative humidity, and cloud cover showing an inverse relation with water use. Once the influence of maximum temperature is controlled, water consumption is only significantly associated with wind speed and daylight length, as indicated by the partial correlation coefficient values. Ordinary least square (OLS) regression models explain between 39 and 61% of the variance in seasonal water use, indicating that approximately one-third to two-thirds of the variation is due to weather variables alone. Daily water consumption in July increases up to 4 liters per person with a one degree increase in maximum temperature. Significant improvement of the modeling of seasonal water use was achieved by developing autoregressive integrated moving average (ARIMA) models, which account for autocorrelation in the time series and explain up to 66% of the variance in water use. Our results indicate that weather plays a significant role in determining water consumption in Seoul, and that has important implications for management of urban water resources under potential future climate change.
Saving water saves energy. Consequently, implementing integrated water management (IWM) measures that reduce potable water consumption, stormwater runoff, and wastewater generation can potentially translate into significant energy savings. In this paper, the energy savings associated with IWM measures of rainwater harvesting and gray-water reuse are estimated, both at national and local utility scales using published data. At the national scale, it is estimated in this paper that up to 3.8 billion kWh and $270 million can potentially be saved annually by replacing landscape irrigation and other outdoor water uses through rainwater harvesting alone, and up to 14 billion kWh and $950 million in combination with gray-water reuse. Similarly, in Charlotte, North Carolina, the local water utility can potentially save up to 31 million kWh and $1.8 million annually. However, annual energy and associated cost savings per household are low at either scale, ranging between 1 and 120 kWh with associated cost savings of less than $10. These results are discussed in terms of energy savings' role in IWM policy considerations and promotion of sustainable water use in urban areas.
Behavior and technological impacts on residential indoor water use and conservation efforts in the United States are identified. Preexisting detailed end-use data was collected before and after toilets, faucets, showerheads, and clothes washers were retrofitted in 96 owner-occupied, single-family households in Oakland, California; Seattle, Washington; and Tampa, Florida, between 2000 and 2003. Water volume, duration of use, and time of use were recorded and disaggregated by appliance for two weeks before and four weeks after appliances were retrofitted. For each appliance, observed differences in water use before and after retrofits are compared to water savings predicted by simple analytical, regression, and hybrid models. Comparisons identify prediction precision among models. Results show that observed and predicted distributions of water savings are skewed with a small number of households showing potential to save more water. Regression and hybrid model results also show the relative and significant influence on water saved of both technological (flow rates of appliances) and behavioral (length of use, frequency of use) factors. Additionally, regression results suggest the number of residents, performance, and the frequency of appliance use are key factors that distinguish households that save the most water from households that save less. Study results help improve engineering methods to estimate water savings from retrofits and allow water utilities to better target subcategories of households that have potential to save more water.
Central to the Smart Growth movement is that compact development reduces vehicle miles traveled, carbon emissions, and water use. Empirical efforts to evaluate compact development have examined residential densities but have not distinguished decreasing lot sizes from multifamily apartments as mechanisms for compact development. Efforts to link design features to water use have emphasized single-family at the expense of multifamily housing. This study isolates the determinants of water use in large (more than fifty units) apartment complexes in the city of Tempe, Arizona. In July 2007, per bedroom water use increased with pool area, dishwashers, and in-unit laundry facilities. We are able to explain nearly 50 percent of the variation in water demand with these variables. These results inform public policy for reducing water use in multifamily housing structures, suggesting strategies to construct and market “green” apartment units.
This paper develops an integrated approach to model heterogeneous household water and energy use and their linkages. The approach considers variations in behavioral and technological water and energy use factors that affect U.S. indoor residential water and energy use for toilets, showers, faucets, clothes washers, and dishwashers. The study uses a recent, large, national, disaggregated household water use data set collected from 11 cities, as well as national energy data on water heater efficiency and setpoint/intake temperatures. First, probability distributions of water and energy use factors are identified and correlated. Then, Monte Carlo simulations are used to calculate probability distributions for estimated household water and energy use. Finally, linear regressions are used to find the relative effects of water and energy factors on household energy use. Results show that water and energy distributions among households are skewed, with the largest 12% of the users consuming 21% and 24% of water and energy, respectively. Water heater setpoint temperature followed by intake temperature, heater efficiency, shower hot water percentage, household size, shower flowrate, and faucet flowrate have the highest relative effect on household energy use and should be targeted to reduce household energy use. The approach improves prior homogenous and deterministic water-energy models and can help utilities select and size cost-effective, collaborative water and energy conservation actions.
Precise information about water use patterns can be gathered by analyzing flow traces obtained from residential customer water meters that are fitted with portable data loggers. Flow traces are precise enough that signatures associated with all major water use categories can be identified. For this study, more than 10,000 water use events were recorded, classified, and entered into a database. The technique is both accurate and reliable and can be used to collect time-specific and disaggregated water use data. Measuring directly instead of inferring measurements from aggregated data is a quick and cost-effective way to analyze water use patterns and directly assess how conservation measures influence water demand.
Water remains an essential ingredient for the rapid population growth taking place in metropolitan Phoenix, Arizona. Depending upon the municipality, between 60 and 75% of residential water is used outdoors to maintain nonnative, water-intensive landscapes and swimming pools. Residential water use in Phoenix should be especially sensitive to meteorological and climatic variations because of the strong emphasis on outdoor water use. This study explores the intraurban spatial variations in the sensitivity of residential water consumption to atmospheric conditions. For 230 census tracts in the city, we developed times series of monthly water use anomalies and compared them with monthly anomalies of temperature, precipitation, and the Palmer Drought Hydrological Index. We found that one third of census tracts have little to no sensitivity to climate, while one tract had over 70% of its monthly variance in water use explained by atmospheric conditions. Greater sensitivity to atmospheric conditions occurred in census tracts with large lots, many pools, a high proportion of irrigated mesic landscaping, and a high proportion of high-income residents. Low climatic sensitivity occurred in neighborhoods with large families and many Hispanics. Results suggest that more affluent, non-Hispanic neighborhoods will be disproportionately affected by increasing temperatures due to urban heat island effects and the buildup of greenhouse gases.