Professional Documents
Culture Documents
Literature Review
2.1 Conventional Urban Stormwater Management System and Its
Limitations
Stormwater management basically means managing the surface runoff. The
urban stormwater has been predominantly managed for the purpose of flood control
(Brown, 2005; Winz et al., 2011). Traditionally, surface runoff from developed areas,
being considered undesired, has been removed as completely and as fast as possible
(Boller, 2004). The aim has been to drain the urban runoff quickly with the help of
channels and pipes. Typically the conventional system comprises road side drains,
channelization, rectification and enlargement of storm drains (Miguez et al., 2014) and
provision of cross drainage works.
In cities, sewer systems may be provided to convey mostly dry weather flow,
coming from domestic and industrial sanitary sewage as well as infiltration flow and
stormwater. Traditionally, in urban drainage two types of sewer systems are adopted:
separate and combined sewers. The former carry dry and wet weather flow separately
into two different networks, while the latter convey dry and wet weather flow together.
In a combined sewer system, stormwater runoff mixes with sanitary flow. If the volume
and rate of stormwater and sanitary flow exceed capacity at wastewater treatment plants
(WWTPs), the combined flow overtops the discharge weirs at regulators, causing
combined sewer overflows (CSOs). If runoff rates exceed the conveyance capacity of
the sewer system, sewer back-ups or street flooding may also occur. This is likely to
create unhygienic conditions.
The sharp rise in urban development has led to an increase in impervious areas
and a decrease in vegetated and pervious surfaces (Leopold, 1968). The conversion of
natural surface into impervious cover like roads, rooftops, and parking lots decreases
infiltration and increases runoff volumes. Conventional urban stormwater
infrastructure, comprising pipes and channels, is provided to remove these excess
flows. The natural stormwater flow paths and runoff velocities get altered due to such
systems (Roesner et al., 2001; USEPA, 1993; Clar et al., 2004; Walsh et al., 2005). As
a result, there is a significant increase in the peak flow rates and frequencies in
urbanized areas, when compared to pre-development conditions (Roesner et al., 2001).
16
The thrust towards such development has led to draught and flooding problems faced
by urbanized areas today (Wong et al., 2000).
The importance of stormwater quality was realized with elevated levels of
stormwater runoff volume (Miguez et al., 2014; Arnold and Gibbons, 1996). Though
the earlier theories considered surface runoff as clean water, the quality of stormwater
may be significantly reduced, causing stream degradation and impacting the estuaries
and bays (Novotny, 1991; Walsh et al., 2005; Wong, 2005; Wong and Eadie, 2000;
Chow et al., 2011; Hatt et al., 2004; Lee et al., 2007). The pollution impact from
secondary treated domestic wastewater may be lesser than that associated with
stormwater (Wanielista et al., 1977). Environmental Protection Agency EPA (2000)
report states the problems of conventional systems which include decrease in
groundwater recharge, increase in runoff volume and change in the timing, frequency
and rate of discharge (Leopold, 1968). These changes can cause flooding, water quality
degradation, stream erosion and the need to construct end of pipe Best Management
Practices (BMPs) (Hollis, 1975; Hall, 1974; EPA, 2000).
The conventional system considers runoff as undesirable water which needs to
be removed from the site immediately for good drainage. This paradigm which
dominated site planning and engineering in the 1960s resulted in the creation of an
efficient system for conveyance of runoff through the network of pipelines typically
discharging it into a creek, river or lake. This resulted in the amplification of the
hydraulic changes due to loss in natural storage, increase in impervious areas, decrease
in runoff travel times and increase in the degree of hydraulic connections (PGDER,
1999). The natural features which decrease travel times and detain or infiltrate runoff
are absent in the developments resulting from conventional site designs.
Traditional practice ignores the importance of stormwater as a precious resource
and considers it as a waste product (Andr′es-Dom′enech et al., 2010; Campbell et al.,
2004). The adverse impacts of the traditional practice of urban stormwater management
like degradation of receiving waterways, combined sewer overflows, diffuse pollution
(Andr′es-Dom′enech et al., 2010; Campbell et al., 2004) have raised growing concerns
about the natural environment. This is also being criticized as an outdated practice
which impedes the implementation of sustainability in stormwater management and
ignores the importance of stormwater as an alternative source of water (Brown, 2005;
Miguez et al., 2014). Urban planners are facing the challenge of balancing of water
shortages and flood risks (Yang and Cui, 2012). The traditional practice of providing
17
additional centralized stormwater infrastructure is employed to cater to the growing
urban development. The implementation of the sustainable stormwater management
model may get hindered by continuing with the conventional practice (Brown et al.,
2009). The traditional metric of stormwater management does not focus on runoff
management at source or prevent or control stormwater pollution. Hatt et al. (2006)
recommend better and sustainable management of the urban water cycle to satisfy
growing water demand without further environmental deterioration. The conventional
approach which includes provision of drainage network and modification of existing
water courses is being complemented or partially substituted by different concepts,
including recycling and reuse of stormwater. The failure of the old paradigm has led to
the development of a new comprehensive, sustainable approach to manage stormwater
known as sustainable urban drainage system (Miguez et al., 2014). The practice of
stormwater management is constantly evolving since then. Stormwater harvesting has
emerged as an important strategy for managing urban water resources to address the
growing concerns of water management throughout the world (Yang & Cui, 2012;
Brodie, 2012).
In the 1960s, the stormwater was managed only for flood prevention, but in
subsequent decades, objectives for stormwater management have diversified to include
quality, ecosystem health, reuse, integration with urban design etc. along with quantity.
There are many issues which today’s stormwater management system has to address,
such as (Roy, et al., 2008; Ellis et al., 2008):
• Quantity: ability of the stormwater system to prevent flooding;
• Quality: provide an improved technology for environmental protection of
receiving waters;
• Ecosystem Health: should have a significant positive effect on stream stability,
habitat structure, base flows and water quality;
• Reuse: its potential usability as a water resource in society;
• Recreation, Aesthetics & other Issues: possibility to enhance the ecological and
aesthetic value of the urban environment;
• Integration with Urban Design: developing a comprehensive stormwater
drainage arrangement closely linked with the Integrated Urban Water System
and;
• Cost: reduce construction and maintenance costs of the stormwater
infrastructure.
18
Thus, stormwater systems should be designed in such a way that the resources would
be used more efficiently. This would in turn require a radical change in the way these
systems have been managed so far. Technologies to reuse water, at source control of
stormwater should be applied wherever practically and economically feasible.
Concurrently, the health of the ecological systems should be maintained by limiting the
pollution in receiving waters and land.
21
2.3.1 Bio-retention
Bio-retention systems utilize a conditioned planting soil bed and planting materials
to filter, infiltrate and store runoff within a shallow depression. Bio-retention can
perform both runoff volume reduction and pollutant filtering functions (USEPA, 2000).
Bio-retention cells filter and cleanse the runoff and discharge the water either in a
storage unit for reuse or into a receiving water body or allow it to get infiltrated into the
soil thus replenishing the groundwater (Beecham, 2003). These systems can be
designed taking into consideration soil types, site conditions and land uses. Typical
components and their functions are listed below (USEPA, 2000).
Grass buffer strips for reducing the velocity of runoff and removing the
particulates.
Sand bed: Provided for aeration and drainage of the planting soil. It also helps
in flushing out the pollutants from the soil.
Shallow surface water ponding area: Provided for storing the excess runoff,
facilitating the settling of particulates and evaporation of excess water.
Soil / organic layer: For filtering pollutants, preventing soil erosion. Provides a
medium for decomposition of organic matter and growth of plants.
Vegetation (Plants): Provided for removing excess water by evapotranspiration
and pollutants through nutrient cycling. Also improve the aesthetics of the area.
Overflow outlet: For removing excess runoff to prevent ponding during rain
events exceeding the design storm.
Under drain: Provided for removing runoff when the soils are unsuitable (low
infiltration rate).
The design components are summarized in Table 2.1 while Figure 2.1 shows the typical
section of a bio-retention area.
22
Table 2.1: Bio-retention Design Components (Source: PGDER, 1999)
Component Specifications
Pre-treatment Area Required where a significant volume of debris or suspended
material is anticipated such as parking lots and commercial
areas. Grass buffer strip or vegetated swales are commonly
used pretreatment devices.
