Chapter 3

RAIN WATER HARVESTING SYSTEM DESIGN

3.1 General
Rain water harvesting (RWH) is practiced in many forms throughout the world. While
surface run-off is collected for agricultural purposes as well as for mitigating flash
floods, roof run-off is used to supplement potable water, mainly to households.

Surveys on water usage prove that per capita consumption can be considered a constant
for a given set of users and the water usage patterns are habitual. Hence, untreated
harvested rain water can be easily used for activities such as flushing of WC and for
vehicle washing. As such, research has improved the collection efficiency and the
performance of various components of the RWH system with particular attention given
to the storage device.

Various storage designs are introduced with a view to cut down the capital cost since
the storage tank is identified as the highest cost component of a RWH system. Hence,
algorithms are developed to determine the optimum tank size for a given demand to
achieve a desired WSE. One such algorithm, developed and presented in graphical
form, was introduced by Fewkes (1999) and validated for Sri Lanka (Sendanayake &
Jayasinghe, 2006).

However, the harvested rain water has to be drawn-off and supplied to user points to
cater to modern conveniences if RWH systems to proliferate. In order to save energy
on pumping, it is desirable if at least part of the harvested rain water can be gravity fed.
Since the storage capacity required for a higher WSE is substantial, placing it at a
higher elevation raises issues such as the need for costly supporting structures and
disturbance to aesthetic appearance of the building envelop. Besides in urban areas,
lack of space demands the storage tank located below ground level. The issues get
further complicated when RWH systems are integrated to multi storey houses. Hence,

69
there is a need for a conceptually new RWH model, which will fully or partially feed
collected rain water by gravity to user points, in multi storey situations.

3.2 Rain water harvesting

The needs and benefits of rainwater harvesting along with the global usage of such
systems will be discussed in detail in the following paragraphs.

3.2.1 Needs of RWH

There exists a growing need for RWH systems worldwide due to a number of factors as
summarized as follows:
• Inadequacy of existing water supply systems in the face of rapid population
growth, creating frequent water shortages and scarcities.
• Degradation of water quality in primary sources such as rivers, ground water
aquifers and natural lakes as a result of wide spread use of chemicals in
agriculture (pesticides, herbicides and fertilizer) and their contamination due to
industrial and human waste.
• Escalating cost of providing water (cost per m3) due to high cost of constructing
reservoirs for storing reticulated water, high costs in pumping from centralized
locations to end user points, filtering and purification costs, distribution system
maintenance costs and financial costs on investments such as opportunity costs.
• Risk of disruption to mains water supply due to break downs or prolonged
draughts. The storage facility of the RWH system can act as the buffer for such
an emergency.
• Non-availability of potable water in isolated areas through conventional
methods due to lack of water bodies in the vicinity, difficulty in reaching ground
water aquifers due to excessive depths and high capital outlay in drilling through
rock, non-availability of power supply inherent to isolated hamlets in arid, semi
arid and mountainous areas.
• Depletion of water levels in underground aquifers thus limiting the draw-offs as
a result of minimal ground water recharging and increased use of ground water.

70
3.2.2 Benefits accrued from RWH
Apart from the obvious benefits of availability of potable water at virtually no cost
excluding pumping cost from the storage tank to end user points, there are a host of
direct and indirect benefits from a well designed RWH system that can be described as
follows:

• Reduced demand on conventional water supply systems by supplementing rain

water for needs which do not require high quality water such as WC flushing,
washing, gardening, vehicle washing etc., thus saving on purified, treated
drinking quality water. This would facilitate managing demand for water and
rationalize new investments.
• Minimized depletion of ground water by recharging in surface run-off
harvesting and preserving it at higher levels and quality, minimizing water stress
during draughts and enhancing the vitality of all life forms.
• Increased decentralized water security and local self reliance whilst encouraging
family level operation and maintenance.
• Facilitating urban home gardening and small-holder food production,
supplementing rural irrigation and stimulating income generation.
• Lowered risk of flash flood situations by taking off a sizable quantity of roof
run-off from the drainage system.
• Reduced national energy consumption and water loss in the treatment and
conveyance of reticulated water.
• Reduced conflictive invasion of rural water sources to cater for urban demand
by meeting requirements close to the point of harvesting.
• Increased domestic water security by reducing the unproductive labor, time and
hazards faced mainly by women and children in fetching water from a distance,
and improved accessibility to safe water for many marginalized communities.
• Minimized consequences of increased salinity intrusion due to sea level rise, and
the threat caused from pollution to traditional sources of water by planned
infiltration.

71
3.2.3 Global use of rain water harvesting

Per capita consumption of water is a relatively elusive figure in practical terms as water
usage patterns vary significantly with life style, draw-off source, and geographical
location of the end user as well as the climatic conditions prevailing in the area. While
per capita water consumption is low in dry and low humid areas, it tends to increase in
areas with abundant rain. It is observed that the relative ease of availability of water
tends to increase the usage while the biggest variation occurs along with life style
differences.

