You are on page 1of 13

Environmental Modelling & Software 23 (2008) 1448–1460

Contents lists available at ScienceDirect

Environmental Modelling & Software


journal homepage: www.elsevier.com/locate/envsoft

Decision support for sustainable option selection in integrated


urban water management
C.K. Makropoulos a, *, K. Natsis b, S. Liu a, K. Mittas c, D. Butler a
a
Centre for Water Systems, School of Engineering, Computing and Mathematics, University of Exeter, Exeter, Devon EX4 4QF, UK
b
Veolia Water Solutions & Technologies, France
c
Environment Agency, UK

a r t i c l e i n f o a b s t r a c t

Article history: Conventional urban water management practices aim to meet water supply-demands while conveying
Received 14 February 2007 wastewater and stormwater away from urban settings. Alternative approaches which consider water
Received in revised form 11 April 2008 demands to be manageable and wastewater and stormwater as valuable resources, although being
Accepted 13 April 2008
increasingly sought, lack reliable site specific implementation methodologies. This paper describes the
Available online 3 June 2008
development of a decision support tool (termed the Urban Water Optioneering Tool (UWOT)) to
facilitate the selection of combinations of water saving strategies and technologies and to support the
Keywords:
delivery of integrated, sustainable water management for new developments. The tool is based on
Decision support
Option selection
a water balance model which allows the investigation of interactions between the major urban water
Sustainability assessment cycle streams. The model is informed by a knowledge library which is populated with technological
Urban water management options and information on their major characteristics and performance. The technology selection is
Water cycle modelling driven by a GA algorithm allowing efficient exploration of the decision space. Quantitative and quali-
tative sustainability criteria and indicators are used to compare between alternative composite water
management strategies while preserving the multiobjective nature of the problem. The tool has been
successfully tested on a case study site in the UK, and the results are presented and discussed. It is
suggested that ‘‘optioneering’’ tools will increasingly become part of urban water management planning
toolkits as practice moves towards more decentralised, integrated, context-specific solutions to address
issues of sustainability.
Ó 2008 Elsevier Ltd. All rights reserved.

Software information Minister (2002), predicting the need for over 4 million new
households by 2016. This concentration of human activities in-
Name of software: UWOT tensifies local competition for all types of resource, with water
Software platform: Simulink/Matlab/GA and Fuzzy Logic amongst the most vital (Zoppou, 2001), rendering it increasingly
Toolboxes difficult to find and utilize new sources of water necessary to satisfy
Type: Research tool growing water demand (Niemczynowicz, 1999). It is argued that
For information on availability contact: Centre for Water water management solutions for new residential developments
Systems, University of Exeter, EX4 4QF: www.ex.ac.uk/ should be based on sustainability considerations due to their far
cws reaching social, economic and environmental implications (Fenner
et al., 2006; Makropoulos et al., 2006a,b). The application of sus-
1. Introduction tainable practices within the urban environment is dependent on
a number of development-specific characteristics requiring both
Demographic change (including internal and international mi- quantitative and qualitative information to be fully taken into ac-
gration) and economic growth considerations are influencing much count (Makropoulos et al., 1999). This necessitates the development
of the urbanisation processes, both in Europe and elsewhere of appropriate decision support tools to help decision-makers to
(Cheshire and Carbonaro, 1996; Skeldon, 2006). This has significant confront the ill-structured nature of the sustainability assessment
implications for current UK policy, with the Office of Deputy Prime process, examine its components and their interactions, allow for
explicit statement of assumptions and enhance the communication
between stakeholders, including researchers (Jakeman et al., 2006).
* Corresponding author. Tel.: þ441392 264049. Within this work the classic view of Sprague and Carlson (1982)
E-mail address: c.makropoulos@exeter.ac.uk (C.K. Makropoulos). was adopted, according to which the role of a DST is to support

1364-8152/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.envsoft.2008.04.010
C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460 1449

rather than replace managerial judgment, while assisting decision- A criterion can be defined as a measure against which option
makers to explore possible solutions and outcomes beyond the performance is assessed along with the degree to which stated
obvious ones (Jakeman et al., 2006). objectives are achieved. Indicators constitute a means of measuring
The work presented here has been developed within the Water the level to which criteria are satisfied (Makropoulos et al.,
Cycle Management for New Developments (WaND) Project in the 2006a,b). The criteria-indicators approach adopted in this work,
UK whose aim is ‘to provide guidelines and decision support tools relates to the framework developed within the SWARD project
for the implementation and assessment of efficient and sustainable which focuses on sustainability for the water industry. Following
water management interventions in new urban developments with the classification introduced in SWARD, four Sustainability Capitals
due consideration to social, environmental and health associated are being used, namely, environmental, economic, social and
factors’. The paper presents the background, development and technical. Each capital contains a set of (quantitative or qualitative)
application of a decision support tool, termed the Urban Water indicators enabling assessment. A subset of the SWARD indicators
Optioneering Tool (UWOT) designed to facilitate the strategic was chosen and implemented in this study, based on their rele-
planning of sustainable water management systems, and is struc- vance to the scope of UWOT and the extent to which they could be
tured as follows: the first section deals with sustainability objec- satisfactorily quantified. This does not imply that other (non-
tives and the use of sustainability indicators as a means of quantifiable) indicators should be ignored, but only that such in-
supporting sustainable option selection. The following section dicators cannot, by definition, be used in a quantitative tool such as
deals with the urban water cycle, its distinct components and UWOT and need to be assessed, within the decision-making pro-
provides examples of urban water management (UWM) technolo- cess, in addition to UWOT’s results. The criteria and indicators
gies, included in the tool. The third section describes in detail the adopted in this study are discussed in brief in the following section
generic structure of the water mass balance model, developed in and are summarised in Table 1. For detailed definitions of these
Simulink, which forms the core of the DST as well as the Technology criteria and indicators the reader is referred to output from the
Library which contains the different technological options for SWARD Project (Ashley et al., 2004).
UWM and their key characteristics. The implementation of a ge-
netic algorithm driving the optimisation process embedded in the 2.2. Option selection
tool is then described and the results of the application of UWOT in
a new residential development are presented and discussed. The The methodology developed in this study, supported by the
paper concludes by discussing the main strengths of the tool and its prototype UWOT software, deals with a specific type of urban water
potential role in the urban water management domain. management problem: the (technological) option selection prob-
lem. It relates to the decision problem of choosing the most suitable
2. Sustainability of the urban water environment technology given case-specific conditions and constraints with re-
spect to technical, environmental, social and economic objectives
Water management in new residential developments is tasked (Balkema et al., 2002).
with the requirement to achieve acceptable levels of service, not
overburden existing infrastructure, have minimal impacts on the 3. Modelling the urban water cycle
natural environment and to be, at the same time, socially and
economically acceptable. In turn, the whole of the urban water The three main components of the urban water cycle are water
management cycle forms part of the broader water resources supply, wastewater disposal and stormwater drainage, making up
management cycle, currently influenced by (in essence) sustain- the urban water system. The traditional approach is to consider the
ability requirements embedded in environmental legislation in the
EU (for example, Directive 2000/60/EC). The emerging EU envi-
ronmental legislation paradigm internalises principles of in- Table 1
Capital, criteria and indicators used in UWOT for the assessment of sustainability
tegrated water management, pollution prevention and a more
complete concept of sustainability which, as stated by Brunner and Capital Criteria Indicator
Starkl (2004), not only features ecological or economic aspects, but Environmental Resource utilisation Water usage (litres/use)
also social and institutional ones, such as the participation of Water loss (litres/use)
Energy use (KWh/use)
stakeholders.
Chemical use (litres/use)
Land use (m2)
Service provision Service provisiona
2.1. Sustainability criteria and indicators
Environmental impact Environmental impacta