Shallow surface Depth = 6 in.
water ponding area
Ground cover layer Mulch layer 2-3 in. thick
Planting soil / Depth = 4 ft. Recommended soil mixtures include sand, loamy
organic layer sand, and sandy loam. Clay content < 10%
In situ soil Infiltration rate >= 0.5 inches/hour without under-drains
Infiltration rate <= 0.5 inch/hour, under-drain required
Vegetation / plants Native species recommended
Inlet and outlet Non erosive flow velocities (0.5 ft/sec)
controls
23
These cells can reduce the annual stormwater runoff by 85 – 90 % if provided
in appropriate soils and help in replenishing groundwater. The article by Beecham
(2010) reports 90% removal efficiency for suspended solids (TSS), 80% for total
phosphorous (TP) and 40% for total nitrogen (TN). The results presented by Parker et
al. (2009) indicate that the bio-retention systems retained significant volumes (55%) of
stormwater along with reduction in peak flow. The removal efficiencies of TSS, TP and
TN were 79%, 43% and 60% respectively.
Bio-retention cells remove metals (Cd, Zn, Pb, etc.) found in stormwater runoff.
The removal efficiencies are 70-97% for lead, 43–97% for copper and 64–98% for zinc.
They also reduce total Kjeldahl nitrogen by 37-80%, nitrate-nitrogen by up-to 26% and
phosphorus by as much as 87% (USEPA, 2000).
Phosphorous removal from stormwater runoff was investigated by Hsieh et al. (2007b).
The bio-retention columns analyzed in this study showed a decrease in phosphorous
loading and this phosphorous trapped in the media would be subsequently used by the
vegetation through nutrient cycling. In another study by Kim et al. (2012), the removal
potential of Escherichia Coli by bio-retention was investigated. The study results
indicate a direct relationship between E-coli removal and hydraulic retention time.
Longer hydraulic retention times yielded higher E-coli removal.
The influence of planting soil mix design on hydrologic and water quality
performance of bio-retention cells was analyzed by Carpenter et al. (2009). This study
confirms the importance of proper design and construction of bio-retention cells to be
an effective stormwater BMP. The experimentation and analysis shows that the soil
planting mix does affect the performance of bio-retention in reducing the volume and
pollutant concentration. The performance of two grassed bio-retention cells, one
undersized (small) and the other large was analyzed for treatment of highway bridge
deck runoff. The results suggest that 50% removal is possible for a bio-retention cell
undersized by half (Luell et al., 2011). The authors concluded that small bio-retention
cells can be used beneficially for retrofitting in areas where space is limited. The
environmental effectiveness of bio-retention was also studied by Hsieh and Davis
(2005). They studied eighteen bio-retention columns and six existing bio-retention
facilities to evaluate pollutant removal performance and media characteristics. All
columns and on-site facilities demonstrated excellent removal for oil & grease and lead,
and good to moderate removal efficiency for TSS. The removal efficiency of TP ranged
widely and that of nitrate and ammonium was observed to be poor. Based on these
24
results, the authors propose two media profiles as design recommendations in this
study. In another study by Hsieh et al. (2007a), nitrogen removal from layered bio-
retention columns was investigated. It was demonstrated that bio-retention cells with
similar media but different configuration can exhibit different nitrogen fate behaviours.
The poor performance of the studied columns was attributed to the high nitrate leaching
potential of the mulch used in the columns. The authors recommend low available
nitrate content for the mulch / organic matter used in bio-retention. They also propose
a media configuration wherein a less permeable media layer is provided below a more
permeable one to establish nitrification / de-nitrification zones. A similar result was
observed for phosphorous removal using bio-retention columns (Hsieh et al., 2007b).
A high hydraulic conductivity media overlaying one with low hydraulic conductivity
was found to be more efficient in phosphorous removal. According to the results
presented in the research work by Randall (2011), TN and TP may be reduced by up to
53 and 79%, respectively, in specially designed bio-retention gardens.
Bio-retention cells also improve the aesthetics and blend well with the landscape.
Rain Garden
Rain gardens, also called as bio-retention cells are landscaped depressions that
capture and treat stormwater runoff. Rain garden is a popular variant of bio-retention
systems typically provided in a home lawn or parking lot. It is able to blend well into
the landscape and serve as a garden area. The main function of a rain garden is to retain
and treat stormwater from rooftops (Fig. 2.2) and parking lots (Fig. 2.3). Some of the
benefits of providing a rain garden include:
Reduction in stormwater runoff
Slowing down the runoff thus reducing erosion
Improvement in stormwater quality
Recharge of groundwater
Improved and better landscape
25
Fig. 2.2: Rain Garden for Rooftop Rainwater
(Source: U S Dept. of Agriculture, Natural Resources Conservation Service)
26
The soil in the rain garden helps to arrest the pollutants through four processes (Jaber
et al., 2012)
Settling: The runoff slows down after entering the rain garden, and thus the
debris and other impurities settle and are removed. Sediments tend to settle on
the top and may clog it. Thus it requires regular maintenance to prevent ponding.
Chemical reactions in the soil: The pollutants adsorb onto the soil particles and
get filtered out from the runoff.
Biological degradation in root zones: Microorganisms decompose the
impurities and other debris into nutrients and other stable products.
Plant uptake: Plants take up the nutrients released from the microbial
decomposition of impurities.
Two types of rain gardens can be used for stormwater retention.
A planted depression placed downstream of the drainage area: When the native
soil has good infiltration rates, this type of rain garden is provided to retain and
treat runoff from a rooftop.
By replacing existing soil and providing an under-drain: When infiltration rate
of existing soil is less, this type of rain garden is provided. Existing soil is
replaced with layers of soils with high infiltration rates, gravel and mulch. An
under-drain is provided to remove excess runoff which cannot percolate in the
soil. This under-drain discharges it into the stormwater drains.
Design considerations
Maximum drainage area of a rain garden should not exceed 1-2 acres. Flows from larger
drainage areas may cause erosion. In areas with clayey soils, rain garden should be at
least 10 feet away from the buildings to prevent any damage to the foundation.
27
Fig. 2.4: Cross Section of a Typical Rain Garden
(Source: PGDER, 1999)
Sizing of a rain garden
Typically, size of a rain garden varies from 7 – 10 % of the catchment area.
Many states in US have provided on-line design recommendations to help home owners
implement rain gardens. Most of these recommendations are summarized in tabular
format in the article by Jennings et al. (2015). It can be observed that the physical design
varies to a great extent. The area requirement is reported either in terms of the service
(drainage) area or typical range of the required rain garden area. These publications
recommend that the rain garden area may be dependent on the depth. The recommended
garden area may vary from minimum 3% to as high as 80% of the service area or
typically from 50 to 400 sq. ft. This clearly indicates that the design recommendations
need to be verified and revised. The research paper by Jennings et al. (2015) presents
an analysis algorithm for assessing the performance of rain garden in reducing
stormwater runoff. The authors comment that these recommendations are too
conservative and need to be revised. The simulations carried out in their research work
show that 85% reduction in volume is possible by providing rain garden having 10%
drainage / service area when the infiltration rate is 6.35 mm/h. The volume reduction
28
becomes 75% if the infiltration rate is half the earlier value. In another study by Abi
Aad et al. (2009), the reduction in volume was 38% corresponding to the rain garden
area of 3.9% of the drainage area.
It can be concluded that bio-retention systems offer many benefits in terms of
reduction in runoff volume and peak flow, and also can achieve good pollutant removal
efficiencies. These systems blend well with the surroundings and offer an attractive
landscape. Rain gardens represent a popular variant of bio-retention and can be
retrofitted in existing high density urban areas in a de-centralized manner. They also
increase the groundwater recharge opportunities and can replenish depleting ground-
waters, particularly in urban areas.
2.3.4 Swales
Swales are primarily simple drainage and grassed channels provided along the
residential streets and highways to reduce runoff velocity and as an infiltration device
(Fig. 2.6). Two types of grassed swales are being used, the dry swale which provides
both quantity and quality control by facilitating stormwater infiltration, and the wet
swales which reduce peak discharge and treat the runoff before discharging it to a
downstream location. Such swales are effective when the flow depth is minimum and
30
detention time is maximum. The primary pollutant removal mechanism is
sedimentation and secondary mechanisms are infiltration and adsorption. These
channels work well for smaller drainage areas with mildly sloping topography. The cost
of structural stormwater conveyance systems is two to three times more than engineered
grass swales (Design, 1998). Specific concerns regarding the swales include
maintenance problems, nuisance problems which may arise due to the ponding of water
or probable dumping of debris in it and impact on pavement stability. Such concerns
may be alleviated by proper design. The significant maintenance requirements include
periodic removal of sediments and mowing.