Research in many countries has shown that modern household equipment and amenities
such as WC in toilets, washing machines, dish washers as well as car washing has
significantly increased water consumption. In this chapter, water usage pattern of a
typical household having WC fitted toilet facilities is surveyed, where sizable quantity
of service water is used for non-drinking purposes. Apart from WC flushing, vehicle
washing and gardening require significant quantity of water, for which harvested
rainwater can be used disregarding its quality aspects.

Studies carried out on water usage patterns reveal that a sizable quantity is being used
for WC flushing, car washing and other external uses which do not require drinking
quality water. For example, in Sweden, 20% of household water use is for flushing
toilets, 15% for laundry and 10% for car washing and cleaning (Villareal & Dixon,
2004). In the UK, 30% of the potable water supplied to the domestic sector is used for
WC flushing and the transportation of foul waste (Fewkes, 1999a). In Australia, studies
of water usage in homes located in different climatic regions indicate that on average
15% of supplied water being used in toilets while 30% being used for external purposes
(Australian Bureau of Statistics, AustStats, 2000)

In Sri Lanka, an extensive survey was carried out (Sendanayake, 2007) and average
usage for WC flushing was found to be about 25% of the total water demand.
Importantly, this demand was found to be approximately a constant as the water usage
in a household is generally of habitual nature. However, it is important to note that

72
harvested rainwater is to be used as a supplementary source of water taking a sizeable
load off the reticulated centralized supply.

3.3 Conventional RWH models and their limitations

Conventional RWH models are the ones widely used and are fundamentally classified
into different types depending mainly on the method of roof run-off collection.

3.3.1 Fundamental types of RWH systems

Design wise RTRWH systems are classified into two basic types. They are as follows:
• Dry systems
A dry system for rainwater collection involves down pipes leading directly into the
storage tanks, so after a rain event, no water remains within the collection pipes as
shown in Figure 3.1

• Wet systems
A wet system usually involves underground pipes with the entry to the storage tank
being above ground level thereby trapping water within the pipes after rain as
shown in Figure 3.2

Figure 3.1: The Dry RTWHS Figure 3.2: The Wet RTWHS

73
The dry system is preferred as the wet system can lead to water trapped in the
conveying pipes going stale and in some cases breeding mosquitoes if the pipe
entrances are not securely sealed. Since this additional volume need to be jettisoned
through the first flush device thereby increasing the capacity required by the first flush
(FF) device (Hermann & Schmida, 1999)

3.3.2 Global RTRWH systems

Many practical RTRWH systems are in use globally and differ to each other mostly on
cost factors and the level of sophistication. While many developing countries use simple
systems similar to what used in Sri Lanka, most of the developed countries use
RTRWH systems as supplementary water sources for existing mains supply. In these
systems the discharge is automated so that when collected rainwater in the storage
facility drops to a predetermined level, provision is made for automatic change over to
mains supply. In the Caribbean Islands and Central American countries, for example,
storage tank is made of steel drums of 200 L capacity, large polyethylene plastic tanks
of 1300-2300 L capacity or underground concrete cisterns of 100000 – 150000 L
capacity and the respective government regulations have made it mandatory that all
developers construct a water tank large enough to store a minimum 400 L of rain water
per 10m2 of roof area (Economic & Social Commission for Asia & the Pacific (ESCAP,
1989)

3.3.3 Main types of global RTRWH systems

There are 4 main types of typical Roof Top Rain Water Harvesting (RTRWH) systems
in use internationally, distinguished according to their hydraulic properties (Hermann &
Schmida, 1999)
They are as follows:
(a) The Total Flow type
(b) The Diverter type
(c) The Retention and Throttle type
(d) The Infiltration type

74
(a) The Total Flow type
The total run-off flow is confined to the storage tank, passing a filter or screen before
the tank as shown in Figure 3.3. Overflow to the drainage system only occurs when the
storage tank is full. It is important that in the case of a clogged screen or filter, that
there is no overflow allowed before the tank.

(b) The Diverter type

The diverter type, which contains a branch installed in the vertical rainwater type after
the gutter or in the underground drainage pipe as shown in Figure 3.4 The collected
fraction is separated from the total flow at this branch and a surplus is diverted to the
sewerage system; most of these branches contain a fine-meshed sieve diverting most of
particles to the sewer. These devices are a typical invention of the period, when
rainwater usage was only looked onto save drinking water and the diversion of storm
water to a sewer was the usual and accepted habit. The ratio of efficiency of the
diverting devices decreases with increasing flow. So, during heavy rain, most of the
run-off is diverted to the sewerage system. At low precipitation rates, a minimum flow
is diverted to the sewer and the efficiency decreases to zero (Winkler, 1991, Graf, 1995)

(c) The Retention and Throttle type

The storage tank here provides an additional retention volume, which is emptied via a
throttle to the sewer as shown in Figure 3.5 (Mall-Beton, 1999)

(d) The Infiltration type

Local infiltration of the surplus tank overflow is a possible alternative to the diversion
to the sewer as shown in Figure 3.6 Hydraulic impacts for an infiltration site were
calculated by Herrmann & Schmida (1999). Hermann, Kaup and Hesse (1999)
described performance examples and showed that by the combination of rainwater
usage and local infiltration, the natural local water balance can be restored and
maintained independent of the infiltration capacity of the soil, and independent of
available surface for infiltration facilities.