Economic Life cycle costs Life cycle costsa


The incorporation of sustainability assessment into decision-
Willingness to pay Willingness to paya
making processes is thus becoming a key task for water service Affordability Affordabilitya
providers – in the UK and elsewhere (Foxon et al., 2002). Quanti- Financial risk exposure Financial risk exposurea
fying or ‘‘operationalising’’ sustainability is, more often than not, Capital cost (£) Capital cost (£)
a debatable concept and an ambiguous process to be handled with Operational cost (£/l) Operational cost (£/litre)
considerable attention to the particular characteristics of the Social Risks to human health Risks to human healtha
problem at hand (Natsis et al., 2006). The approach adopted here is Acceptability Acceptabilitya
the selection of appropriate criteria linked to specific indicators. Participation/responsibility Participation/responsibilitya
Public awareness Public awareness
The assessment of sustainability through the use of sustainability Social inclusion Social inclusiona
indicators is a well established and documented approach,
Technical Performance Performancea
including, but not restricted to work within the Sustainable Water
Reliability Reliabilitya
industry Asset Resource Decisions (SWARD) project (Ashley et al., Durability Durabilitya
2004), the study by the International Hydrological Programme Flexibility/adaptability Flexibility/adaptabilitya
together with the American Society of Civil Engineers (Loucks and a
These indicators do not have units as they quantify qualitative sustainability
Gladwell, 1999), or the work of Balkema et al. (2002) in the stock. Instead they are rated between 0 and 5 with 0 being the worst score and 5 the
Netherlands. best score.
1450 C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460

infrastructure that delivers potable water, separately from the in- Outside supply
frastructure that disposes of wastewater and separately to the 4%
Shower
provision of drainage for stormwater. There is growing need to re- Kitchen sink
5%
evaluate this approach in order to seek ways to, inter alia, minimise 15% Basin
the environmental impact of urban areas on supply sources and 9%
receiving waters. This necessitates the investigation and sub-
sequent exploitation of possible interactions between the three
main components of urban water. Until recently wastewater and
rainwater were considered waste streams that needed to be con-
Bath
veyed away from the urban environment and disposed of. However, 15%
they are increasingly being seen as resources that need to be
exploited rather than unavoidable by-products of urbanisation.
WC
Planning and operating the systems separately, foregoes the Dishwasher
31%
potential for utilisation of urban stormwater and wastewater for 1%
beneficial purposes, and disregards the fact that these systems are
inter-connected and inter-related. Changing one flow can induce Washing
a reaction in the other systems. Potable water demand, for example, machine
is directly related to wastewater production. Technologies that 20%
reduce water demand would influence wastewater conveyance
Fig. 1. Water consumption share of different micro-components of the household in
systems. Another obvious example would be that water recycling
the industrialised world (Source: Memon and Butler, 2006).
techniques, apart from reducing water demand also reduce
demand for sewage and sewage treatment. There emerges there-
fore, a need to take a more holistic view, allowing water supply,
greywater used for purposes other than drinking, e.g., toilet
wastewater disposal, and stormwater drainage to be considered as
flushing and garden irrigation, reduces demand and thus assists the
interacting components within a single system (Mitchell et al.,
preservation of valuable water resources (Nolde, 2000).
2001). An integrated modelling framework is required, in order to
The main issues, relevant to the applicability of greywater sys-
investigate and quantify the interactions and transformations of
tems include social acceptability and water quality. A freshly pro-
the three water flows and to identify the future possibilities and the
duced greywater usually does not have any objectionable odour.
limitations of different systems within the context of sustainable
However, it requires early treatment after collection. If stored un-
water management for new developments. Prior to describing the
treated for long periods, oxygen deficient conditions will develop
structure of the UWOT prototype developed to address this need
and scum will be formed that can float or sink in the collection tank
and analysing the functionality of its various components, a brief
(Memon and Butler, 2005). Moreover, the bacterial population
overview of five streams of urban water, that are distinguished by
tends to increase with increased storage duration (Dixon et al.,
UWOT, is provided. These are: potable (or whitewater), greywater,
1999). In general, treating greywater prior to recycling is more
treated greywater (or greenwater), wastewater (or blackwater) and
socially acceptable, and renders it suitable for more uses. It should
runoff. Note that these do not correspond necessarily to the three
be noted, however, that water quality standards for in-house
urban water components (water supply, wastewater disposal and
greywater use (or indeed greenwater – see below) have yet to be
stormwater drainage) discussed above as they are introduced ex-
defined in the UK.
plicitly to allow for the conceptualisation of interactions between
these components.
3.3. Greenwater
3.1. Potable water
Greenwater is a term used in this work to denote treated rain-
Potable water is by definition water whose quality meets drink- water and treated greywater.
ing water standards. Treating water to potable standards is an
expensive and energy consuming process. However, as can be seen  Rainwater usually carries small pollutant loads (depended inter
from Fig. 1, only a small proportion (appr. 15–20%) of in-house water alia on location, roof building materials and collection system
demand is actually used for purposes requiring drinking water construction) and its harvesting system consists of three basic
quality (incl. water used for drinking, cooking and cleaning dishes). elements: the collection system, the conveyance system and
Water consumption patterns can vary significantly from house to the storage system. The main disadvantage is the unpredictable
house, depending on the household occupancy, the social and cul- and often irregular supply which results in large storage space
tural conditions as well as on the type of the water consuming requirements (Dixon et al., 1999). Light treatment and disin-
appliances installed in the houses (Butler and Davies, 2004). fection is generally adequate for rainwater treatment to non-
potable standards.
3.2. Greywater  Greywater requires more treatment than rainwater to reach an
acceptable standard. The level of treatment required depends
Greywater is the dilute wastewater stream originating from on the scale and purpose of use. At small scale, a two-stage
domestic activities such as showering, bathing, washing hands, treatment consisting of filtration of coarse pollutants (hair and
tooth brushing, dishwashing, washing clothes, cleaning and food suspended impurities) followed by disinfection with chlorine,
preparation. The water contains some organic material, for exam- bromine or UV may be sufficient (Memon and Butler, 2005).
ple, food remains, with pathogens, and inorganic material, such as Greywater recycling at medium to large scales may be more
detergents, sand and salt (Balkema, 2003). The advantage of reus- viable but requires more complex treatment. Options include
ing greywater is that the supply is regular and not dependent on biological aerated filters, membranes, bioreactors, UV treat-
external phenomena (such as rain). As a result, the storage space ment, Titanium Dioxide (TiO2) dosing, membrane aeration
required could be substantially smaller than in the case of rain- bioreactors and coagulation/flocculation with alum or ferric
water systems. Moreover, the substitution of potable water with (Memon and Butler, 2005).
C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460 1451