31
2.3.5 Rain Barrels and Cisterns
Rain barrels (Fig. 2.7) and cisterns are storage devices that provide above
ground or underground retention storage volume for managing rooftop rainwater. They
are respectively called barrels and cisterns if they provide above ground or underground
storage. Rainwater from the rooftops is directed into the barrels or cisterns using gutters
and down spouts. Filters are provided to treat the first flush and filter the runoff to
prevent clogging by debris. An overflow bypass is provided to bypass runoff from large
storm events. The water is conserved and can be reused for lawns and gardening. This
can reduce water utility costs considerably.
A study was conducted by Steffen et al. (2013) to analyze the performance of
residential rainwater harvesting (RWH) systems (providing cisterns). The objectives
included quantification of water supply and stormwater runoff reduction benefits. The
water saving efficiency benefits were more than 90% for regions with high precipitation
and 20-50% for regions with less precipitation. It was observed that the water supply
benefits depend on several factors including cistern size, water use pattern, and
precipitation. Stormwater runoff reductions ranged from 1-17% depending on the water
use pattern and climatic conditions. Regions with higher precipitation were found to
provide higher water saving efficiency, with a lower stormwater management potential.
The technical feasibility and economic viability of domestic RWH systems was
assessed in the context of Portugal by Silva et al. (2015). The efficiencies of RWH and
water saving were analyzed and payback period was estimated. The authors propose a
simple rule for estimating optimum tank capacity for single-family households in
Portugal. Another paper by Alam et al. (2012) explores the possibility of harvesting
rainwater in rural communities and thickly populated areas of Sylhet as an alternative
source of drinking water. The authors conclude that a carefully planned rainwater
storage tank may be able to satisfy the entire annual domestic water demand of a family
in rural areas of Bangladesh. In another study by Islam et al. (2014), low cost RWH
was assessed as an alternative source for salinity affected coastal region of Bangladesh.
A provision of a 2000 liter capacity rainwater storage tank was found to be adequate
for a rural house with 6 members. A computer model to simulate RWH system
performance for 208 liter rain barrels and larger cisterns was developed by Jones and
Hunt (2010). It was observed that appropriately sized cisterns designed for the
anticipated water usage are required for efficient functioning of the RWH system. It
32
was also found that rain barrels provided inadequate storage and overflowed frequently.
They have a very small role in limiting runoff.
It can be observed that rain barrels and cisterns offer very good potential for
storing water which can be reused for potable / non-potable applications. They can be
used to supplement drinking water particularly in rural areas where there is water
scarcity. The capacity of barrels or cisterns is an important factor affecting effectiveness
as an alternative source and the cost. They have a limited role in attenuating stormwater
and reducing the runoff volume.
33
The stormwater enters through an inlet pipe and excess water is removed with
an overflow pipe (Fig. 2.8). It stores the stormwater temporarily until it
percolates into the soil. Pretreatment is necessary to remove sediments, leaves
and debris.
b) Retention trench: It consists of a trench filled with gravel, sand or loam. This
trench is lined with a geo textile fabric and comprises a layer of coarse gravel
placed under a 300 mm layer of sand or loam. An inflow pipe conveys the
stormwater to the trench and an overflow pipe is provided to remove excess
water (Fig. 2.9). A sediment trap which prevents clogging of the trench is
provided before the stormwater enters the trench.
34
c) Infiltration basin: An infiltration basin is a depression filled with coarse gravel
and 300 mm thick top soil covered with grass. These basins reduce peak
discharge and runoff volumes, and improve the quality of stormwater
discharged to the receiving environments. They are suitable where plenty of
space is available.
d) Seepage pits: A Dry Well or Seepage Pit is a variant of an Infiltration system
for temporarily storing and infiltrating rooftop runoff.
Infiltration devices are not suitable at sites where the hydraulic conductivity of soils is
less than 0.36 mm/hr. They should not be installed on slopes greater than 5%. These
are not recommended for areas where the water table is rising or the salinity of
groundwater is increasing.
Design
The basic data requirements for design of a leaky well include calculation of runoff
volume from a roof and hydraulic conductivity of soil, kh, in m/s. The assumptions in
the design (Argue et al., 2009) are as follows:
Well is empty at commencement of inflow;
Well fills over time 𝜏 minutes (time base of the design storm runoff
hydrograph);
Percolation through floor is at full rate of kh for period of 𝜏 minutes;
Percolation through walls is distributed hydrostatically, hence this component
of outflow is half saturated (outflow) rate for period of 𝜏 minutes;
Perforated wall offers no restriction to outflow;
Groundwater level is significantly below the floor of the well.
The following equations can be used to design the leaky wells (Argue et al., 2009).
𝑉
𝐷=√ 𝜋
(𝐻 + 120 ∗ 𝑘ℎ ∗ 𝜏 ∗ 𝑈) ∗ ( )
4
where D is diameter of the leaky well in m; V is design volume of runoff; H is height
of the well in m and U is moderation factor.
Another form of this equation is
4(𝑉 − 30 ∗ 𝜋 ∗ 𝐾ℎ ∗ 𝜏 ∗ 𝐷2 ∗ 𝑈)
𝐻=
𝜋 ∗ 𝐷2
35
To improve the existing unsatisfactory situation of drainage in Dhaka,
Bangladesh, a proposal of leaky wells using water sensitive urban design principles was
investigated by Ahammed et al. (2013). The authors propose two leaky wells with 2m
height and 2m diameter to be provided for every 500 m2 allotment to improve the
situation.
36
Enhancement of environmental quality and preservation of ecology
(Oberndorfer et al., 2007);
Providing wildlife habitat and biodiversity enhancement.
The review paper by Rowe (2011) discusses comprehensively how green roofs
influence air pollution, CO2 emissions, carbon sequestration, longevity of roofing
membranes, quality of stormwater runoff and noise pollution. This review has evaluated
published research on green roofs in terms of pollution mitigation and suggested future
directions of research. Apart from the benefits mentioned above, this paper highlights
the role of green roofs in reducing the CO2 in the atmosphere. The authors also predict
that green roofs could be incorporated as carbon trading credits in future. Another
benefit of green roofs mentioned in this article is the higher longevity of green roofs as
compared to conventional roofs resulting in fewer roofing material reaching the landfill,
in turn reducing pollutant leaching opportunities (Oberndorfer et al., 2007).
Green roofs have a potential for providing an attractive green space in
downtown areas where the green space on the ground is limited or simply non-existing.
In many countries interest in green roofs is increasing (Berndtsson, 2010).
Generally, green roofs are classified as intensive or extensive. An intensive roof
garden utilizes considerable soil depth with a high diversity of plants as against
extensive roofs which need less water and maintenance. Consequently the weight and
cost of an intensive green roof is much higher than that of extensive roofs (Rowe, 2011).
Being technically less complex, extensive roofs can be provided for sloping roofs and
are also appropriate for large sized rooftops. Their energy and stormwater management
benefits are relatively lower in comparison with intensive roofs (MacIvor et al., 2013).
Though costly, intensive roofs create an aesthetically appealing natural environment
with improved biodiversity and provide better insulation (Berardi et al., 2014).
A number of research articles on various aspects of green roof application have
investigated and analyzed the technique from different perspectives. Some studies have
focused on the runoff quality to investigate if the green roofs are sources or sinks of
various metals (Na, K, Ca, Mg, Al, Fe, Cu, Cd, Pb, Zn, Mn, Cr, Ni, etc.), inorganic
anions and cations. Vijayaraghavan et al. (2012) conclude that concentrations of various
chemical components in roof runoff depend on the nature of substrates used in the green
roof and also on the volume of rain. Gregoire and Clausen (2011) report that green roof
(extensive roof) acts as a sink for NH3-N (Berndtsson et al., 2006; Berndtsson et al.,
37
2009), Zn and Pb but not for total and PO4-Phosphorous (Berndtsson et al., 2009) and
total Cu. The authors identify growing media and slow release fertilizers as the probable
sources of P and Cu in green roof runoff. Berndtsson (2006), (2010) also report soil
material (compost) and added fertilizers as the source of nutrients in green roof runoff.