75
Figure 3.3: The Total Flow type

Figure 3.4: The Diverter type

76
Figure 3.5: The Retention and Throttle type

Figure 3.6: The Infiltration type RWHS

77
3.3.4 RWH systems in Sri Lanka

Rain Water Harvesting (RWH) systems in Sri Lanka are mainly classified according to
the positioning of their storage tanks.

3.3.4.1 RTRWH system with above ground Ferro-Cement tank

This model is introduced to rural areas by the Ministry of Urban Development and
Water Supply of Sri Lanka as shown in Figure 3.7. However, space requirement for the
tank hinders use in small dwellings where land area is limited.

Figure 3.7: RTRWHS with above ground Ferro-Cement tank

3.3.4.2 RTRWH system with partial underground tank

This model, as shown in Figure 3.8, is introduced to the rural areas by, the Ministry of
Urban Development and Water Supply of Sri Lanka. The ease of draw-off due to lower
depth is an advantage. However clearing sediments is the biggest drawback.

78
Figure 3.8: RTRWHS with partial underground Ferro Cement tank

3.3.4.3 RTRWH system with below ground brick tank

In this system the space and aesthetics are saved as shown in Figure 3.9, but cleaning of
sediments and ease of draw-off is hampered. Another practical difficulty encountered is
the roots of nearby vegetation damaging the brick/cement structure of the underground
tank. Therefore, for this particular model plastic tanks are recommended.

Figure 3.9: RTRWHS with below ground tank

79
3.4 Components of rain water harvesting systems

An operational RTRWH system consists of five basic components. They are, the
collector surface also known as the effective roof area or the catchment area, the
conveyance system or the piping to convey rain water to the tank, the storage facility or
the tank, various filtering devices and a suitable draw-off device.

A typical RTRWH system, as shown in Figure 3.10(a) and 3.10(b), has its storage tank
at ground level, requiring a pump to supply collected rain water to end user points. Such
a pump will require either grid connected power supply or can be connected to an
alternative power source, such as a photo voltaic module.

Figure 3.10(a): Typical RTRWHS for multi-story house

80
Figure 3.10(b): Typical RTRWHS for multi-story house (schematic drawing)

3.4.1 Collector surface

The collection area in most cases is the roof of a house or a building. The effective roof
area and the material used in constructing the roof influence the efficiency of collection
and water quality. All catchment surfaces must be made of non-toxic material. Painted
surfaces should be avoided if possible, or, if the use of paint is unavoidable, only non-
toxic paint should be used. Lead, chromium or zinc based paints are not suitable for
catchment surfaces due to presence of heavy metals. Overhanging vegetation should
also be avoided. Steep galvanized iron roofs have been found to be relatively efficient
rainwater collectors, while flat concrete roofs are very inefficient. (Edwards & Keller,
1984)

Rooftop catchment efficiencies range from 70% - 90%. It has been estimated that 1 cm
of rain on 100 m2 of roof yield 10000 L. More commonly, rooftop catchment yield is
estimated to be 75% of actual rainfall on the catchment area, after accounting for losses
due to evaporation during periods when short, light showers are interspersed with
periods of prolonged sunshine (Edwards & Keller, 1984)

81
3.4.2 Conveyance system
A conveyance system usually consists of gutters or pipes that deliver rainwater falling
on rooftop to tanks or other storage vessels. These should be properly supported and
sufficiently strong to carry and keep loaded water during the heaviest rain.

The conveyance system should be constructed of chemically inert materials such as

plastic, aluminum, or fiberglass in order to avoid adverse effect on water quality.

3.4.3 Storage facility

storage tank or recharge tank can be stationed above ground, partly underground or
fully underground depending on the design and spatial arrangements and can be made
of reinforced cement concrete (RCC), Ferro cement, masonry, plastic (polyethylene) or
metal (galvanized iron) sheets. All rainwater tank designs should include as a minimum
requirement:
- A coarse inlet filter
- An overflow pipe
- A manhole, sump and drain to facilitate cleaning
- An extraction system that does not contaminate the water.(A tap or a pump)

Additional features might include;

- A device to indicate the amount of water in the tank
- A second sub-surface tank to provide water for livestock etc.

3.4.4 Filtering devices in RWH systems

Filters are used to filter out the debris that comes with the rooftop water and prevent
them being added to the storage tank. These are of two broad types:

(a) Mesh Filters

A wire mesh fixed at the mouth of or on the down pipe to prevent leaves and debris
from entering the system. While preventing larger objects these filters alone are not

82
sufficient to obtain a reasonable quality rain water collection. Also mesh filters tend to
corrode over time unless the wires are plastic coated. A typical mesh filter is shown in
Figure 3.11

Figure 3.11: A typical mesh filter

(b) First Flush (FF) devices

First Flush (FF) device is a valve that ensures the run-off from the earliest rains is
flushed out and does not enter the system. The first flush of run-off water that occurs at
the beginning of a storm event has been reported to contain a high proportion of the
pollutant load (Fewkes, 1999a). The main cause of this phenomenon is the deposition
and the accumulation of pollutant material to the roof during dry periods. The longer
the dry period, the greater the probability of a higher pollutant load in the first flush. It
is relatively straightforward to install a device for diverting the first flush away from the
collection system (Forster, 1991)

The sizing of the FF devices can follow a simple equation relating to the collection area
and estimated pollution load on the roof.