Greenwater, regardless of the scale of the recycling scheme or beginning of the programming process, UWOT followed the Argent
origin, could be a viable alternative water supply, and can poten- and Grayson (2001) paradigm in assuming that the availability of
tially substitute potable water in some water uses within the house, an initial system to react to future needs and be improved upon,
with the obvious exception of drinking water or food preparation. may actually overcome the users’ initial difficulties to interact with
Studies have shown (Nolde, 2000) that service (green) water made the system and thus increase its usefulness.
available from rainwater or greywater systems can be cost effective
and with proper operation presents no hygienic risk or comfort loss 4.1. The tool platform
for the consumer. The unavailability of water quality standards for
greenwater in-house use, however, is an obvious barrier to its The initial idea during the planning stage of UWOT was to de-
adoption as an alternative form of supply, at least in the UK. velop it based on simple and commonly used software platforms,
such as Microsoft Excel, which is widely available and familiar to
3.4. Wastewater most professionals in the field. However, spreadsheets have limi-
tations, the most serious of which being their inability to handle
In conventional water supply and drainage systems, the whole complex decision-making algorithms, their computational impre-
volume of delivered drinking water becomes wastewater requiring cision and their limited database management capability (Liu et al.,
treatment before being released to the environment. Increasing 2005). Liu and colleagues, suggest that computational capacity and
water demand also means that additional wastewater will have to sophistication can be readily added to a spreadsheet by linking it to
be disposed of, often in areas where existing sewage treatment one of the less widely available, but well known, high-level tech-
works may be at capacity or in areas where discharge potential to nical computing environments such as MATLAB by Mathworks.2
local watercourse is limited by environmental concerns. If recycling The development environment selected for the water mass
schemes are implemented the volume of wastewater produced can balance model, which constitutes the core of UWOT, is Simulink,
be significantly reduced. Within UWOT, greywater is considered which forms part of the MATLAB suite of tools developed by
separate from wastewater, the latter only used to describe water Mathworks. Simulink enables the modelling, simulation, and
from the toilet that leaves the household water cycle to be treated analysis of dynamic systems, whose outputs change over time.
at a central scale. However, if a no recycling setup is modelled Simulink was chosen as it facilitates the representation of processes
within UWOT, then the greywater is added to the wastewater and flows in a visually intuitive manner.
(blackwater) stream. UWOT integrates Simulink/MATLAB and Microsoft Excel (via
Excel’s Visual Basic for Applications (VBA) programming environ-
3.5. Runoff ment) into a computationally strong and user-friendly decision
support tool (Fig. 2). Excel is used for data input, storage of tech-
New developments have a direct impact on existing drainage nology options characteristics and results visualisation. MATLAB/
infrastructure and the surrounding environment (Butler and Mak- Simulink computes the water mass balance model, for the partic-
simovic, 2001). They increase the area of paved surfaces, thus ular combination of technological options selected for evaluation
reducing infiltration, while causing surface runoff to exhibit higher by the user and processes its outputs in a form suitable for sus-
peak flows, larger volumes, shorter times to peak and accelerated tainability assessment, optimisation and visualisation, finally pre-
transport of pollutants and sediment from urban areas. (Niemczy- sented in Excel. The details of the mass balance model and its
nowicz, 1999; Makropoulos et al., 1999). This results in pollution of interaction with the technology options at different spatial and
the receiving watercourses and increased flood risk within the temporal scales are presented below.
development. Controlling surface runoff thus becomes a key ele-
ment in working towards urban sustainability. Surface runoff can 4.2. The issue of scale
be reduced either by collecting rainwater for recycling (through
harvesting1) – with the additional advantage of the reduced pota- One of the main research and policy questions in the field of
ble water demand – or by the installation of non-piped solutions to UWM today, at least in the UK (Butler and Makropoulos, 2006) is
urban drainage (known as ‘‘sustainable drainage systems’’ (SUDS) whether it is possible to attain more sustainable urban water
in the UK (Butler and Parkinson, 1997; Butler and Davies, 2004; management through improving the existing centralised systems
Makropoulos et al., 2006a,b)). or whether it is necessary to switch to new decentralised systems
or is there an optimal mix of decentralised and centralised man-
agement that can combine the advantages of both systems,
4. Building a decision support tool
depending on the situation at hand?
UWOT was designed to enable the comparison of a wide variety
UWOT follows from extensive research carried out by the
of water management technologies and schemes (including recy-
authors in sustainable urban water management, decision-making
cling, treatment and drainage) at different scales.
and computer modelling (e.g. Makropoulos et al., 2006a,b; Natsis
Within the urban water cycle, water quantities are continuously
et al., 2006; Mounce et al., 2006). A previous attempt towards the
exchanged between spatial scales (e.g. household, development) in
development of a tool with similar functionality – but of different
the form of potable water supply and generated wastewater, while,
structure and rationale – was carried out by Sakellari et al. (2005),
within the household, water is used to satisfy various types of de-
within the framework of the WaND initiative. At that stage issues
mand (Sakellari et al., 2005). The appropriate spatial resolution for
of modularity and upgradeability were identified as areas of
modelling the urban water cycle, is therefore an integration of
improvement and UWOT was developed with an aim to maximise
distinct spatial layers that simulate the different scales, which,
the potential of the system to be upgraded and improved upon in
when combined, form the totality of the urban water development.
the future, without significant re-programming effort requirements.
In UWOT the most detailed level of modelling is the in-house
Although this modularity imposes a greater work load at the
micro-component, followed by the household and finally, the
development level.
1
Although this is not the primary function of rainwater harvesting systems and
their operational objectives can be in conflict with those of rainfall-runoff
2
management. See also: www.mathworks.com.
1452 C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460

MATLAB/SIMULINK EXCEL
Boundary
Conditions
and User
preferences
at a given
time and
place

Tech
Library to
select and
evaluate
tech.
options

OR

POTABLE WATER DEMAND Scenario


RELIABILITY WASTE WATER Evaluation
PRODUCED
OPTIONEERING ASSESSMENT
Through an of user-defined RISKS TO RUNOFF
HUMAN HEALTH
EVOLUTION options.
STRATEGY
SOCIAL ENERGY
ACCEPTABILITY

CAPITAL COST LAND USE

OPERATIONAL COST

Fig. 2. Schematic overview of the UWOT system.

1. The micro-component represents a single water using appli- space. It is the spatial scale at which centralised treatment and
ance, for example, a washing machine, a shower, or a household recycling schemes, as well as centralised SUDS. The water
scale greywater treatment system. Modelling separately all ap- balance model inputs at this level are rainfall and potable water
pliances that use water, is fundamental to the modelling concept from the WSP while the only outputs are runoff and waste-
of UWOT, since each appliance is considered a technological water produced. Water evaporated or infiltrated at the de-
option (either selected by the user or proposed by the optimi- velopment scale, constituting a sink in the water balance at this
sation algorithm) with specific characteristics affecting both the scale, is calculated and subtracted from the water balance but
water cycle and the subsequent sustainability assessment. All not modelled in detail.
micro-component options are contained, as suggested before in
a library which includes multiple technological options for each In terms of spatial variability, the UWOT prototype assumes
water use within the household (e.g. multiple types of washing uniformity within the development and does not attempt to model
machines, each with specific characteristics). physical processes. Spatial variability is represented by exploiting
2. The next modelling scale is the household which represents the multiple household type functionality and the associated dif-
a single building, which can have private, commercial or public ferences in characteristics that can be assigned to these types by the
function. The household represents an aggregation of the mi- user (including percent of impervious cover etc). The most spatially
cro-components it contains. The UWOT prototype developed dependant inputs to the model are the rainfall data (in the form of
can model six types of households depending on the specific time-step-average rainfall) and the evapotranspiration potential.
configuration of appliances and user practices (e.g. frequency of Since the model is run at large time-steps (see discussion below),
use) within them. Inputs to the household are (a) rainfall falling no account is taken of the hydraulics of flows within and between
on the impervious and pervious areas of the plot, (b) potable households.
water from the water service provider (WSP) and treated
greenwater water if a centralised recycling scheme is installed. 4.3. Temporal resolution
The basic outputs of this scale are wastewater and runoff. In the
case of centralised recycling schemes, greywater and/or har- The scope of UWOT is to facilitate the strategic planning and
vested rainwater are also exchanged between the household preliminary design of sustainable water management systems,
and the development scales. rather than to provide support for detailed design and imple-
3. The development represents a group of households, possibly of mentation of the technologies. A daily time-step was therefore
different types, as well as the associated roads and public open selected as appropriate for the water balance model. This level of
C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460 1453