In a study by USEPA (2009a), runoff from green roofs was found to contain higher
concentrations of most of the nutrients and ions evaluated. Despite the conflicting data
on the influence of green roof on water quality, the overall effect could be positive.
According to Rowe (2011), while newer green roofs may be a source of pollutants,
established vegetation and substrates can improve the quality due to absorption and
filtration of pollutants.
A number of studies report the hydrologic benefits and performance of green
roofs. Many researchers have documented the effectiveness of green roofs in reducing
stormwater runoff (Gregoire and Clausen, 2011; Dinsdale et al., 2006; Alfredo et al.,
2009; Fioretti et al., 2010; Mentens et al., 2006; Berghage et al., 2009; Simmons et al.,
2008; Hilten et al., 2008; Stovin et al., 2010, 2012). The hydrologic analyses of green
roofs have shown that they are highly effective for small storms while being less
effective for larger storms (Speak et al., 2013) and they extend the runoff duration
(Hilten et al., 2008). The hydrologic impact of green roof application at a watershed
scale was modeled by Carter and Jackson (2007). This study was carried out at a variety
of spatial scales and identified areas in the watershed where the reduction in total
impervious area would be significant. For a better understanding of the hydrologic
processes within the green roof and to accurately predict its hydrologic performance, a
number of models have been developed worldwide (She and Pang, 2010; Stovin et al.,
2013; 2015). Yio et al. (2013) have attempted to model the detention effects of green
roof substrates. It was shown that detention increased as a function of substrate depth
(Mentens et al., 2006) and organic matter content. Stovin et al. (2013) have developed
a hydrological flux model for prediction of runoff retention performance of extensive
green roofs. The authors observed that retention capacity of green roofs is strongly
influenced by the climatic conditions and restoration of the capacity after dry periods.
The apparent detention characteristics of a specific green roof system change when
exposed to different climatic conditions (Stovin et al., 2015). The retention in green
roofs is significantly lower in winter than in summer (Mentens et al., 2006; Berghage
et al., 2009). It also depends upon the antecedent moisture condition (Stovin, 2010). A
detention modeling approach applied with a suitable retention model helps in both, long
38
term statistical evaluation and detention performance evaluation based on design storm
(Stovin et al., 2015).
One of the important benefits of green roof application is its energy
performance. It was shown by Fioretti et al. (2010) that green roof outperforms the
reference roof and thus reduces the daily energy demand. The attenuation of solar
radiation through the vegetation layer as well as the thermal insulation performance of
green roof structures was good despite the less favourable climatic conditions in the
Mediterranean context. Green roof temperatures were cooler than the conventional
roofs at the roof membrane as well as inside the building (Simmons et al., 2008). The
article by Castleton et al. (2010) highlights the strong potential for green roof retrofit in
UK to avail the energy savings advantage offered by them.
The influence of plant species on runoff from green roofs was studied by Nagase
and Dunnett (2012). They found that grasses were most effective in reducing the runoff.
It can be concluded from this comprehensive literature review that green roofs have
multiple advantages in terms of energy, runoff and water quality management. The
environmental benefits and economic feasibility of green roofs is indisputable. They
have a good retrofitting potential in urbanized spaces with flat concrete roofs.
41
frequency and severity of combined sewer overflows (CSOs) (NRC, 2008; Carmon and
Shamir, 2010; Novotny et al., 2010).
A number of studies have tried to investigate the impact of alternative
stormwater management strategies on hydrology. Many researchers have proposed and
investigated infiltration based low impact design to reduce the negative impact of
conventional stormwater management systems. Holman-Dodds et al. (2003) compared
three basic scenarios including an undeveloped landscape, a fully developed landscape
with traditional stormwater system, and a fully developed landscape with infiltration
based low impact design, using simple engineering tools. The manipulation of
urbanized landscape with LID was able to reduce the impact on hydrology but was
sensitive to rainfall event size and soil texture. They also observed that greatest
reductions are possible for small, relatively frequent rainfall events and more pervious
soils. The study concluded that the benefits are more for events that generate largest
relative increases from urbanization.
The impact of infiltration techniques on various development types was also
studied by Brander et al. (2004) and Williams & Wise (2006). The authors of both these
studies state that such techniques are effective in reducing runoff and confirm the
observations of Holman-Dodds et al. (2003) regarding sensitivity to soil texture
(Brander et al., 2004) and storm event type (Brander et al., 2004; Williams & Wise,
2006).
Hood et al. (2007) compared the effects of LID development and traditional
development in Waterford, Connecticut and observed that LID decreases peak
discharge rates, runoff volumes and the average runoff coefficient while decreasing the
lag times and runoff threshold. This study strengthens the conclusion by Holman-Dodds
et al. (2003); Brander et al. (2004) and Williams & Wise (2006) that alternative
techniques are most effective for smaller storms with shorter durations.
The effect of cisterns on peak discharge reduction, trap efficiency and runoff
volume reduction was analyzed by Schneider and McCuen (2006). The results showed
that the effectiveness of peak discharge reduction is a function of the size of storm
which increases with decrease in storm size. In another study by Gilroy and McCuen
(2009), the effects of both, location and quantity of cisterns, and bio-retention pits were
analyzed for a micro-watershed. The effectiveness of these BMPs is highly influenced
by both, the spatial location and storage volume. This study further corroborates the
findings by Schneider and McCuen (2006) regarding runoff reduction efficiency with
42
use of cisterns. They also observed that cisterns are capable of controlling rooftop
runoff for small storms. The authors suggest that BMP volume selection should
consider the design objective (which could be peak or volume control). The study
reports the concept of maximum BMP storage beyond which there will be little effect
on runoff volume or peak control. In South Carolina, results showed that although
smaller cisterns could be used for water supply, larger cisterns were needed for
adequate stormwater control (Jones and Hunt, 2010). Crowley (2005) determined that
installation of a 17,034 l (4,500 gal) tank at houses in a Portland, Oregon neighborhood
could reduce the average annual runoff volume by 68%. Gilroy and McCuen (2009)
calculated a reduction of peak runoff rate and volume of more than 30% for the one-
year recurrence interval storm for a single hypothetical single-family residential lot
served by four cisterns with the dimensions of 0.75 m diameter and 0.91 m height. Less
than 10% reduction was found for the two-year event.
44
saving potential is more dependent on the daily precipitation distribution than the
annual precipitation up to a particular limit.
Water savings and financial viability for RWHS has been studied by Rahman et
al. (2010) for multistorey residential buildings in Sydney. Various scenarios in relation
to site area and floor arrangements were generated for a hypothetical multistorey
building and water balance model was developed for each scenario. The most important
factor affecting the financial benefits was found to be roof area. The authors concluded
that RWHS for multistorey buildings can be sustainable under some favourable
scenarios.
Although RWH is being looked upon as a novel solution to tackle the problems
of stormwater management, it is yet to become an accepted practice. Other than
economic aspects, effective implementation of RWH systems further gets impeded by
technical, social and institutional barriers. The article by Ward et al. (2012) addresses
this issue for RWH systems in UK, demonstrates the need for greater interaction
between macro, meso and micro levels and identifies strategic areas for future action.
The authors identify technical (product development), social (in terms of capacity
building) and institutional (support services) development areas in their strategic
framework for overcoming the barriers.
RWH systems have recently been looked upon as a runoff management tool for
urban areas. The potential of RWH for urban runoff management and for better
utilization of water resources, particularly for arid urban centers, is investigated by
Mahmoud et al. (2014). This paper then explores the potential of RWH in Khartoum
City Center as an option for storm water management, since the drainage system covers
only 40% of the study area. The potential runoff is computed using the United States
Natural Resources Conservation Services CN method (NRCS-CN). The authors
propose harvesting of rainwater from rooftops of commercial and business district area
to overcome the inadequacy of existing drainage network. They stress the importance
of harvesting especially for the hot dry season during which the harvested water could
be utilized for domestic and cooling purposes.
While many rural and urban areas are dependent on groundwater as a source of
drinking water supply, this source is challenged due to the increasing depth of water
table and degrading quality of the water. The contamination of groundwater in most of
the districts in Bangladesh by Arsenic poses a threat for its use as a source of drinking
water. According to Alam et al. (2012), rainwater harvesting can be a pragmatic
45
solution to this problem. The authors suggest that rooftop rainwater harvesting may be
able to fulfill the annual domestic water demand in rural areas of Bangladesh.