Flush Volume (L) = Roof Area (m2) x Pollution Factor x 100 [3.1]
Pollution factors are 0.0005, for nil to light pollution, and 0.001 to 0.002, for heavily
polluted sites. This corresponds to 1 mm to 2 mm of initial rainfall (Zobrist, 2000). As
a rule of thumb, the first 1 mm rainfall on a catchment area is to be released through the
FF device.

83
FF devices have a slow release valve which allows the captured water to slowly drain to
the garden or storm water outlet and thereby empty and reset for the next rain event.
The concept is to flush the contaminants from the roof and gutter into the device which
then closes mechanically when full, allowing the remaining roof water to flow into the
tank. The release of the FF water commences immediately and the study by Miller
(2003) showed that this release rate can be significant to the efficiency of the storage
system. A typical First Flush device is shown in Figure 3.12

Figure 3.12: A typical first flush device

3.4.5 Draw-off devices used in RWH systems

Draw-off devices are used to deliver stored rainwater from the tanks to end user points
and can vary according to the design of the particular RTRWH system. A draw-off
device can be:
- A simple outlet to the tank
- A hand pump which are widely used with underground and partial
underground storage devices as shown in Figure 2.17
- A centrifugal or positive displacement pump which can be used to pump
collected rainwater from storage facility on the ground to an over head tank.

84
3.5 Optimization of storage size

It has been shown that Fewkes generic curves for water saving efficiencies (WSE) can
be used to determine the optimum storage capacities for a given demand and for a
desired WSE. The curves are validated for Sri Lanka by Sendanayake & Jayasinghe
(2007) These minimum annual rainfall figures defining the boundary of the domain in
which Fewkes curves hold true are below the minimum annual rainfall figures in the dry
zone of Sri Lanka. As such, the curves given in Chart 2.1 can be used for RWH model
system sizing in any region of the country and can be accepted as universal within Sri
Lanka. However, as the sizing applications move towards drier regions, unless the
capture area is significantly increased D/AR tends to increase thus falling into regions
of lower WSE of the curves. To maximize the WSE for the given D/AR value, S/AR
values will have to be chosen beyond the 0.15 range, indicating bigger storage tanks. A
similar scenario can be seen when the demand (D) for harvested rain water increases,
even in the wet zone.

3.5.1 Space and weight restrictions

It is observed that the harvested rainwater can be utilized for WC flushing and cleaning
purposes the where the amount of water used is approximately 40% of the total water
usage. However, such requirements need the delivery of collected rainwater to utility
points at a sufficient pressure to be used at any given time. One possible energy
efficient arrangement is to position the storage tank at an elevation near the capture area
(at roof level) so that the collected water can be fed to utility points through gravity.
However, when the tank size increases, the space and strength requirements to support
the tank will be beyond the meaningful utilization of harvested rainwater. Further, due
to limited availability of ground space in urban multi-story buildings, positioning of a
larger storage tank above ground will not be feasible and the entire quantity of
harvested rain water will have to be pumped up to utility points. Therefore, typical
sizes of storage tanks will have to be studied to make the model more practical.

85
Considering a typical household in the wet zone, where the annual rainfall is the highest
(1500 mm to 6000 mm), with a capture area of 50 m2, the daily water usage for four
occupants can be taken as 800 L (at per capita demand of 200 L)
If harvested rainwater is utilized only for WC flushing and cleaning,
then the demand for harvested rain water is 800x40% = 320L/day (116.8 m3/year)

As the minimum annual rainfall in the wet zone, Rmin-wet = 1500 mm

The value for D/AR can be calculated as D/AR = 1.56

(It should be noted that the minimum rainfall values are selected as a safety factor for
performance reliability)

From the WSE curves (Chart 2.1), the maximum possible WSE that can be achieved is
found to be 65% and the corresponding value for S/AR = 0.15 giving an optimum
storage size (S) of 11.25 m3. Even when the capture area is doubled (100 m2), it would
still give a value of 1.5 m3 as the storage capacity for the same WSE of 65%. If
however, a WSE of 95% is desired, then the optimum storage capacity (S) will be 15
m3 . Therefore, if a reasonably high and economically acceptable WSE is to be
employed (typically over 80%), then a higher value for the optimum tank size (S) to be
expected. Moreover, as the minimum annual rainfall figure (Rmin) tends to be smaller
for the intermediate and dry zones, higher tank capacities are required if the WSE to be
achieved above 80%.

It can be observed that in order to provide running water facility, the storage tank has to
be placed at a higher elevation-which is not feasible due to volumes concerned. While
such bigger tanks can be accommodated in rural single story houses with abundant
ground space, for urban multistory houses with the necessity of running water will need
a different model to use rain water harvesting effectively and meaningfully.

