Elsevier

Journal of Cleaner Production

Volume 60, 1 December 2013, Pages 129-146
Journal of Cleaner Production

Revealing the determinants of shower water end use consumption: enabling better targeted urban water conservation strategies

https://doi.org/10.1016/j.jclepro.2011.08.007Get rights and content

Abstract

The purpose of this study was to explore the predominant determinants of shower end use consumption and to find an overarching research design for building a residential water end use demand forecasting model using aligned socio-demographic and natural science data sets collected from 200 households fitted with smart water meters in South-east Queensland, Australia. ANOVA as well as multiple regression analysis statistical techniques were utilised to reveal the determinants (e.g. household makeup, shower fixture efficiency, income, education, etc.) of household shower consumption. Results of a series of one-way independent ANOVA extended into linear multiple regression models revealed that females, children in general and teenagers in particular, and the showerhead efficiency level were statistically significant determinants of shower end use consumption. Eight-way independent factorial ANOVA extended into a three-tier hierarchical linear multiple regression model, was used to create a shower end use forecasting model, and indicated that household size and makeup, as well as the showerhead efficiency rating, are the most significant predictors of shower usage. The generated multiple regression model was deemed reliable, explaining 90.2% of the variation in household shower end use consumption. The paper concludes with a discussion on the significant shower end use determinants and how this statistical approach will be followed to predict other residential end uses, and overall household consumption. Moreover, the implications of the research to urban water conservation strategies and policy design, is discussed, along with future research directions.

Introduction

Water is one of the most vital resources on earth. Due to climate change consequences such as the increasing frequency and severity of droughts and the unpredictable changing rainfall patterns, water availability is becoming more variable. Drought, together with growing populations which results in an escalating urban water demand are making water a scarce resource in many regional and urban centres (Dvarioniene and Stasiskiene, 2007, Giurco et al., 2010, Hubacek et al., 2009, Willis et al., 2009a, Willis et al., 2010b). Scarcity of water is forcing many governments and public utilities to invest significantly in the development and the implementation of a range of water strategies (Correljé et al., 2007, Stewart et al., 2010), including dual supply schemes (Willis et al., 2011b), shower visual display monitors (Willis et al., 2010a) and the installation of rainwater tanks (Tam et al., 2010). These strategies aim at improving urban water security through a more sensible and sustainable water consumption to meet future demand (Mahgoub et al., 2010, Palme and Tillman, 2008). This scenario is common in Australia and to some extent the world (Commonwealth of Australia, 2011a, Giurco et al., 2010, Inman and Jeffrey, 2006).

South-East Queensland (SEQ), Australia has been suffering a long drought period, varying rainfall patterns, and a rapid increasing population. These factors together have lead to the enforcement of water demand management (WDM) strategies. Such strategies include water restrictions, rebate programmes for efficient fixtures, water efficiency labelling, and conservation awareness programs (Inman and Jeffrey, 2006, Mayer et al., 2004, Nieswaidomy, 1992). In spite of reductions in water consumption resulting from the implementation of such WDM strategies, government usually follows reactionary-based approaches rather than proactive-based approaches (Beal et al., 2010). Additionally, their effectiveness is dependent on differences in location, community attitudes and behaviours (Corral-Verdugo et al., 2003, Turner et al., 2005, Stewart et al., 2011). Further, estimations of water savings yielded from the implementation of such strategies and programs are often calculated based on limited evidence and with many assumptions due to the lack of appropriate data at the end use level, thereby deriving understated or grossly inaccurate values for water savings associated with them (Willis et al., 2009d). Therefore, the development of effective urban water conservation strategies, policies and forecasting models is essential to better manage our urban water resources.

The development of effective strategies and policies requires more detailed information on how and where residential water is consumed (e.g. shower, washing machine, dish washing, tap, bathtub, etc.) (Mayer and DeOreo, 1999, Willis et al., 2009a). This detailed knowledge of water consumption can provide a greater understanding on the key determinants of each and every water end use, and in return, will allow for the development of improved long-term forecasting models (Blokker et al., 2010, Stewart et al., 2010). The formulation of such models is paramount, especially when there is a distinct lack of micro-component level models that have been created from empirical water end use event data registries into forecasts for total urban residential connection demand as presented in the herein study.

The advent of advanced technology such as water smart metering, which encompasses high resolution data capturing, logging and wireless communication technologies has facilitated the collection, wireless transfer, storing and analysing of abundant detailed and useful water end use information (i.e. time and quantity of each and every end use) (Willis et al., 2009d). The alignment of such detailed and accurate water end use data with a range of socio-demographic, stock inventory, residential attitude and behavioural factors, will aid the development of models that are capable of revealing the determinants of each and every end use; thereby providing the foundations for more robust urban water demand forecasting models.

Many residential water demand forecasting models have been developed based on historical billing data, existing statistical reports, or technical information from stock appliance manufacturers (Beal et al., 2010). Such models are not able to provide an accurate disaggregation of consumption into water end use categories. Therefore, long-term actual measurement and disaggregation of water end use data (i.e. micro-component analysis) using smart metering technology and computer software is considered the most robust and accurate foundation for the development of urban water demand forecasting models.