The type of roofs affects the quality and quantity of rooftop runoff. For RWH
systems to succeed in providing solution to the conflicting issues of flooding and
scarcity of water, roofs should be properly selected for maximizing availability and
ensuring better quality of runoff. Farreny et al. (2011) have provided criteria for roof
selection by comparing four types of roofs, viz., sloping roofs with clay tiles, metal
sheet and polycarbonate plastic, and a flat gravel roof. They found that the sloping
smooth roofs have 50% more RWH potential than flat rough roofs and present better
quality. This type of study can provide guidelines to urban planners for effectively
implementing RWH.
Zobrist et al. (2000) analyzed the quality of rooftop rainwater from an inclined
tile roof, an inclined polyester roof and a flat gravel roof. They stated that the quality
of runoff is a function of the type of roof cover, drainage systems and their interaction
with the atmospheric deposition. Rooftop rainwater quality was also studied in the
article by Hamdan (2009). The article states that smooth tile roofs have little influence
on quality and concrete roofs have a positive effect on removing heavy metals.
Rainwater utilization has the potential to tackle the flooding and water scarcity
problems, particularly in urban areas. However, rainwater utilization may be impeded
due to its contamination. Kim et al. (2005) have focused on developing technologies to
minimize the contamination level of rainwater in urban areas. They have proposed
novel techniques utilizing TiO2, sunlight and Bauxsol to reduce the concentrations of
particles, micro-organisms and nutrients.
RWH systems are gaining popularity due to their inherent advantages and are
being increasingly adopted to tackle the threats posed by urbanization and climate
change, which in turn are threatening the health and availability of water resources.
Many countries are trying to adapt to new practices, the major one being RWH. A
survey was conducted by Thomas et al. (2014) to understand the common practices of
RWH adopted in the United States. The results indicate that the harvested rainwater is
mainly being used for irrigation and is used for potable purposes by more than 25%
respondents.
46
2.7 RWH Applications on GIS Platform
RWH is being increasingly adopted in many countries and decision support
systems are being developed. In order to integrate RWH systems in overall water
resources management, assessment of their hydrologic impact at watershed scale and
identification of suitable areas for its implementation are required. Kahinda et al. (2008,
2009) have developed rainwater harvesting decision support system for South Africa to
satisfy both the above mentioned objectives. It uses a GIS platform to assess the RWH
suitability and results of hydrologic impact are given in terms of runoff reduction.
A model to evaluate the site suitability for farm ponds to augment water supplies in
agricultural areas was developed by Napoli et al. (2014). A methodology integrating
GIS with a NRCS-CN model was developed and utilized to identify potential RWH
sites. The accuracy of the method was verified by field surveys. In another study by
Mbilinyi et al. (2007), a GIS based decision support system using remote sensing (RS)
and limited field survey was developed to identify potential RWH sites in Tanzania.
The outputs of the system are maps showing potential sites of water storage systems
(ndiva), stone terraces, bench terraces and borders. The study proves the capabilities of
RS, GIS and limited field surveys in identifying potential RWH sites. The methodology
proposed by Jha et al. (2014) uses remote sensing and conventional field data to select
suitable RWH structures. This methodology also uses NRCS-CN method to generate
runoff coefficient maps in the GIS environment. Using integrated geo-spatial and
MCDA techniques, suitability zones for various types of RWH structures such as check
dams, percolation structures, etc. were identified at a watershed scale.
47
2.8 Development of Models and Strategies for Sustainable Stormwater
Management
As the urbanization and climate change effects are increasing, many countries
are developing strategies to manage the stormwater sustainably so as to reduce the risks
of flooding and deterioration of the quality of receiving waterways. A number of
models and decision frameworks are being developed worldwide.
A probabilistic model was developed by Zhang et al. (2014) for assessing the
effectiveness of runoff treatment strategy to satisfy the regulatory limits. This model
determines the fraction of storm runoff that needs treatment under the constraint of
regulatory discharge limit. The model when applied to an impervious surface runoff
showed that runoff management strategy is a function of the selection of the target
pollutant.
Quality of stormwater runoff is an important factor for developing effective
stormwater management strategy. The quality parameters of storm runoff from a small
urban residential area in a semi-arid region were analyzed in a study by He et al. (2010).
The study results indicate significant correlation between rainfall intensity and event
mean value (EMV) of Total Suspended Solids (TSS), while no significant correlation
was found between antecedent dry period and EMVs of all analyzed pollutants. Another
significant result regarding effect of first flush on TSS indicates that the entire storm
event should be targeted for TSS rather than just the initial stages.
An attempt to simulate runoff quantity and quality (TSS and TP) in residential,
industrial and commercial catchments using Storm Water Management Model
(SWMM) was made by Chow et al. (2012). They observed that prediction of runoff
quantity and quality using SWMM improved after considering antecedent moisture
condition, depression storage value and local estimates of build-up and wash off
parameters respectively.
The study by Ahmed et al. (2011) evaluated the effect of depression storage on
peak flow for an urban area in Hyderabad, India. The study concluded that the effect
on reduction in peak flow is significant if depth of depression storage is more in
pervious areas. Hence rainwater harvesting structures for every lot should be made
mandatory in order to sustain ground water levels and to minimize flooding during
storms in urban areas.
48
A decision support tool DUWSiM was developed by Willuweit and O’Sullivan
(2013) to model the effects of climate change and urbanization on water supply and
demand and urban water cycle. It uses a dynamic land-use model to assess the effects
of urbanization on various components of the water cycle such as stormwater runoff,
water demand, etc. using a scenario based approach.
The study by Burns et al. (2012) compares and contrasts the hydrologic effects
of the drainage-efficiency focused and pollutant-load-reduction focused conventional
approaches to urban stormwater management with those of a proposed alternative
approach. The authors propose a flow regime management approach to be applied at
smaller scale for protection of urban stream quality. However the cumulative effects of
measures applied at smaller scales on the catchment scale hydrology are not yet fully
understood. The authors suggest development of models to predict such performance
on the catchment-scale hydrology.
Scenario based approach was also used by Fletcher et al. (2007) to study the
effects of urban stormwater harvesting on water conservation and environmental flows.
The results indicated that stormwater harvesting is able to bring the flow and water
quality back to pre-development levels. However, the authors warn about potential of
over-extraction of flow and stress on the need to optimize the harvesting strategy. This
would ensure that both the supply and environmental flow objectives are satisfied.
Brodie (2012) has explored the compatibility between stormwater harvesting and
frequent flow management (FFM) systems. Since both these techniques offer similar
environmental and hydrologic benefits, the potential for their integration was also
explored. It was observed that stormwater storages in both these methods are
compatible to each other and thus integration of stormwater harvesting and FFM
systems can be beneficial.
The selection of suitable stormwater management alternatives depends to a
large extent on the level of development and consequently different decision levels
(regional, local or political) demand different strategies to deal with stormwater. This
necessitates sufficient information and clarity about the possibilities being assessed.
Also, the long term impacts of each decision need to be considered during the
evaluation (Barbosa et al., 2012). Hence a comprehensive and flexible approach to
manage the stormwater is required which will consider the local, societal and regional
priorities of the stakeholders. Different decision scenarios can be formulated based on
the technical and economic constraints. The authors further mention that the best
49
solutions may not be always the most innovative solution or involving high technology.
The choice of appropriate solution may vary based on available information and
stakeholder choices and should reflect environmental and social priorities.
54
However, the effective implementation of the sustainable urban stormwater
approach may be challenging, particularly in developing countries (Silveira and
Goldenfum, 2004; Silveira et al., 2001). Initiation and implementation of sustainable
drainage schemes are inhibited by uncontrolled urban development, leaving very few
open spaces to accommodate infiltration and detention devices. Highly contaminated
nature of runoff further restricts the adoption of infiltration based stormwater
management options. Thus, the factors hindering the adoption of centralized retention
and detention approaches for stormwater management in developing countries may
include maintenance tasks (McCuen and Moglen, 1988), space constraint (Burns et al.,
2012; Clar & Coffman, 2001) and non-point source pollution (USEPA, 1996; Teemusk
and Mander, 2011). Providing detention reservoirs may be challenging in densely
urbanized environments. Thus, detention or retention techniques have limited scope,
particularly in highly urbanized cities. Decentralized stormwater management
approaches can be advantageous in such conditions. Techniques which promote
artificial recharge of groundwater may avoid such problems.