3.5.2 Alternative methods of storage tank positioning

Various methods of positioning bigger sized storage tanks, which can be used to
provide running water to utility points and the corresponding plumbing configurations

86
possible for typical households, having Roof Top Rain Water Harvesting (RTRWH)
systems supplementing the service water provided by either mains supply or from a
well/bore hole, are presented below. The practical water supply situations in both
single and two story household situations are looked at in five scenarios.

a) The storage tank at ground level, and draw-off through pressure operated
pump (PP)

Collected rainwater is fed to a separate pipeline, feeding WC end user points, at a

higher pressure than the mains. A level sensor operates the pressure pump, to prevent
the pump running dry. The system can be used in multi-storey situations, but no energy
saving is possible. A 5000 L tank connected to a roof area of a minimum 45 m2 is
recommended. A schematic diagram is shown in Figure 3.13

Figure 3.13: Plumbing configuration for RTRWHS – scenario (a)

b) The storage tank mounted on eve of multi-storey house

Rainwater is supplied through gravity, hence no energy consumption occurs.

However, supply of water to upper stories is not possible due to lack of head.
Since the tank is mounted on the eve, space restrictions could occur. Also, a
strength analysis of the eve for its load bearing capacity is required

87
 Figure 3.14: Plumbing configuration for RTRWHS – scenario (b)

It should be noted that if the capture area is > 200 m2, a smaller tank of 2000 L can be
utilized, so that the eve can support the additional weight since the tank size is smaller
compared to that for a smaller capture area. A schematic diagram is shown in Figure
3.14

c) Rainwater pumped from storage facility to an Overhead Tank

In this situation an extra energy input is required to pump collected rainwater to the
OHT. Therefore, the overall system efficiency could be low. A level sensor to operate
the pump P1 fixed in the OHT could improve the efficiency in water saving. This
system is suitable for locations, where ground water levels drop seasonally. A 5000 L
capacity tank connected to a roof area of minimum 45 m2 and a suitable filtering system
in between the rain water Tank and the OHT is recommended. A schematic diagram is
shown in Figure 3.15

88
 Figure 3.15: Plumbing configuration for RTRWHS – scenario (c)

d) Rainwater collected in split sump

To mitigate the unreliability of mains water supply, many households utilize

underground sumps. By partitioning the sump so that one part receives roof collection
while the other part receives the mains supply, savings can be made on service water. A
5000 L capacity tank connected to a minimum roof area of 45 m2 is recommended for
WC flushing water requirement. A schematic diagram is shown in Figure 3.16

89
 Figure 3.16: Plumbing configuration for RTRWHS – scenario (d)

e) Rainwater collected in sump with draw-off through filtration

Employing a series of filters such as Carbon and Sediment filters and a UV sterilizer,
drinking quality water can be obtained from the collected rainwater. It can be envisaged
that, by selecting suitable storage capacities and collection surfaces, substantial water
saving efficiencies can be achieved. A 10000 L tank connected to a minimum roof area
of 200 m2 is recommended for this configuration. However, a higher capacity tank will
ensure water security even in prolonged draught situations. A schematic diagram is
shown in Figure 3.17

90
 Figure 3.17: Plumbing configuration for RTRWHS – scenario (e)

Except in scenario (b), in all other scenarios the requirement of a pump to provide the
harvested rainwater either to an overhead tank or directly to the utility points can be
observed. Such arrangements while preserving water utilizes energy to transfer the
entire quantity of collected rainwater and as such cannot be considered as energy
efficient or as promoting the principles of sustainable development for built
environments.

3.6 Cascading multi tank model

In the following paragraphs a rain water harvesting model is introduced with the new
concept of decentralizing the storage capacity where the roof collection cascading down
through storage tanks located at different floor levels.

3.6.1 Description of concept

In any RWH situation, the storage tank has to be placed at a lower elevation than the
collection area, thereby facilitating the flow of collected rain water into the tank under
gravity. However, the retention volume required for improved WSE levels pose a

91
problem in space requirements in built up areas beside the bigger problem of pumping
back the harvested rain water in to service points for the system to be on par with the
centralized systems as far as the user convenience is concerned. Such a system will
negate the positive contribution of rain water harvesting on sustainability principles by
consuming energy in pumping. In order to minimize the energy requirement in
transferring collected rainwater, a Cascading Multi Tank Rain Water Harvesting
(CMTRWH) model is proposed and analyzed as shown in Figure 3.18.

Figure 3.18: CMTRWH system for a two storey house

In the model, a number of smaller volume tanks are positioned at each floor level, with
the top most tank just below the collection area, and a bigger volume tank at ground
level. Rain water is fed first to the upper tank, the overflow of which will cascade down
to the lower tanks finally ending up in the parent tank at ground level. Supply to each
floor is from individual smaller capacity tanks by gravity floor and make-up water is

92
pumped from the parent tank to the top most tank as and when required. Essentially the
concept of MTRWH model attempts to distribute the storage capacity of the RWH
system at various floor levels so that the requirement for pumping is minimized for the
same or marginally improved overall WSE.