aggregation results in the loss of diurnal variations in flows, but was this is not a straightforward process, but it is important to stress
considered adequate to provide sufficient information for the com- that the tool is flexible enough to ignore any indicators for which
parison of different water management strategies and technologies such information is unavailable, unreliable or irrelevant. Apart
especially for long-term simulations, necessary for sustainability from the sustainability indicators, the technology library also con-
assessment. tains operational parameters, including technical and operational
characteristics for each technology that are necessary for the calcu-
4.4. The technology library: a collection of urban water lation of the water balance for the total modelled urban water cycle
management technologies whenever a specific technology is selected for use (e.g. water use per
flush for a specific type of toilet). The technology library currently
UWOT aims to facilitate the comparison of different water comprises 14 separate spreadsheets. Table 2 presents the different
management technologies and their combination at different technology categories and classifies them into different application
application scales. In order to achieve this aim, a knowledge base scales. Each technology spreadsheet includes information on a series
of existing water management technologies was developed and of technologies of the same technology type.
populated. This knowledge base, termed ‘‘Technology Library’’ was
implemented in multiple Excel spreadsheets, one for each category 4.5. The Simulink model
of technological option (e.g. baths, showers, greywater treatment
systems etc) and contains data and information on their major The Simulink water balance model is at the core of UWOT and
characteristics and performance. The information included in the was created to investigate the effect of selecting different combi-
library is based on the environmental, economic, social and tech- nations of technologies (at various scales) on the water cycle of the
nical indicators introduced in earlier sections. Some of these in- development. The Simulink model consists of distinct, (drag-n-
dicators are quantitative (e.g. installation costs for a particular tap) drop) components. This is facilitated by the Simulink modelling
while others are qualitative and require input in the form of units called ‘‘blocks’’. By developing and linking together various
judgement for the particular situation for which the tool is used independent blocks, a powerful yet modular application has been
(e.g. social acceptability of a particular greywater recycling scheme built minimising the effort required for future model extension.
in a particular development). This judgement can be generated by
domain experts, or (more appropriately) by stakeholder engage- 4.5.1. Individual blocks and model components
ment and consultation. The process of generating context-specific In Simulink, a ‘‘block’’ is an independent unit (an object) which
indicators for specific technologies at a case-by-case basis does not can simulate a process by calculating a set of internal and/or ex-
fall within the subject matter of this paper. It is acknowledged that ternal functions, including MATLAB M-files or Simulink S-files (Liu

Table 2
Technology types and scales of use in the Technology Library

Technology type Comment Technology use at:

In-house scale Household scale Dev. scale


Washing machine A number of washing machines from different manufacturers X
with diverse specifications.
Toilet Different types of toilet cisterns and toilet technologies. Distinction X
between different types is mainly based on flushing water consumption.
Shower Different types of shower heads, which can deliver various flow rates. X
Distinction between different technologies is mainly based on water
consumption and user satisfaction.
Bath Different types of bath tabs. Distinction is mainly based on bath tab X
holding capacity.
Hand basin Different types of hand basin tabs. Distinction is based on water flow X
delivery (laminated or aerated flow) and technique diversion to achieve
certain water flow.
Dishwasher A number of dishwashers from different manufacturers with X
diverse specifications.
Kitchen sink Different types of kitchen sink tabs. Distinction is based on water flow X
delivery (laminated or aerated flow) and technique diversion to achieve
certain water flow.
Garden Techniques and devices used for garden watering. X
Outside use Devices and features for domestic outside use, other than garden X
watering devices. For example, water tanks, pumps, swimming
pools and water fountains.
SUDS local Sustainable Drainage Systems (SUDS) which can be used at a local scale X
to serve a property or a small number of properties. Differentiation
between SUDS is mainly based on functionality.
SUDS centralised SUDS which can be used at large scales to serve a region or a catchment X
area. Differentiation between SUDS is mainly based on functionality.
Greywater
Treatment local Decentralised rainwater harvesting and greywater reuse systems at the X X
household level or serve a small number of properties. Differentiation
between the systems is based on design specifications.
Greywater treatment centralised Greywater treatment and potential recycling at regional or X
catchment level. Technologies mainly include different types of
wastewater treatment plants.
Rainwater Treatment Centralised Rainwater treatment and potential recycling at regional or catchment X
level. Technologies mainly include different types of rainwater
treatment plants.
1454 C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460

Influent quantity
a model development perspective to facilitate understanding of
Effluent quantity
Water Cycle how the model is constructed.
Component
Influent quality Effluent quality
4.5.3. The household composite block
Fig. 3. A component of the household water cycle represented by a Simulink block The household block, which can represent any type of building,
(adapted from Liu et al., 2005). is created by aggregating several components and modelling their
dynamic interactions. The block is divided into four subsystems:
et al., 2005). For the water cycle at the household scale, the ele- source, allocation, water use and switches/multipliers. The source
ments of the system, such as a water source (e.g. water supplied subsystem handles the integration of the water source compo-
from external WSPs), water use appliances (e.g. washing machine, nents. The output from the source subsystem represents the total
toilet and shower) and treatment units (e.g. local greywater recy- water available in the house separated into water quality volumes
cling system) are all viewed as objects that encapsulate specific (white, green, grey and black) and is (re)calculated at each time-
attributes and behaviours and can interact with other objects by step. The water use subsystem contains appliance components and
exchanging water quantity and quality information (see Fig. 3). it is here that the demand for water is generated. The allocation
These objects are used to model the components of the water cycle. subsystem matches supply and demand for different water quali-
A schematic representation of an example configuration for ties based on user-defined preferences and water quality priorities.
a household block can be seen in Fig. 4 to assist in the visualisation This is probably the most complex subsystem and will be presented
of the model description. The figure illustrates the basic idea of the in greater detail in the following paragraphs. The switches/multi-
modelling components presented in the discussion below, but does pliers subsystem defines the existence (or not) of recycling schemes
not represent the detailed configuration of the Simulink blocks for greywater and rainwater harvesting, and is operated on the
developed for UWOT. basis of either user input or optimisation.
For the purposes of model construction, components are dis- The source subsystem is composed of three water sources
tinguished into sources, appliances and sinks. Sources are the ele- (white, green and greywater) that are modelled as water tanks,
ments that ‘‘bring’’ water into the system, such as an (external) with a fixed storage capacity and a variable water level. The green
potable water supply component or a rainwater collection tank. and the greywater tanks represent the actual storage tanks that are
Appliances are all elements that ‘‘use’’ water (e.g. a washing ma- installed in the house to balance supply and demand. The potable
chine). Finally, sinks are the elements that convey water out of the water tank on the other hand is conceptual and is included for
system, such as a wastewater conveyance system. reasons of symmetry of representation and used to model (exter-
nal) potable water supply from the WSP. In principle this could also
4.5.2. Composite blocks be set to some fixed capacity to model water scarcity or rationing,
Several blocks can be grouped according to their functionality to but in the current development stage it has been assigned an
form a new composite block. UWOT uses this functionality to infinite storage capacity.
manage the different scales within the urban water cycle. Below, A crucial issue in modelling the water cycle within the house-
the composite blocks employed in UWOT are presented from hold is the balance between demand for water and its provision.

HOUSEHOLD BLOCK
SOURCE

Potable Water
Tank

WATER USE (Appliances)

Green
Rainfall
Water OUT
IN Green Water
Rainwater Harvesting Tank ALLOCATION

Potable
Water Grey Water
IN OUT Grey Water
Shower Tank

Green
Water
Grey Water SWITCH
IN
Washing Mashine OUT
Grey Water
(Developme
nt Scale)
Grey
Water Black
IN Water OUT Waste
Toilet water

SINK

Fig. 4. Schematic of example components within a household block.