Harvesting at the site scale may prove to be more beneficial as the collected
rainwater can either be used for multiple purposes or infiltrated into ground. Retrofitting
of site-scale watersheds with LID measures is one of the latest techniques for site scale
harvesting. The benefits of implementing RWH as a stormwater control measure and
as an alternative source of water for US cities and individual residents are evident from
the results presented by Steffen et al. (2013). Thus, micro scale, infiltration based LID
techniques, applied in a decentralized way, promise significant benefits for cities in
developing countries like India. Table 2.2 provides detailed information about available
sustainable stormwater management options and describes their advantages,
disadvantages and suitability in Indian context.
55
Table 2.2: Alternative Stormwater Management Measures and Their Suitability in the Indian Context
Stormwater
Management Advantages Disadvantages Suitability in Indian conditions
Alternative
Bioretention (Rain 1. Reduction in runoff volumes and peak flows (Davis, 1. High sediment may cause premature 1. May not be suitable as a
gardens and bio 2008; Hunt et al., 2008) failure centralized means for
swales) 2. Generally requires less space and is more economical 2. Cannot be provided for large drainage management of stormwater due to
3. Requires less maintenance areas high sediment load.
4. Removes pollutants (Carpenter and Hallam, 2010; 2. Can be utilized in a decentralized
Kim et al., 2012) way for small drainage areas
5. Has aesthetic value (Luell et al., 2011)
6. Offers good retrofit opportunities for existing urban
landscapes
Grass swales 1. Reduces runoff volume, peak flow and pollutants 1. Open channels may be potential 1. May lead to favorable conditions
(Parker et al., 2009) nuisance problems for proliferation of vectors or
2. Application is primarily along residential streets and 2. Moderate or high maintenance cost carriers of tropical disease
highways (Silveira, 2002)
3. Adaptable to a variety of site conditions 2. Maintenance issues may further
4. Flexible in design and layout complicate adoption of this
5. Less costly than conventional storm drain pipe system measure
56
Stormwater
Management Advantages Disadvantages Suitability in Indian conditions
Alternative
Green roofs / 1. Reduces percentage of impervious spaces in urban 1. High initial cost 1. May not be suitable for old, small
Vegetated roof areas 2. Moderate maintenance cost residential buildings due to
covers 2. Reduction in runoff volume (Fioretti et al., 2010; 3. Climatic condition structural considerations
Simmons et al., 2008) 2. May be adopted for large roofs in
3. Reduce peak discharge rates (Alfredo et al., 2010; commercial zone where open
Fioretti et al., 2010; Getter et al., 2007; Stovin et al., space is limited
2012; Teemusk and Mander, 2007)
4. Provides aesthetic benefits (Banting et al., 2005)
5. Better thermal performance (Simmons et al., 2008)
6. Decrease in total energy consumption of buildings
(Banting et al., 2005; Castleton et al., 2010)
7. Mitigates the urban heat island effect (Banting et al.,
2005; Berardi et al., 2014)
8. Increases the longevity of roof membranes
(Oberndorfer et al., 2007)
57
Stormwater
Management Advantages Disadvantages Suitability in Indian conditions
Alternative
Permeable / Porous 1. Effective in reducing imperviousness in a drainage 1. Costlier than conventional pavements 1. May not be suitable due to high
Pavements basin (USEPA, 2000) sediment load in stormwater
2. Recharges the groundwater 2. Suitable for low traffic areas such as
3. Improves the quality by arresting the pollutants parking lots and sidewalks (Fletcher et
(Collins et al., 2010) al., 2008; Scholz and Grabowiecki,
2007)
3. Clogging problems may arise due to
high sediment load (Chopra et al., 2009;
Fassman and Blackbourn, 2010)
4. Maintenance is costly (Ahmed et al.,
2011)
Infiltration devices 1. Reduces peak flow (Holman-dodds et al., 2003) 1. Require pretreatment to remove 1. Can be provided for small
(Leaky wells, 2. Recharges the groundwater (Moura et al., 2011) sediment drainage areas in a decentralized
Retention trenches, 3. Improves the groundwater quality 2. Unsuitable for soils with very low manner (Ahammed et al., 2012a)
Infiltration basins) 4. Reduces runoff volume (Ahammed et al., 2012a) hydraulic conductivity 2. Can be connected to any
3. Cannot be installed on steep slopes conventional RWH system which
4. Not suitable in areas with rising water has a filter to trap the debris
table or where salinity of groundwater is
increasing
58
Stormwater
Management Advantages Disadvantages Suitability in Indian conditions
Alternative
Detention Basins 1. Attenuates peak flow 1. Little reduction in runoff volume 1. Unsuitable due to large space
(dry ponds, 2. Simple to design and construct 2. Has large space requirement hence may requirement and highly polluted
extended detention 3. Easy to maintain not be suitable in ultra-urban areas nature of stormwater (Silveira,
basins, detention 4. Can also function as a recreational facility 3. Provides moderate removal of 2002)
ponds, extended 5. Can be used with lining where groundwater is pollutants (Yang and Cui, 2012) 2. Regular maintenance is required
detention ponds) vulnerable 4. May turn into mosquito breeding sites if to prevent unhygienic condition
improperly managed
5. Normally provided towards the end of
sustainable urban drainage management
train
Retention Ponds 1. Reduces peak flow 1. Land requirement may limit use in 1. Not suitable due to scarcity of
(stormwater ponds, 2. Provides good stormwater treatment dense urbanized landscapes space and safety hazards.
wet retention ponds, 3. Provides high amenity and aesthetic benefits 2. Pose health and safety risks (Goldenfum 2. May lead to favorable conditions
wet extended 4. Adds value to local properties et al., 2007) for proliferation of vectors or
detention ponds) carriers of tropical disease
59
2.12 Need for Multi Criteria Decision Making
The comprehensive literature review on sustainable stormwater management and
on the possibilities of implementing such systems in developing countries suggests that
urban stormwater management is complex and there is a need for decision support to Urban
Local Bodies (ULBs) and planning agencies. The selection of appropriate strategies for
stormwater management often involves multiple criteria such as costs, environmental
performance, safety, ecological risks and community perception. There is a significant
growth over the last decade in Multi Criteria Decision Making (MCDM) applications in
environmental decision making (Huang et al., 2011) and in stormwater management
decision support systems (Baptista et al., 2007; Chung et al., 2011; Ellis et al., 2006, 2004;
Lee et al., 2012; Moura et al., 2011, 2007; Sugumaran et al., 2004). The next section
describes various Multi criteria Decision Analysis (MCDA) methods and discusses their
applicability in various decision making situations.
60
2. Expressing stakeholders’ preferences for each criterion and;
3. Determination of best alternative(s) by combining preferences and scores of all the
criteria
Decision makers find the third step most difficult and hence MCDM methods can be used
to identify the preferred alternative. For determining the preferences of the stakeholders,
many modes are available which mainly include direct rating, pairwise comparisons and
lotteries. While selecting the mode, it’s compatibility with the available data and
convenience of stakeholders should be considered.
The final result obtained after comparison of alternatives is sensitive to the choice
of MCDA method. Thus choosing an appropriate MCDA method may be the most crucial
step in the analysis, and a careful analysis of the available methods and its appropriateness
to the selected problem is very critical.
Two approaches can be used to include stakeholders’ preferences for comparison of
alternatives:
1. Multi-criteria selection methods in which preferences are used simultaneously for
ranking of alternatives.
2. Aggregative methods in which all the indicators along with their weights are
aggregated in a single variable, called as Performance Index (PI), reflecting the
performance of each alternative.
Zhou et al. (2006) have classified Decision Analysis methods into three main groups (Fig.
2.10).