3.6.2 Assumptions adopted in system operation

In developing an algorithm for the operation of a CMTRWH system, the following are
assumed to be valid:
• The height differences between each floor level are a constant.
• The water usage at any given floor level remains constant for a given set of
operating parameters.
• No loss of water occurring in system operation. i.e., in cascading down or
pumping up of collected rain water.
• All tanks installed at floor levels other than the ground level are taken as of
equal capacity.

3.6.3 Advantages and limitations of CMTRWH systems

3.6.3.1 Advantages are the following:

• Possibility of gravity feeding the total usage to service points
• If pumping is required for higher demands, the reduced energy utilization.
• Lower spatial and strength demand on the building structure.
• Reduced adverse impact on the aesthetic appearance of the building envelops.

3.6.3.2 Limitations are the following:

• Reduced supply pressure at user points
• Requirement of additional storage tanks for upper floor levels.

93
3.7 System dynamics

Development of an algorithm to describe the dynamics of the system is an important

step to understand the operational aspects fully.

3.7.1 Development of a system algorithm for CMTRWH systems

It can be shown that both the upper and lower tanks, individually and collectively obey
the Yield After Spillage (YAS) algorithm (Jenkins, 1978) used to develop Fewkes
(1999) generic curves. Therefore, Fewkes generic curves, which have been validated
for Sri Lanka have been used extensively to analyze the system dynamics of CMTRWH
model.

In order to analyze the performance of the system, the amount of water that has to be
pumped up from the lower tank to the upper tank has to be determined. The model can
be considered performing optimally if the demand is met by the upper tank with the
minimum amount of water transferred.

If the water saving efficiency (WSE) of the upper tanks are ηi and the parent tank is ηp
for a given capture area A (m2), annual rainfall R (m) and demand D (m3/year), and the
tank capacities are Si and Sp respectively, from YAS algorithm and Fewkes generic
curves;

ηi = f{ Si, D, A, R}

ηp = f{ Sp, D, A, R}

This can be used to determine the optimum storage tank capacities.

For a given A, R and D, D/AR can be calculated. Then for a desired efficiency (ηp) the
optimum tank size, Sp can be found.
As space and weight restrictions dictate for the installation of a smaller capacity for the
upper tanks, a suitable tank size, Si is selected. (Ideally 1 m3 capacity tank can be
selected for Si) Then for (AR)i and Di, ηi can be found from the curves.

94
3.7.2 Effective run-off and pumping requirement

For cascading multi tank situations, the following algorithms are valid.
For each floor, If the yield is Yi, for i = 1 to n
Pumping requirement Qi ;
Qi = Di - Yi = Di(1- ηi) [3.2]

Then for the ith floor (ith tank),

When the demand is Di, supply is (AR)i
But, (AR)i = (AR)i+1 – Yi+1
Since Yi+1 = Di+1* ηi+1
(AR)i = (AR)i+1 – Di+1* ηi+1 [3.3]

Further, if the total demand is D,

n
D= ∑ Di
i =1
[3.4]

The overall WSE for the system is denoted as ηo

Therefore, if the number of floors are n and the ground floor is taken as i = 0,
it can be shown that;
The amount of water that can be pumped up in CMTRWH system, Q,
n n n
Q= ∑ Qi - ∑ Qi (1- ηP) = ∑ Qi * ηP
i =1 i =1 i =1

From Equation 3.2,

n n
Q = ηP { ∑ Di - ∑ Diηi } [3.5]
i =1 i =1

When,
n
(AR)i = AR - ∑ Di *ηi
i =i +1
[3.6]

95
3.7.3 System limits

The above equations are true when,

(AR)i > 0 and 0.25(AR)i ≤ Di ≤ 2.00(AR)i

Therefore, for the model to function effectively, Di/ (AR)i for each floor level should
behave within the limits validated for Fewkes WSE curves. Further, in order to obtain
the maximum WSE in the multi tank situation, Storage capacity for the parent tank, SP
is taken so that, S/AR ≥ 0.1 This will ensure that the ratio S/AR falls in the stable
region of the WSE curves developed by Fewkes (Chart 2.1)

3.7.4 System equations for equal loads at each floor level

When the demand at each floor level is taken as Di, and the total system demand is
taken as D, for i = 1 to n;
Since ∑ Di = D,
D1 = D2 =………..= Dn = D/n
Therefore, from equations 3.5 and 3.6,

n n
Q = ηP { ∑ Di - ∑ Diηi }
i =1 i =1

n
Q = ηPD{1 – 1/n ∑ηi } [3.7]
i =i +1

n
(AR)i = AR – D/n ∑ηi [3.8]
i =i +1

3.8 Determining the validity of CMTRWH algorithm

To determine the validity of the algorithm developed for CMTRWH systems, the
performance of such should be evaluated under different operating conditions.

96
3.8.1 Methodology

A prototype cascading multi tank model with three tanks are installed in a two storey
house located in Colombo, Sri Lanka with a roof collection area of 50 m2. The capacity
of the parent tank is taken as 12.5 m3 and the upper tanks at 1 m3 each. System
performance is monitored for a daily demand of 200 L per floor with the yield from
each upper tank measured and tabulated daily with the pump in operation. The pump is
connected to floater switch arrangement to cut-in when the water level in the closest
tank to the parent tank drops. The above methodology makes use of the curves
validated for Sri Lanka, initially presented by Fewkes (1999) given in Chart 2.1.