C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460 1455

This includes the allocation of water by the tanks to competing obtain information on their performance and tailor their assess-
water requests and the prioritisation of water uptake by the ment to the particular problem at hand. UWOT can be run in two
appliances between alternative water sources. In UWOT, each modes: (a) assessment and (b) optimisation.
micro-component is associated with a list of different water qualities
that the micro-component can receive in order of (user-defined)
preference. A toilet, for example, may be specified to ‘‘prefer’’ to use 4.7. Assessment mode
greenwater and only if this is not available to accept potable water
for flushing. The tanks on the other hand, supply water using a first In this mode, UWOT evaluates a user-defined configuration of an
come, first serve basis. Since in the Simulink mass balance model all urban water management system (for a specific urban de-
requests would happen simultaneously within one time-step, the velopment). To initiate model simulation, the user needs to specify
randomness of the actual phenomenon3 is simulated by creating through the GUI the micro-components involved in the different
a random water request sequence. Clearly dependencies between types of houses and the different scales by configuring an on/off
appliances could also be built into this random sequence but this has flag, which is associated with each technological option. Moreover,
not been implemented in this version of the UWOT prototype. the user is asked to indicate whether water recycling schemes are
The appliances subsystem contains all the water components to be included, specify the type of water to be treated and recycled
and generates the demand. This is where water quality is altered (rainwater or greywater) and the spatial scale of the recycling’s
based on its use. For example, in the toilet components, water input implementation (household or development scale). Recycling
is transformed to blackwater irrespective of the quality of the input. schemes can be applied independently to each household type, by
The shower components transform all input to greywater, and in configuring each household’s recycling switches. This approach
the treatment components, greywater is transformed to enables the comparison of different combinations of technologies
greenwater. and urban water management strategies.4 Finally, the user is asked
The switches subsystem manages rainwater harvesting (to be to specify the priority by which each type of micro-component
used as a source for greenwater) and greywater recycling. It is demands a particular water quality. Additional user inputs include
responsible for diverting the appropriate water streams according information on the number of houses per household type and their
to user preferences. The switches can be set (by the user) to 0, 1 or 2 corresponding occupancies, as well as data on their pervious, im-
for no recycling, local recycling and centralised recycling, re- pervious and rain harvesting areas at the various scales. Finally, the
spectively. If the rainwater harvesting switch is set to 0, for exam- user is asked to specify the desired duration of the simulation, after
ple, then the rainwater stream will be connected directly to the which UWOT populates the urban water cycle model with the
runoff stream and flow out of the development. If it is set to 1 the specific micro-components suggested by the user, runs the full
rainwater stream will be harvested within the household, and if it simulation processes the results within the MATLAB environment
set to 2, the rainwater stream will be diverted to a centralised and exports them to Excel for visualisation.
rainwater treatment facility. The same applies to the greywater Two types of results are produced by the tool: (a) numerical
streams. values for all water streams, including potable water demand,
wastewater and runoff generated, presented in total, daily and daily
4.5.4. The development composite block per capita values and plotted in bar diagrams to allow for a quick
The development scale is created by an aggregation of house- impression of the water balance of the simulated configuration;
holds. Six different types of household (composite) blocks are (b) a wider assessment created by obtaining values for the sus-
supported in the current version of the model and can be used to tainability indicators in Table 1 for all the technologies present in
represent different types of premises with different occupancy and the simulated configuration, for which the user has provided
water use. Additional components at this scale are roads and public (context-specific) values for the particular situation by entering
open space associated with the development, centralised rainwater them into the technology library. In both cases, the indicator values
and greywater treatment facilities and service reservoirs for the of the proposed solution are standardised through fuzzy inference
greenwater water resulting from these facilities. A schematic rep- systems (FIS) against those of a user-defined benchmark scenario
resentation of the concept can be seen in Fig. 5. Sources of water at (Natsis et al., 2006). The benchmark could be a system configura-
this scale are the potable water supplied by the WSP and rainfall, tion that the user considers as ‘‘business-as-usual’’. Such a config-
while the sinks represented are wastewater and runoff leaving the uration may include, for example, standard appliances, no recycling
development, as well as water that is infiltrated or evaporated. or harvesting and classic end-of-pipe drainage and wastewater
collection technologies (which are also present in the technology
4.6. Running UWOT library). This approach allows the inclusion in the same assessment
process of both quantitative and qualitative criteria with different
Efforts were devoted to designing a user-friendly interface, units and values and physical meanings. Consequently, the per-
motivated by the fact that user-friendly interfaces have been formance of different system configuration scenarios are assessed
reported to help overcome, at least partly, some stakeholder’s re- with respect to a (user-defined) point of reference, making the
luctance towards use of formal decision support systems (Lu et al., assessment context aware. This context-awareness is based on two
2001). The design of the user interface for UWOT considered issues elements: (a) the inclusion in the technology library of case-specific
of accessibility, flexibility, and interactiveness, as discussed in information (through, for example, consultation with local stake-
Malczewski (1999). In particular, the interaction between the user holders) and (b) the fact that the assessment is a ‘‘comparison’’
and the model is handled by the Excel-based Graphical User rather than a ‘‘score’’ between a standard solution, given the cir-
Interface (GUI), which facilitates the use of UWOT by users cumstance, and a potential alternative one. This context dependency
unfamiliar with MATLAB and Simulink. Excel also contains the is an issue of paramount importance in the sustainability assessment
Technology Library spreadsheets, allowing the user to browse debate (Ashley et al., 2008).
through the available water management technological options,

4
UWOT can therefore compare rainwater harvesting to greywater recycling (as
3
The phenomenon in this case being the sequence with which appliances re- strategies) as well as one greywater recycling system with another (as technological
quest water within the household. options).
1456 C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460

DEVELOPMENT
Household (1) Household (2) Household (3)

Rain Water
Black Water

Green Water Grey Water

Potable
Water Runoff

Household (4) Household (5) Household (6)

Infiltration

Fig. 5. A schematic of the development scale model formed by an aggregation of household blocks.

4.8. Optimisation mode The fitness of each solution is calculated on the basis of its sus-
tainability against the indicators identified in Table 1. Since the GA
Genetic algorithms (GA) were used to provide optimisation ca- used in this work is based on minimising a single objective, while
pabilities to UWOT. GAs and evolutionary programming (EP) have the indicators (and consequently the objectives they represent) are
recently seen a rise in their application in the water field (e.g. multiple, the computed values for all indicators, standardised
Farmani et al. (2005), Keedwell and Khu (2005), Makropoulos and against the benchmark scenario using FIS (as in the assessment
Butler (2005)). This is primarily because GAs can be used to effec- mode), are aggregated using two separate weighting schemes.
tively solve large, complex, non-linear, discrete optimisation
problems, such as found in urban water management. A discussion  A simple weighted averaging allows the incorporation of direct
on GAs and their principles falls outside the scope of the present user preferences enabling the user to tailor the search to par-
paper.5 The GA implementation in UWOT was developed using the ticular objectives. For example, allow the optimisation algo-
MATLAB’s GA Toolbox (Mathworks, 2000) and all parameters (ini- rithm to minimise water demand and wastewater production
tial population, mutation and crossover rates as well conversion only.
criteria are set by the user through UWOT’s optimisation part of the  Ordered Weighted Averaging (OWA) (Yager, 1988). This enables
GUI). For a detailed discussion of the algorithm and its setup the the aggregated fitness value to be influenced by the degree of
reader is referred to Natsis (2005). The user specifies the boundary optimism or pessimism of the decision-maker. A comprehen-
conditions of the problem, as in the assessment case (incl. size of sive discussion on this aggregation process in the context of
development, occupancy etc), and is also presented with the option UWOT is included in Natsis et al. (2006).
of pre-specifying a number of technological options that should be
included in the final composite solution, without being subject to Both aggregation procedures can be employed at the same time
selection through optimisation. A solution includes: (a) a set of if the nature of the problem requires it. The algorithm finishes
specific technological options for all micro-components (that are when one of a set of end criteria are satisfied, which include
not pre-specified by the user) and (b) a set of configurations for the incremental fitness improvement and maximum number of itera-
switches, specifying the inclusion (or not) within the solution of tions. Once an optimal solution is reached, the corresponding set of
greywater recycling and/or rainwater harvesting. The algorithm is technological options is identified and its sustainability ratings are
restricted to provide feasible solutions: some configurations are communicated to the decision-maker through the graphical user
linked together (e.g. the choice of the installation of a greywater interface, in their original disaggregated and non-weighted form.
recycling scheme necessitates the existence of a greywater treat- In both the case of assessment and optimisation, no further
ment micro-component), while others are not allowed (e.g. in- formal support is provided to rank the compared solutions, under
appropriate treatment trains) through the use of a look-up table. the assumption that trade-offs between criteria need to be visible
to the decision-maker rather than hidden behind aggregate values.
It is assumed that the results of UWOT will be used as input to
5
For a comprehensive discussion on GA and Evolutionary Programming the a further decision process, including possibly negotiation between
reader is referred to Michalewicz (1996). stakeholders with diverging priorities and agendas, which will be
C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460 1457