61
Fig. 2.10: Decision Analysis Methods (Source: Zhou et al., 2006)
2.13.2 Multiple Attribute Decision Making (MADM) and Multiple Objective Decision-
Making (MODM) methods
Decision making requires consideration of decision maker’s preferences to choose
the best alternative among options involving a number of often conflicting objectives. A
systematic modeling of these preferences can yield most appropriate results. The selection
of alternatives is usually based on more than one criterion. Thus, in practice, decision
making problems are often referred to as multi-criteria decision-making (MCDM)
problems (Kalbar et al., 2012). A variety of MCDM techniques ranging from simple rating
to highly sophisticated systems are documented in literature (Hwang and Masud, 2012;
Hwang and Yoon, 1981; Greco et al., 2005). Hwang and Masud (2012) define two
categories of MCDM methods: MADM (Multiple Attribute Decision Making) and MODM
(Multiple Objective Decision-Making). MADM involves selection of an alternative from
a finite number of feasible alternatives which are evaluated using set of attributes. While
MADM methods can handle discrete decision problems, MODM methods are suitable for
continuous decision problems in which there are infinite numbers of feasible alternatives
(Kalbar et al., 2012; Rao, 2013). MODM can optimize many conflicting objectives by
subjecting them to a set of defined constraints (Kalbar et al., 2012). This study involves a
finite set of feasible alternatives to be evaluated against a set of defined criteria and
indicators. Thus MADM methods are most suitable for this analysis.
62
2.13.3 MADM Methods
Decision makers frequently come across the problem of selection of the most
suitable alternative from a wide range of feasible options. This selection is based on
attributes such as technical, economic, environmental, social, etc., which are often
contradictory. MADM methods help in making preference decisions by evaluating and
ranking the alternatives based on conflicting attributes (Zhou et al., 2006). The processing
of attribute information varies with the MADM method adopted. Various MADM methods
are available for solving decision making problems, which use different ways of processing
of attributes to arrive at a choice (Rao, 2013). Two major approaches for processing of
attribute information are non-compensatory approach and compensatory approach (Hwang
and Yoon, 1981). The non-compensatory approach is mainly founded on the pairwise
comparisons of alternatives with respect to individual criteria (Collette and Siarry, 2013).
Tradeoffs between attributes are not permitted by non-compensatory models. This
indicates that an un-favourable value in one attribute cannot be offset by a favourable value
in some other attribute. Thus, such approaches allow the comparisons to be made on an
attribute by attribute basis. Some of the non-compensatory MADM methods include
dominance, maximin, maximax, conjunctive constraint, disjunctive constraint and
lexicographic methods. A compensatory model allows explicit tradeoffs between
attributes. Although it is more complex in terms of perception and knowledge base, the
outcomes from such models may be closer to the optimal. The chosen alternatives may be
better and more rational as compared to non-compensatory models (Yoon and Hwang,
1995). Multiple attribute utility theory (MAUT) is the basis for most of the compensatory
models in which a single overall criteria is proposed and optimized (Shanian and Savadogo,
2009).
Compensatory models are further classified into three subclasses:
1. Scoring models: Alternative with the highest score or maximum utility is chosen
by considering all of the attributes together at one time. e.g., simple additive
weighting;
2. Compromising models: The alternative that is closest to the ideal solution is
selected. e.g., TOPSIS;
63
3. Concordance models: Preference rankings are generated based on how best they
satisfy a given concordance measure. e.g., ELECTRE (Kalbar et al., 2012).
4. Evidential Reasoning (ER) Approach: ER approach uses an extended decision
matrix. Each attribute of an alternative is described by a distributed assessment
using a degree of belief concept (Xu and Yang, 2001). The distributed assessment
helps in modeling precise data while capturing different types of uncertainties.
Based on the type of information available from the decision maker, there can be three
classes (Source: Hwang and Yoon, 1981; Yoon and Hwang, 1995).
1. No information: Examples include dominance, maximin, maximax.
2. When information on attributes is available, this category can be further sub-
divided into
a. Standard level (Conjunctive, Dis junctive)
b. Ordinal (Lexicographic, Elimination, Permutation)
c. Cardinal (Linear assignment method, SAW, Hierarchical additive weighting
method, ELECTRE, TOPSIS), and
d. Marginal (Hierarchical tradeoffs)
3. When information on alternatives is available, it can be classified as
a. Pairwise preference (LINMAP, Interactive SAW method)
b. Order of pairwise proximity (MDS with ideal point)
Other well-known MADM techniques include PROMETHEE (Preference Ranking
Organization Method for Enrichment Evaluation) (Brans & Vincke, 1985) and AHP
(Analytical Hierarchy Process) developed by Thomas L. Saaty (1995). An extensive
classification of MADM methods is available in Yoon and Hwang (1995). This
classification is based on the type (ordinal or cardinal) and the nature of the available
information. This summary of various MADM methods dominantly includes those
methods which are developed for processing information available in cardinal form
(Kalbar, et al., 2012).
Weighted summation is one of the most commonly used compensatory methods
(Marler and Arora, 2010). The method can convert incomparable attributes into
comparable ones (usually on a scale of 0-1). The attributes are prioritized by assigning
weights. Finally the weighted standardized scores are aggregated and alternatives are
64
ranked. This method has a well-established theoretical background and is easy to explain
and use, though loss of information is possible during standardization. The assumption
regarding the attributes being independent of each other is rather unrealistic in many cases.
Multiple Attribute Utility Theory (MAUT) uses the concept of a complex utility function.
An alternative is assessed by the decision maker with respect to each attribute. The decision
maker assigns importance weights to each attribute. The utilities across the attributes are
aggregated by using an appropriate combination rule. Each evaluation is carried out with
respect to its relevant value dimensions (Von & Edwards, 1986). The utility for the
evaluator forms the common denominator of all these dimensions. This method may
produce better rankings because the decision makers’ preferences are captured in terms of
the utility of attributes and a complex utility function can be used for aggregating the
utilities. Many researchers have applied MAUT to decision problems in various fields such
as water resources planning (Keeney, 1973); research and development planning (Keefer,
1977); energy and environmental management (Buehring et al. 1976), etc.
Another category of MADM methods is the one which is based on an outranking
approach. In these methods, an outranking relationship among alternatives is developed to
select the most satisfying alternative. The outranking relation represents the preference
ranking of the finite set of alternatives. The decision maker can express the preference in
terms of four preference relations: a strict preference, an indifference, weak difference and
incomparability (Roy, 1973; Seppala et al., 2001). The two most commonly adopted groups
of outranking methods are: (a) the elimination and choice translating reality (ELECTRE)
methods and (b) the preference ranking organization method for enrichment evaluation
(PROMETHEE) methods.
1. ELECTRE method
The MAUT models aggregate all criteria values in a single index to rank the alternatives
while the outranking approach in ELECTRE methods can be very different from this
approach. The outranking methods consider bad performance of an alternative on one or
more criteria irrespective of its potential good performance on some other criteria. Basic
MAUT models such as SAW prefer an alternative that demonstrates a superior
performance in the majority of criteria even though it is very weak in the remaining criteria.
65
The ELECTRE method was developed by Benayoun et al. (1966). The method has since
evolved and a number of versions have appeared subsequently (Roy, 1968; 1991; 1996;
Nijkamp and van Delft, 1977; Voogd, 1983; Mousseau and Slowinski, 1998). The
fundamental concept being the same, these versions are operationally somewhat different.
This method models the decision makers’ preferences through outranking relations and the
concept of concordance and non-discordance. The decision maker provides inter-criteria
information interpreted as the criteria weights. These weights represent the relative
importance among the objectives. The alternatives are ranked using the concept of
concordance and discordance. These matrices are estimated using their attribute score and
preference weights. The concordance matrix reflects the differences among the weights
whereas differences among attribute scores are represented by means of the discordance
matrix (Hwang and Yoon, 1981). The outranking relations are built using these indices and
threshold values. Using these relations, alternatives are classified as preferred and non-
preferred. The exploitation of these relations is useful in alternative selection.
Limitations of ELECTRE method
The scientific credibility of the method is questioned by the proponents who say that there
is lot of confusion and arbitrariness in the interpretations. The method and the applied
parameters are in-transparent. These methods are also criticized as ‘difficult to verify
empirically as models of human preferences’ (Stewart, 1992).
Voogd in (1983) has presented another method based on outranking approach called as
EVAMIX (Evaluation of mixed data). This method estimates the dominance scores of each
alternative on criteria by criteria basis. The quantitative and qualitative criteria are
aggregated with the help of different aggregation algorithms.
2. PROMETHEE
Although the ELECTRE method was successful in solving different problems, it required
determination of values of many parameters. Since the decision makers and analysts assign
these values, the method was found to be intricate by them. It was difficult to estimate
values of parameters such as concordance discrepancies and discrimination thresholds
clearly which did not have a real economic meaning. Their influence on the results could
also not be understood properly (Brans and Vincke, 1985). In order to avoid such
difficulties, a new approach which was simple and easy to understand was proposed by
66
Brans (1982). This approach is called as Preference Ranking Organization Method for
Enrichment Evaluation (PROMETHEE). This method is based on the extension of the
notion of the criteria which takes into account the extent of deviations between the
evaluations. These extended criteria represent the natural notion of intensity of preference.