In general, there are few residential water end use studies that have been conducted using high resolution smart metering technologies. Internationally, a number of end use studies have been conducted in the United States of America (Mayer and DeOreo, 1999, Mayer et al., 2004) and more recently in New Zealand (Heinrich, 2007) and Sri-Lanka (Sivakumaran and Aramaki, 2010). Additionally, in South Africa, a conceptual end use model was developed by Jacobs (2004). Moreover, a number of water end use studies (also called water micro-component studies) have been conducted in the United Kingdom (Barthelemy, 2006, Creasey et al., 2007, Sim et al., 2007). In Australia, three major studies have been completed to date in Perth (Loh and Coghlan, 2003), Melbourne (Roberts, 2005) and most recently in Gold Coast City, Queensland (Willis et al., 2009a, Willis et al., 2009b, Willis et al., 2009c, Willis et al., 2009d, Willis et al., 2010a, Willis et al., 2010b, Willis et al., 2011a, Willis et al., 2011b). Table 1 summarises established averages of total and indoor daily per capita water consumption volumes, as well as the indoor water end use breakdown percentages of previous studies conducted in Australia.

In 2010, a South-east Queensland Residential End Use Study (SEQREUS) was commissioned with the objective to gain a greater understanding on water end use consumption in this large urbanised region. This study was funded by the Urban Water Security Research Alliance (UWSRA), which is a partnership between the Queensland Government, CSIRO’s Water for Healthy Country Flagship, Griffith University, and University of Queensland. The main aim of this alliance was to address SEQ’s emerging urban water issues to inform the implementation of enhanced water strategy (Beal et al., 2010). The primary objective of the greater study was to quantify and characterise mains water end uses of single detached dwellings across four main regions (i.e. Sunshine Coast Regional Council, Brisbane City Council, Ipswich City Council, and Gold Coast City Council) in SEQ, Australia, as shown in Fig. 1 (Beal et al., 2011).

This herein described study utilises information collected in the SEQREUS July 2010 baseline data, where a Permanent Water Conservation Measures (PWCM) daily target of 200 L per person per day (L/p/d) was set by the State Government (Beal et al., 2011). Both the reported SEQREUS and Queensland Water Commission (QWC) water use averages of 145.3 L/p/d and 154 L/p/d, respectively, fell well below the government set target as shown in Fig. 2 (Beal et al., 2010a, Queensland Water Commission, 2010). PWCM are not considered restrictions but mainly guidelines for the efficient use of potable water for irrigation purposes (e.g. irrigating lawns after 4 pm when less heat, etc.). Moreover, PCWM guidelines only provide very broad guidance on efficient indoor consumption. Thus in summary, there was not any restriction regime in place at the time of data collection related to this study that could have directly influenced householders’ indoor consumption.

This paper describes a component of this greater SEQREUS study. The herein described research study seeks to formulate a bottom-up residential end use demand forecasting model, which includes a comprehensive listing of predictor variables.

There are several factors influencing water consumption that have been reported previously. Such factors are socio-demographic and water stock efficiency related factors. Socio-demographic factors like household size and household income have been found to influence water consumption (Kim et al., 2007, Loh and Coghlan, 2003, Mayer and DeOreo, 1999, Renwick and Archibald, 1998, Turner et al., 2009). Additionally, other previous studies (Athuraliya et al., 2008, Heinrich, 2007, Mayer et al., 2004, Willis et al., 2009d, Willis et al., 2010a) have shown that the use of water efficient appliances and fixtures reduces water consumption.

As argued, smart metering and comprehensive end use studies provide immense opportunities to significantly improve current understanding on the determinants of residential water consumption, as well as the accuracy of demand forecasting models. A discussion on the relationship between a range of household descriptive characteristics, socio-demographic and stock efficiency characteristics and shower end use consumption is provided below.

Section snippets

Determinants of shower end use consumption

While the greater SEQREUS has a repository of all residential water end use events, this study has been focussed on the shower end use category. The reason for this is that shower end use consumption, is often the highest indoor demand in residential households. Greater understanding on the primary determinants of shower end use consumption, will aid the preparation of strategic plans (e.g. showerhead rebate/replacement programs, social behavioural marketing, etc.) to reduce consumption during

Research objectives

As shown in Table 1, previous studies have revealed that showering is a major end use component representing around one third of the indoor consumption, and a significant contributor to both residential energy demand and resulting GHG emissions. Furthermore, the shower is one of the discretionary end uses from which residential households have the greatest potential to conserve water (Bonnet et al., 2002, Stewart et al., 2011, Willis et al., 2010a, Willis et al., 2011a). Therefore, a greater

Research design

To achieve such comprehensive study objectives, a mixed method research design has been applied using both quantitative and qualitative approaches to obtain and analyse water end use data. This complex design allows the use of multiple methods to address research objectives (Creswell and Plano Clark, 2007). This mixed approach is adopted in data collection through collecting quantitative natural science data in the form of end use water consumption data, quantitative stock inventory data,