aided by a clear display of the assessment of all solutions across showerheads, which are selected (manually) by the user.6 This
a range of sustainability indicators. scenario is realistic as the technologies employed are already
developed and both the market and society are progressively
5. Results and discussion adopting them. Similar to the benchmark scenario, the Water Save
scenario assumes no centralised greywater recycling and rainwater
UWOT was applied to the problem of selecting water-related harvesting and the only source of water is the WSP external supply.
technologies for a new UK urban development, Elvetham Heath The Water Save scenario shows a decrease in demand for
(EH), to test the water balance model, technology library and potable water, indicating a per capita consumption of 109 lcd,
functionality of the decision support interface. UWOT was first which is in agreement with what literature suggests as possible
applied to simulate the present situation in EH and afterwards to following the installation of water saving devices (Styles, 2005;
assess the sustainability of various alternative water management Butler and Makropoulos, 2006).
practices that could be applied in the new households during the A series of quantitative outputs, used for the subsequent as-
development’s planned expansion. sessment of the scenarios discussed above, can be seen in Fig. 6.
Elvetham Heath is a 126 ha development situated in Hamp- The Benchmark scenario values are used to define the setup of
shire in the south east of England. It contains a natural reserve, the FIS that is then used to evaluate the Water Save scenario and are
car parks and 62 ha of a residential area, which houses, approx- always placed at equal distances from the centre of the spider di-
imately, 4300 residents. The public space contains 42 ha of paved agram (Fig. 7). A scenario value lower than the benchmark indicates
surfaces and 20 ha of unpaved surfaces. 1340 households had poorer performance. Similarly, high scenario values are placed
been constructed and occupied by April 2005 and 528 new outside the benchmark values. In this example, we observe that, for
households are planned to be completed by 2008, increasing, the example, indicators ‘‘potable water demand’’ and ‘‘wastewater
total number of residents to 6000. The 1340 households consist produced’’ have improved at the expense of ‘‘energy’’.
of 29 1-bedroom apartments, 378 2-bedroom flats, 395 3-bedroom Although the actual values of these indicators would depend on
flats, 376 4-bedroom flats, 118 5-bedroom flats and 44 6-bedroom the specific technologies used, what can be generally suggested is
flats. Some alternative UWM practices have been successfully that changes in the water cycle, through the introduction of water
implemented in Elvetham Heath in the form of SUDS. The village managing technologies have implications on the entire system and
runoff drainage system includes 21 swales/retention ponds that impact on a series of (sustainability) metrics. This complex, mul-
are successfully operating and also contribute to the aesthetic ticriteria (and ultimately multiobjective) nature of the problem is in
value of the village. fact the raison d’être of UWOT. In Fig. 7, we use the term objectives
A series of scenarios was developed to provide a basis for to refer both to the criteria of the assessment mode and the use of
thinking about alternative development possibilities in the case the same criteria as ‘‘objectives’’ for the optimisation mode.
study area. A discussion of scenarios of possible futures and their An Optimisation scenario, was developed to assess the potential
impact on urban water management is included in Makropoulos of UWOT to identify solutions beyond what was immediately ob-
et al. (2008). In this paper we present only two of the scenarios vious to the end-user (represented here by the Water Save sce-
developed as a proof of concept of the tool’s operation. A bench- nario). As discussed, the optimisation algorithm can handle
mark scenario is also presented, to allow for a common basis for multiobjective problems through weighting. To allow for a direct
comparisons. comparison between the Water Save and the Optimisation sce-
The benchmark scenario, or ‘business as usual’, resembles resi- narios we have chosen to optimise exclusively for potable water
dential developments using conventional, water intensive tech- demand, by setting its simple weighted averaging weight to 1 and
nologies. The scenario uses population and household data specific allowing all other weights to assume the value of 0. An initial
to Elvetham Heath and makes use of conventional technologies at population of 100 individuals (each representing a potential solu-
the household level assuming no rainwater harvesting, greywater tion to option selection process) was allowed to evolve for 30
recycling or SUDS. It also assumes that all water services, i.e. water generations. It can be seen (Fig. 6) that the algorithm, has identified
supply, stormwater and wastewater conveyance and treatment, are a set of technological options that can achieve an even more pro-
delivered using centralised systems. Since rainwater harvesting and nounced reduction in water demand, lowering the per capita con-
greywater recycling is not allowed in this scenario, the only source sumption to 93 lcd.
of water supply is potable water from the WSP (through external An assessment of the optimisation scenario can be seen in Fig. 7.
water mains). Similar to the Water Save scenario, some of the indicators (in this
The mean per capita consumption corresponding to the EH case, for example, ‘‘potable water demand’’, ‘‘wastewater pro-
development under the Benchmark scenario, calculated over duced’’ and ‘‘runoff’’) have improved, while others (such as ‘‘en-
a simulated period of 10 days, is approximately 168 l per capita per ergy’’ or ‘‘operation costs’’) have deteriorated. The effect is more
day (lcd). This is approximately 10% higher than the 150 lcd which pronounced as the optimisation scenario, to achieve a significant
is identified as a typical value for the UK (see also Butler and reduction in potable water demand has employed, in this case,
Makropoulos, 2006). This can be attributed to the fact that the case greywater recycling at the local level. This can be observed through
study has a significant amount of gardens and associated outdoors the indicators ‘‘greenwater used’’ and ‘‘greywater used’’ in Fig. 6. It
water demand. Wastewater generation is approximately 85% of the is not suggested, that these results are necessarily optimal, as
water demand, which is in agreement with generally accepted evolutionary algorithms can only guarantee near optimal solutions,
losses of 5% within the house (Butler and Davies, 2004) augmented and in any case the extent of the search was not sufficient to ade-
by the 10% of demand due to gardening (water used in garden quately cover such a complex solution space. It can be seen, how-
irrigation is not contributing to the measured wastewater flow). ever, that even with a few optimisation runs the tool can suggest
The results provide confidence that the model is well con- solutions that outperform reasonable ‘‘best guesses’’ by end-users.
structed and able to correctly simulate the water cycle within It is expected that as the technology library is updated with new
a development. entries of available technologies thus making the solution space
The Water Save scenario is based on the same base data
(number of houses, occupancy, pervious and impervious areas of
the Elvetham Heath case study), but uses water saving technologies 6
For example, after an inspection of the technologies available in the technology
at the household level; for example, dual WC cisterns and low flow library.
1458 C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460

700000
Optimisation
600000 Water Save
Benchmark
500000
Water (lcd)
400000

300000

200000

100000

0
POTABLE WATER GREEN WATER GREY WATER WASTEWATER RUNOFF WATER
DEMAND USED USED PRODUCED INFILTRATED

Fig. 6. UWOT results for selected indicators for the different scenarios.