3.8.2 Calculation

Annual yield from the upper tanks and rain water collection AR, are calculated using 15
day moving average method. The moving average method uses a technique where the
average value of a number of consecutive data are averaged and developing a
progression of average values so that a vastly higher number of data can be obtained
from a limited number of data. If the system follows the algorithm, then the maximum
yield possible from the total system, i.e. Dηo should be delivered by the two upper
tanks.

i =n
Therefore, Yo = ∑ Yi generally and Yo = Y1 + Y2 in this case.
i =1

If so, ηo calculated for ηo = ΣY/ΣD from measured (Y1 + Y2) should be equal to ηo
obtained from Fewkes WSE curves for a set of given D, S and AR.

3.8.3 Results and discussion

The results are shown in Chart 3.1.

97
 120

100

80
WSE %
WSE (Chrt)
60
WSE (Calc)
40

20

0
0 10 20 30 40 50
Period Number

Chart 3.1: Comparative WSE values obtained from prototype CMTRWHS

From the results, it can be seen that the calculated WSE, ηo(Cal) and the WSE obtained
from Fewkes generalized curves, ηo(WSE) are almost the same with the margin of error
attributed to system losses.

3.9 Limiting values of Demand (D) for total gravity feed

It will be useful if the limiting values of annual demand (D) can be given in a
generalized form for total gravity fed scenario.

3.9.1 Calculation of limiting values

For the CMTRWH system to confirm to the validity limits of WSE curves, for any Di;

0.25 ≤ Di/(AR)i ≤ 2

n
For Di/(AR)i ≥ 0.25 and (AR)i = (AR)i = AR - ∑ Di *ηi
i =i +1

n
Di ≥ 0.25[AR - ∑ Di *ηi ]
i =i +1

However, when the load is distributed equally among the floors,

Di = D/n where n is the number of floors.

98
 n
Therefore, D/n ≥ 0.25[AR – D/n ∑ηi ]
i =i +1

n
D/AR ≥ 0.25n/[1 + 0.25 ∑ηi ] for n ≥ 2
i =i +1

n
However, for D/AR to be a minimum ∑ηi should be a maximum.
i =i +1

i.e., ηi = 1.00 for all i ≥ n + 1

Therefore, for the minimum D/AR and n ≥ 2

∑ηi = (η2 + η3 + η4 + …………..+ ηn ) = (n – 1)

i =i +1
for ηi = 1.00

Therefore, the limiting value for D/AR,

D/AR = 0.25n/[1 + 0.25(n – 1)] [3.9]

Chart 3.2: Lower limiting values for D/AR for different floor levels

However, it can be shown from WSE curves that;

η = 1.00 when S/AR ≥ 0.05 for D/AR ≤ 0.5

99
 n
In multi storey situations, STotal = Sp + ∑ Si
i =1

It is also seen from WSE charts (Chart 2.1) that,

η = 1.00 when S/AR ≥ 0.02 for D/AR ≤ 0.25
Therefore, for housing units of 2 storey, for ηo = 1.00 and ηi = 1.00
D/AR ≤ 0.4 for STotal/AR ≥ 0.05 and Si/AR ≥ 0.02
And for housing units of 3 storey, for ηo = 1.00 and ηi = 1.00
D/AR ≤ 0.5 for STotal/AR ≥ 0.05 and Si/AR ≥ 0.02
It implies therefore that a CMTRWH system can be designed with STotal/AR ≥ 0.05 and
Si/AR ≥ 0.02 for total supply reliability, without the requirement of pumping, where the
total demand can reach 0.5AR in 3 storey situations and 0.4AR in 2 storey situations.

For example, for a two storey house in Colombo, Sri Lanka where R = 2500 mm/year
and a roof collection area of 50 m2, when Sp and Si are selected as 6.25 m3 and 2.5 m3
respectively, the total demand can be a maximum of 0.4*AR, i.e. 50 m3 per year at
136.9 L/day. Such a demand will ensure that both floor levels are supplied with
collected rain water at 100% WSE. It implies that, by increasing the roof collection
area A, the desired demand can be met for a CMTRWH system where the pumping
requirement is no longer exists.

However, in certain months the rainfall is so low that when converted to annual values,
it may be only about 700 mm per year (Jayasinghe, 2001). Therefore, for a foolproof
design the month with the lowest average rainfall, the month of February, can be
selected to calculate the annual average rainfall though with the disadvantage of having
to select the sub-optimum roof collector area.