Research method

For the purpose of this study, all factors presented in Table 2 were classified as categorical variables. In other words, each variable is composed of mutually exclusive categories. For instance, as shown in Table 2, the household size characteristic labelled number of adults (A) is composed of households with one adult (1A), two adults (2A) and three adults or more (3A+). To achieve the objectives of this study, a series of one-way independent ANOVA extended into a set of multiple regression

Data analysis and results

Flow trace end use event disaggregation for the SEQREUS resulted in an average total indoor water consumption of 335.9 L per household per day (L/hh/d) for the sampled 200 houses over a 2-week data collection period and average occupancy of 2.6 persons per household. This represents an average per capita indoor consumption of 129.2 L/p/d. Fig. 5 illustrates that the shower end use category is the largest portion of indoor consumption with an average of 111 L/hh/d or 42.7 L/p/d representing 33%

Conclusion

A mixed method research design was applied to collect both quantitative and qualitative data from over 200 households in SEQ, Australia. This design required the implementation of a range of collection approaches, including smart metering technology, questionnaire surveys, diaries, and household water stock inventory audits. All such data collection requirements were essential in order to accurately disaggregate residential meter flow data into each and every water end use event. The

Study implications

Given that showering is often reported as the highest indoor consumption category, and that shower end use event volumes and frequency, are generally much higher than is required for sanitary purposes (i.e. showering is often considered as a leisure activity), this water end use category has the potential to be substantially reduced in drought periods. In such periods, or as a core long-term water conservation measure of the community, the herein described study findings can assist water

Future work

The next stage of this investigation is to follow a similar research method to that described herein to reveal the significant determinants of all other indoor end use categories (e.g. toilet, tap, bathtub, clothes washing and dishwashing). Moreover, a modularised micro-component forecasting model will be built for each of these end uses combining significant predictors of that particular end use category. The summation of all end use predictions from such complex models can provide an

Limitations and future research directions

Water end use studies using high resolution smart metering technology is costly and time consuming, thereby prohibiting large and widespread sample sizes. Nonetheless, the cost of this technology will reduce over time and enable larger samples to be examined over longer time periods; thereby enhancing the statistical power of the forecasting model. For instance, sample size constricted the number of dummy coded determinant categories and limited the level of detail that could be explored (i.e.

Acknowledgements

This research utilises data collected by the SEQREUS team based at Griffith University and funded by the Urban Water Research Security Alliance (http://urbanwateralliance.org.au/). Assistant Professor Michael Steele from Bond University, Australia is also acknowledged for his invaluable advice on statistical methods.

References (63)

  • Aquacraft

    Trace Wizard® Software Version 4.1., 1995–2010

    (2010)
  • A. Athuraliya et al.

    Yarra Valley Water 2007 Appliance Stock and Usage Patterns Survey

    (May 2008)
  • Barthelemy, O.T., 2006. Untangling Scenario Components with Agent Based Modelling, PhD, Manchester Metropolitan...
  • C. Beal et al.

    South East Queensland Residential End Use Study: Baseline Results – Winter 2010. Technical Report No. 31 for Urban Water Security Research Alliance

    (2010)
  • C. Beal et al.

    SEQ residential end use study

    Journal of the Australian Water Association

    (2011)
  • W.D. Berry

    Understanding regression assumptions, Sage University paper series on quantitative applications in social sciences, 07–092

    (1993)
  • E. Blokker et al.

    Simulating residential water demand with a stochastic end-use model

    Journal of Water Resources, Planning and Management

    (2010)
  • B.L. Bowerman et al.

    Linear statistical models: An applied approach

    (1990)
  • J. Cohen

    Multiple regression as a general data-analytic system

    Psychological Bulletin

    (1968)
  • Water Efficiency Labelling and Standards Scheme

    (2011)
  • Water Efficiency Labelling and Standards Scheme: WELS Products

    (2011)
  • J. Creasey et al.

    Microcomponent-based Forecasting for the AMP5 WRP: Final Report (August 2007)

    (2007)
  • J.W. Creswell et al.

    Designing and Conducting Mixed Methods Research

    (2007)
  • W.D. DeOreo et al.

    Retrofit realities

    American Water Works Association Journal

    (2001)
  • H.P. Duncan et al.

    A Stochastic Demand Generator for Domestic Water Use. Proceedings of Water Down Under 2008

    (2008)
  • J. Durbin et al.

    Testing for serial correlation in least squares regression, II

    Biometrika

    (1951)
  • A.P. Field

    Discovering statistics using SPSS

    (2009)
  • Gato, S., 2006. Forecasting urban residential water demand, PhD, RMIT...
  • D. Giurco et al.

    Developing industrial water reuse synergies in Port Melbourne: cost effectiveness, barriers and opportunities

    Journal of Cleaner Production

    (2010)
  • M.A. Hardy

    Regression with Dummy Variables, Sage University Paper series on Quantitative Applications in Social Sciences, 07–093

    (1993)
  • M. Heinrich

    Water End Use and Efficiency Project (WEEP) - Final Report, BRANZ Study Report 159

    (2007)
  • Cited by (0)

    View full text