more complicated, the gap between optimisation and best guess study. Each number is associated with a specific technology. For
assessment would increase. The results of this example scenario example, technology no. 2 of the Toilet component is a toilet
may not even be desirable from a sustainability perspective, as the equipped with a dual valve 4/2 litre flush, whose average water
algorithm was asked to minimise potable water demand to the demand is 2.4 l/use.
expense of all other criteria. It is suggested that a more balanced The presence of greywater recycling in the optimisation sce-
weight selection, would result in less pronounced water demand nario can be seen from Table 3, in the selection of a treatment
reductions, but would achieve this without significant decrease in technology (in this case technology number no. 1 as opposed to no
the performance of other indicators. Such a multiobjective han- treatment (indicated by zero) in the other two scenarios). This se-
dling of the problem could also prove to be less intuitive to the user lection has implications in the selection of other technologies as the
and it is in such situations that the tool’s capabilities would become (significant) reductions in potable water achieved by greywater
more advantageous. recycling allows the algorithm to select suboptimal technologies in
In any case, a main advantage of the proposed tool and meth- other components of the cycle. In the example above, the optimi-
odology is making trade-offs visible and allow decision-makers to sation algorithm suggests washing machine no.3 although it is
understand the interaction between and implications of specific (slightly) inferior in terms of water consumption to washing ma-
technology selection. chine no.2.
Table 3, presents an example of the technologies selected for What is optimised here is the overall performance of the water
one of the 6 possible types of households represented in UWOT, cycle in the development and not necessarily of each household
under the 3 scenarios. Household type 1, in this case, represents type individually. For the special case of one water source (potable
single occupancy households, of which there are 11 at the EH case water) and one criterion (for example, water demand reduction)

Water Save
Benchmark
POTABLE WATER DEMAND
Optimisation 1

RELIABILITY WASTEWATER PRODUCED

KSTO HUMAN HEALTH RUNOFF

OCIAL ACCEPTABILITY ENERGY

CAPITAL COST LAND USE

OPERATIONAL COST

Fig. 7. A comparison of the Water Save and Optimisation scenarios with the Benchmark scenario against multiple sustainability objectives.
C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460 1459

Table 3 (using processes and tools appropriate to each individual


Example of technological options selected for one type of household indicator – not discussed here) to be consistent with the situ-
Water cycle components Household Type 1 ation in a particular case study and/or include new metrics.
Optimisation Benchmark Water save
2. Metrics can be either quantitative or qualitative indicators
standardised through fuzzy inference systems.
Washing machine 3 3 2
Toilet 2 1 2 3. The technology library can be updated with new technologies
Treatment 1 0 0 without any need of updating the tool.
Shower 2 1 1 4. Sustainability assessment is handled as a context-specific issue,
Bath 5 0 2 since the tool always assesses the sustainability of specific sce-
Handbasin 5 3 1
Kitchen sink 4 3 1
narios in comparison to user-specified (case study dependent)
Dish washer 4 5 1 benchmark scenarios.
Garden 3 5 6 5. The decision-maker’s perception of risk can be taken into
Outside use 3 0 0 account in the process through OWA.
SUDS local 2 0 0
6. Advanced optimisation in the form of GAs is used to explore
the complex solution space and suggest alternatives based on
the full range of sustainability metrics and decision-maker
then it can be argued that the optimal solution will always be the preferences.
one that minimises all component solutions (in this case all
household types). When multiple water sources are possible and UWOT was tested using data from the Elvetham Heath resi-
the availability of some sources (such as grey or greenwater) is dential development. The results of the tool’s application reveal
linked to water demand for certain water uses (for example, the use that the modelling approach adopted is valid and that the funda-
of washing machines) then the solutions become non-linear. For mental algorithms for water allocation and the underlying flow
example, better reductions in energy may be achieved by allowing logic between model components are reasonable.
a washing machine type that requires more water provided that it Results indicate that the tool is useful in identifying and pre-
then produces enough greenwater for other uses (for example, senting to the user, trade-offs across a series of sustainability
toilet flushing) that would otherwise request potable water, while indicators resulting from the interaction between urban water
not increasing overall potable water demand.7 UWOT is able to flows. Specifically, the tool allows the investigation of trade-offs
investigate such interactions and come up with insights on these between quantitative indicators (incl. water wastewater, energy,
and other trade-offs – notably between water, energy and land land use etc) and qualitative indicators (incl. health risk and social
(Butler and Makropoulos, 2006). acceptability) in a visually comprehensible manner that facili-
tates group decision-making and negotiation (as also advocated in
6. Conclusions Jakeman et al., 2006), without oversimplifying matters by pre-
senting only aggregated values.
This paper presents a decision support tool (UWOT) supporting Following the application presented in this paper as well as in
the planning of water cycle management for new urban de- Natsis et al. (2006) and Makropoulos et al. (2006a,b) it is suggested
velopments. The core of the tool consists of a component-based, that UWOT can:
flexible water mass balance model, an extensible technology library
and a set of user-defined sustainability metrics. The underlying o assist the investigation and quantification of interactions and
concepts, methodology and software development of the tool are transformations of the three urban water flows within the
discussed and its application to a case study in the UK presented. same modelling framework,
UWOT provides a holistic view of the urban water system within o facilitate the selection of combinations of water managing
a new development, allowing water supply, wastewater collection technologies and
and disposal as well as stormwater drainage to be considered as o ultimately support the delivery of integrated, sustainable water
components of a single modelling framework. management for new developments.
The tool was developed based on the following principles,
which are embedded in its conceptual and methodological As technology changes in response to social, economic and
development. environmental drivers and urban water management solutions
become more customised to address context-specific issues within
o The sustainability discourse requires wide boundaries and both centralised and distributed water management systems, the
consequently a wide range of metrics to allow for a mapping of solution space for informed option selection to manage the water
those boundaries. cycle will become more complex. The emergence of multiple actors
o Sustainability and its assessment must be relevant to the con- (Marquez and Maheepala, 1996) with potentially conflicting views
text at hand and cannot result in the same solution in every and priorities suggest the need for thinking (software and model-
case. ling) platforms that can facilitate discussion and negotiations by the
o Sustainability is essentially an archetypal multiobjective provision of information and the quick assessment of alternative
problem. what-if scenarios (de Kok and Wind, 2003). UWOT, developed
within the WaND initiative, is such a prototype tool able to assist
These principles were operationalised in UWOT in the following the integrated provision of water management solutions taking
way. into account both the interactions between water flows and the
trade-off between multiple qualitative and quantitative sustain-
1. The use of user-specified sustainability metrics: the user can ability criteria.
update the values of the metrics in the technology library
Acknowledgement