The parameters have a real economic meaning. This makes it easy for the decision maker
to build them (Brans and Vincke, 1985). This method is most applicable to situations in
which the number of alternatives is finite and is involving several (may be conflicting)
criteria (Kalbar et al., 2012). The following six generalized criteria functions cover most
of the practical applications. They are: a usual criterion, a quasi-criterion, a criterion with
linear preference, a level criterion, a criterion with linear preference & indifference area
and a Gaussian criterion (Brans et al., 1986). A review of various PROMETHEE methods
and their applications can be found in Behzadian et al. (2010).
3. Analytic Hierarchy Process ‐ AHP
Another popular method having its own mathematical foundation is the Analytic Hierarchy
Process ‐ AHP method. Numerous applications of this method in a variety of areas have
been documented. The method is very flexible and can be integrated with other techniques
like linear programming, fuzzy logic, etc. (Vaidya and Kumar, 2006). It is one of the most
widely used multiple criteria decision making tools. AHP is a structured technique for
organizing and analyzing complex decisions. It was developed by Thomas L. Saaty in the
1970s (Saaty, 1977) and has been extensively refined since then. A decision problem is
structured using a comprehensive and rational framework in this method. It uses a multi-
level hierarchical structure of objectives, criteria, sub-criteria and alternatives. These
elements of the hierarchical structure can be comprehended and analyzed independently.
They are quantified by decision makers using pairwise comparisons of the alternatives
based on their relative performance against each criterion. The decision makers can use
concrete data as well as their judgments about the elements' relative meaning and
importance for comparing them. Thus the decision makers’ subjective assessments of
relative importance are converted into a set of overall scores or weights (Saaty, 1980). This
is based on the logic that humans are more capable of making relative judgments than
absolute judgments (Kalbar et al., 2012). A numerical scale of 1-9 is used to capture
decision makers’ preferences. The consistency can be checked by verifying the consistency
67
ratio. A numerical weight or priority is derived for each element of the hierarchy by
comparing them in a rational and consistent way. Finally, numerical priorities are
calculated for each of the decision alternatives. These numbers represent the alternatives'
relative ability to achieve the decision goal. Using these values, alternatives are ranked in
the order of their preference. The method is criticized for the rank reversal problem which
is encountered when new alternatives or criteria are added or old ones are deleted. This
problem would not arise if the newly introduced alternatives are not merely copies of
earlier ones and independence of the criteria among themselves and with the alternatives
is maintained. The second concern is about the inconsistency of the judgments. Such
inconsistent judgments may affect the aggregated results and the priorities derived from
them. The condition of order preservation holds for consistent judgements. While dealing
with inconsistent judgements, this condition may or may not hold (Saaty, 2008). According
to Saaty (2008) ‘It is axiomatically imposed, sacrificing the original intent of the AHP
process to derive priorities that match the reality represented by the judgments without
forcing consistency’.
4. Technique for Order Preference by Similarity to Ideal Solutions (TOPSIS)
The limitation of MAUT models which does not consider bad performance of an alternative
with respect to one or more criteria can be alleviated by using ideal and nadir solution
measures. The ideal solution is a hypothetical optimal alternative whose criteria values are
chosen to be the best of each column in a given decision matrix. In contrast, the nadir
solution is a hypothetical unfavorable alternative whose criteria values are chosen to be the
worst of each column in a given decision matrix. A preferred alternative should be as close
as possible to the ideal solution while it is far from the nadir solution (Kangas et al., 2001).
A good example of these methods is TOPSIS. It is a method developed by Hwang and
Yoon (1981) which has a sound mathematical basis. It is based on the compromising model
employing multiple attribute utility theory. The final ranking is based on an aggregating
function representing closeness to the ideal. The alternatives are ranked on the basis of the
shortest distance from the positive ideal solution (PIS) and the farthest from the negative
ideal solution (NIS). The distances to both PIS and NIS are considered while ranking and
a preference order is generated according to their relative closeness and a combination of
these two distance measures (Hwang and Yoon, 1981). Shih et al. (2007) address four
68
distinct advantages of this method. They are as follows. ‘(1) a sound logic that represents
the rationale of human choice; (2) a scalar value that accounts for both the best and worst
alternatives simultaneously; (3) a simple computation process that can be easily
programmed into a spreadsheet; and (4) the performance measures of all alternatives on
attributes can be visualized on a polyhedron, at least for any two dimensions’. Besides,
fewest rank reversals are observed for TOPSIS among the eight methods in the category
(Zanakis et al., 1998). Each alternative is compared directly depending on data in the
evaluation matrices and weights (Cheng et al., 2002). It also provides more distinct and
clear scores for the alternatives. This feature of TOPSIS facilitates the selection of best
alternative at each preference level (Kalbar et al., 2015).
MADM techniques offer useful and practical tools to solve real life problems with
discrete alternatives. The category of techniques in which cardinal information of attributes
is available, is found to be more convenient to decision making. This is owing to their
explicitly represented procedure. In this category, TOPSIS is one of the most
straightforward and hence popular methods. The requirement of paired comparisons is
lessened and the applicability is not limited by the number of alternatives or attributes (Shih
et al., 2007). Hence TOPSIS is most suitable while dealing with problems involving a large
number of attributes or alternatives and when information on cardinal scale is available. A
few weaknesses of the method are presented by Shih et al. (2007). They say that it does not
provide for weight elicitation and consistency checking for judgments.
There are numerous applications of TOPSIS in a variety of fields. The high flexibility
inherent in its methodology makes it more suitable for group decision making. Shih et al.
(2007) propose a group TOPSIS model for decision making. The authors claim that this
model does not cause additional computational burden and is efficient and robust. Since
weight elicitation is not involved in this methodology, either AHP or other suitable
methods can be used to obtain the weights (Shih et al., 2001).
69
2.13.4 Discussion
We collect a lot of information for decision making and today it has become a
mathematical science (Greco et al., 2005). One of the branches of Operations Research
(OR) models, dealing with decision making, is multiple attribute decision making. Most
decision problems involve multiple as well as conflicting criteria along with a set of pre-
defined constraints. Thus various methods are developed which will yield solutions that
can satisfy all criteria and constraints simultaneously. The MAUT forms the basis of the
models based on compensatory approach. The methods based on outranking approach are
preferred in most of the decision problems due to the equivalent importance being given to
the bad as well as good performance of an alternative with respect to one or more criteria.
The ELECTRE method includes this logic through the concordance and discordance
indices and TOPSIS through the ideal and nadir solutions. Both these concepts are
analogous in this sense. Hence both these methods have wide applications in many fields
of decision making. Two major limitations of the ELECTRE method are the confusion and
arbitrariness in the interpretations and the difficulty in empirical verifications as models of
human preferences. These limitations can be overcome by using TOPSIS which offers a
sound logic following rationale of human choice. It is also computationally simple with
fewest rank reversal problems.
70
2.13.5 Applications of MADM Methods
A wide range of MADM methods are available for generating hierarchy of options
in management decisions. MADM methods have been applied extensively for
environmental decision making. Some of the applications include diversion of water in a
watershed (Alipour et al., 2010), chemical selection in treatment of textile wastewater
(Aragonés-Beltrán et al., 2009), urban water supply (Abrishamchi, 2005), selection of
landfill sites (Melo et al., 2006), eco-environmental vulnerability assessment (Huang et al.,
2010), selection of appropriate wastewater treatment technology (Kalbar et al., 2012),
design of sustainable environmental management systems for cleaner production (Khalili
and Duecker, 2013) and selection of efficient solid waste management options (Vučijak et
al., 2016).
71
implementation and O&M costs. ELECTRE TRI method was selected by Moura et al.
(2011) since it is a sorting method. This methodology suits the decision problem in this
work which aims at assigning strategies to predefined ordered categories (Moura et al.,
2007). EVAMIX method was used for developing a decision support tool for urban
drainage by Al-Ani et al. (2011); Sidek et al. (2008) and Sidek & Ezlin (2011).
PROMETHEE was used by Inamdar (2013, 2014) to rank potential stormwater harvesting
sites in urban areas.
72