7
This is limited by the supply-demand balance for recycled water at a given This work is developed by the ‘Water Cycle Management for
development, which makes UWOT’s assessment case and context-specific. New Developments’ (WaND) project [www.wand.uk.net] funded
1460 C.K. Makropoulos et al. / Environmental Modelling & Software 23 (2008) 1448–1460

under the Engineering & Physical Science Research Council’s Makropoulos, C., Butler, D., 2005. A multi-objective evolutionary programming
approach to the object location spatial analysis and optimisation problem
‘‘Sustainable Urban Environment’’ Programme by EPSRC, UK gov-
within the urban water management domain. Civil Engineering and Environ-
ernment and industrial collaborators. mental Systems 22 (2), 85–108.
Makropoulos, C., Butler, D., Maksimovic, C., 1999. GIS supported evaluation of
source control applicability in urban areas. Water Science and Technology 39
(9), 243–252.
References Makropoulos, C., Liu, S., Natsis, K., Memon, F., Butler, D., 2006a. Supporting the
Choice, Siting and Evaluation of Sustainable Drainage Systems in New Urban
Argent, R.M., Grayson, R.B., 2001. Design of information systems for environmental Developments. Urban Drainage Modelling and Water Sensitive Urban Design
managers: an example using interface prototyping. Environmental Modelling Conference, Melbourne, Australia.
and Software 16 (5), 433–438. Makropoulos, C., Memon, F.A., Shirley-Smith, C., Butler, D., 2008. Futures: an
Ashley, R., Blackwood, D., Butler, D., Jowitt, P., 2004. Sustainable Water Services: exploration of scenarios for sustainable urban water management. Water Policy
a Procedural Guide (SWARD). IWA Publishing, London. 10 (4), 345–373.
Ashley, R., Butler, D., Hurley, L., Memon, F., 2008. Criteria and indicators in Makropoulos, C., Morley, M., Memon, F.A., Butler, D., Savic, D., Ashley, R., 2006b.
delivering sustainable water systems: from Sustainable Water Asset Resource A decision support framework for sustainable urban water planning and
Decisions (SWARD) to Water cycle management for New Developments management in new urban areas. Water Science and Technology 54 (6–7),
(WaND). In: Thevenot, D.R. (Ed.), DayWater: an Adaptive Decision Support 451–458.
System for Urban Stormwater Management. IWA Publishing, London. Malczewski, J., 1999. GIS and Multicriteria Decision Analysis. Wiley and Sons, New
Balkema, A.J., 2003. Sustainable Wastewater Treatment: Developing a Methodology York.
and Selecting Promising Systems, PhD thesis, University of Eindhoven, 122. Marquez, L.O., Maheepala, S., 1996. An object-oriented approach to the integrated
Balkema, A.J., Preisig, H.A., Otterpohl, R., Lambert, F.J.D., 2002. Indicators for the planning of urban development and utility services. Computers, Environment
sustainability assessment of wastewater treatment systems. Urban Water 4 (2), and Urban Systems 20 (4–5), 303–312.
153–161. Mathworks, 2000. Genetic Algorithm and Direct Search Toolbox Manual, MatLab,
Brunner, N., Starkl, M., 2004. Decision aid systems for evaluating sustainability: the Language of Technical Computing.
a critical survey. Environmental Impact Assessment Review 24 (4), 441–469. Memon, F.A., Butler, D., 2005. Greywater reuse in households. In: Encyclopaedia of
Butler, D., Davies, J.W., 2004. Urban Drainage, second edn. Spon Press, London. Water. John Wiley and Sons.
Butler, D., Makropoulos, C., 2006. Sustainable Water Infrastructure: Technological Memon, F.A., Butler, D., 2006. Water consumption trends and demand forecasting
Options and Future Scenarios. UK Environment Agency. 250. techniques. In: Water Demand Management. IWA Publishing, London.
Butler, D., Maksimovic, C., 2001. Interactions with the environment. In: Michalewicz, Z., 1996. Genetic Algorithms þ Data Structures ¼ Evolution Programs.
Maksimovic, C., Tejada-Guibert, J.A. (Eds.), Frontiers in Urban Water Manage- Springer-Verlag, New York.
ment. Deadlock or Hope. IWA Publishing, London. Mitchell, V.G., Mein, R.G., McMahon, T.A., 2001. Modelling the urban water cycle.
Butler, D., Parkinson, J., 1997. Towards sustainable urban drainage. Water Science Environmental Modelling and Software 16 (7), 615–629.
and Technology 35 (9), 53–63. Mounce, S.L., Hurley, R., Ashley, L., Sharp, C., Makropoulos, Morley, M., 2006. The
Cheshire, P., Carbonaro, G., 1996. Urban economic growth in Europe: testing theory Development of a Flexible, Knowledge Based Framework, for Evaluation of the
and policy prescriptions. Urban Studies 33 (7), 1111–1128. Relative Sustainability of Water Cycle Management. Urban Drainage Modelling
Dixon, A., Butler, D., Fewkes, A., 1999. Water saving potential of domestic water and Water Sensitive Urban Design Conference, Melbourne, Australia.
reuse systems using greywater and rainwater in combination. Water Science Natsis, K., 2005. Decision Support for Sustainable Option Selection in
and Technology 39, 25–32. Integrated Urban Water Management, MSc dissertation, Imperial College
Farmani, R., Savic, D.A., Walters, G.A., 2005. Evolutionary multi-objective optimi- London, UK.
zation in water distribution network design. Engineering Optimization 37 (2), Natsis, K., Makropoulos, C., Liu, S., Butler, D., Memon, F., 2006. A fuzzy logic multi
167–183. criteria assessment in urban water management decision support. In: Pro-
Fenner, R.A., Ainger, C.M., Cruickshank, H.J., Guthrie, P.M., 2006. Widening engi- ceedings of the Seventh International Conference On Hydroinformatics, Nice,
neering horizons: addressing the complexity of sustainable development. vol. 4 2725–2732.
Proceedings of the Institution of Civil Engineers Engineering Sustainability 159 Niemczynowicz, J., 1999. Urban hydrology and water management – present and
(ES4), 145–154. future challenges. Urban Water 1 (1), 1–14.
Foxon, T.J., McIlkenny, G., Gilmour, D., Oltean-Dumbrava, C., Souter, N., Ashley, R., Nolde, E., 2000. Greywater reuse systems for toilet flushing in multi-storey
Bulter, D., Pearson, P., Jowitt, P., Moir, J., 2002. Sustainability criteria for decision buildings – over ten years experience in Berlin. Urban Water 1 (4),
support in the UK water industry. Journal of Environmental Planning and 275–284.
Management 45 (2), 285–301. Office of Deputy Prime Minister, 29 July 2002. Projections of Households in England
Jakeman, A.J., Letcher, R.A., Norton, J.P., 2006. Ten iterative steps in development 2021 Available from: http://www.housing.odpm.gov.uk/research/project/01.
and evaluation of environmental models. Environmental Modelling and Soft- htm.
ware 21 (5), 602–614. Sakellari, I., Makropoulos, C., Butler, D., Memon, F., 2005. Modelling sustainable
de Kok, J., Wind, H.G., 2003. Design and application of decision-support systems for urban water management optionsProceedings of ICE. Engineering Sustainability
integrated water management: lessons to be learnt. Physics and Chemistry of 158 (3), 143–153.
the Earth, Parts A/B/C 28 (14–15), 571–578. Skeldon, R., 2006. Interlinkages between internal and international migration
Keedwell, E., Khu, S.T., 2005. A hybrid genetic algorithm for the design of water and development in the Asian region. Population Space and Place 12 (1),
distribution networks. Engineering Applications of Artificial Intelligence 18 (4), 15–30.
461–472. Sprague, R.H., Carlson, E.D., 1982. Building Effective Decision Support Systems.
Liu, S., Butler, D., Makropoulos, C. and Memon, F., 2005. An excel-matlab based Prentice-Hall, Englewood Cliffs, N.J.
decision making framework and its application in urban water management, Styles, M., 2005. Sustainable Communities: Potential Water Savings through
Proceeding CCWI2005 Conference, University of Exeter, Exeter, UK. Available Technologies. UK Environment Agency. Internal discussion paper.
Loucks, D.P., Gladwell, J.S., 1999. Sustainability Criteria for Water Resource Systems. Yager, R.R., 1988. On ordered weighted averaging aggregation operators in multi-
Cambridge University Press, Cambridge. criteria decision-making. IEEE Transactions on Systems, Man and Cybernetics
Lu, H., Yu, H., Lu, S.S.K., 2001. The effects of cognitive style and model type on DSS 18 (1), 183–190.
acceptance: an empirical study. European Journal of Operational Research 131 Zoppou, C., 2001. Review of urban storm water models. Environmental Modelling
(3), 649–663. and Software 16 (3), 195–231.

You might also like