Similarly, for a CMTRWH system to operate within the validated limits of WSE curves,
D/AR ≤ 2.0 and Di/(AR)i ≤ 2.0
It can be shown that
n
D/AR ≤ 2n/[1 + 2 ∑ηi ]
i =i +1

100
 n
Since D/AR is a maximum when ∑ηi is a minimum,
i =i +1

n
It can be shown that, the minimum value for ∑ηi is when
i =i +1

∑ηi = 0.5n
i =i +1

Therefore, in CMTRWH situations, for Di/(AR)i ≤ 2.00,

The maximum demand that can be sustained by the system is limited by,
D/AR = 2n/(1 + n)
Therefore, for CMTRWH in multi storey situations,
D/AR ≤ 2n/(1 + n) [3.10]

2
1.8
1.6
D/AR (limiting value)

1.4
1.2
1
0.8
0.6
0.4
0.2
0
0 2 4 6 8 10
Number of levels, n

Chart 3.3: Upper limiting values for D/AR for different floor levels

3.9.2 Operating phases of a CMTRWH model

From Equation 3.5, the quantity of collected rainwater that can be pumped up for a
CMTRWH system with n floors is given by,
n n
Q = ηP { ∑ Di - ∑ Diηi }
i =1 i =1

101
However, when the demand load is distributed equally among the floors, from equation
3.7,
n
Q = ηPD{1 – 1/n ∑ηi }
i =i +1

If the overall system is equated to a conventional single tank RWH system with overall
n
WSE of ηo having a total storage capacity of S, where S = STotal = Sp + ∑ Si , and a total
i =1

demand D for the same A and R,

Then the quantity of collected rainwater that can be pumped up from such a system is
given by QO, where
Qo = Dηo

Therefore,
n n
Q/Qo = ηP/ Dηo { ∑ Di - ∑ Diηi } [3.11]
i =1 i =1

However, when the system is having an equally distributed demand load among the
floors, it can be shown that,
n
Q/Qo = ηP/ηo{1 – 1/n ∑ηi }
i =i +1

It is also shown in equations 3.9 and 3.10 that for ‘n’ floor levels and when SP > Si, the
lower and upper Limiting values of D/AR are 0.25n/[1 + 0.25(n – 1)] and 2n/(1 + n)
respectively for equally distributed demand loads.

Therefore, the performance of a CMTRWH system integrated to a building of ‘n’ floor

levels can be evaluated in 4 phases.

i.e. when D/AR ≤ 0.25n/[1 + 0.25(n – 1)]

0.25n/[1 + 0.25(n – 1)] ≤ D/AR ≤ 1
1 ≤ D/AR ≤ 2n/(1 + n)
D/AR ≥ 2n/(1 + n)

102
When D/AR ≤ 0.25n/[1 + 0.25(n – 1)], the system is operating at 100% WSE through
gravity with no pumping required from the parent tank. Hence Q = 0.

When D/AR increases from the lower limit to 1.00, ηi as well as ηP decrease. However,
the rate of decreasing of ηP can be seen as less than that of ηi (from WSE Chart) though
with the combined effect of increasing Q with respect to increase of D (Chart 3.4).
Hence,
dQ > 0
dD

Therefore, D = AR is identified as the limiting demand at which the maximum amount

of collected rain water can be extracted (the yield, Y) from the system for given storage
capacities (Chart 3.4). In other words, the maximum quantity of water that can be
pumped up occurs at D = AR for CMTRWH system. Therefore, it is important to focus
the attention on supplying the pumping energy required using the most cost effective
and efficient pumping system.

Further, in this phase, when the number of floor levels ‘n’ increases for the same
demand, it can be seen that the quantity of water that can be pumped up reducing for all
D. In other words, the yield is more at D = AR when the number of floor levels, ‘n’
increase.

When D/AR increases beyond 1.00 and reaching towards the upper limit, 2n/(n+1), the
amount of collected rainwater that can be pumped up decreases as both ηi and ηP
decrease. However, from the WSE Chart it can be seen that the rate of decreasing of ηP
is much higher compared to that of ηi.

Therefore, the yield Y drops as a result of reduced Q and thus,

dQ < 0
dD

103
When the number of floor levels increase for a given demand D, the quantity of
collected rainwater that can be pumped up Q decreases thus indicating the increased
yield from the system.

Beyond the upper limit for D/AR, the behavior of the system with regard to continuity
is unpredictable. Mostly, the feed tanks at lower floor levels will not receive the
cascading effect and hence η for lower floor levels will be zero, discontinuing the
operation.

This process can be visualized by using Fewkes generalized WSE charts with regard to
dropping of WSE values for a given set of storage capacities (S) and AR values when
the demand (D) varies. Also, it should be emphasized that the above behavior is true
only when SP > Si and Si/AR ≥ 0.01 hence beyond the un-defined area of the critical
zone of the curves.

Further, the above explained behavior is more pronounced when SP/AR ≥ 0.05 where
the D/AR lines are in the stable area of WSE curves (Chart 2.1).

3.10 Energy requirement in CMTRWH systems

Pumping of a certain percentage of collected rain water in any CMTRWH system is

necessary in order to supply the demand in par with the reticulated centralized systems.
Therefore, it is important to ascertain the requirement of energy in such situations to
select and supply from a reliable source to satisfy the needs of the user.

3.10.1 Energy required in pumping harvested rainwater

Energy required in pumping collected rain water in two types of houses, namely single
story and two story houses, are analyzed for daily demands of 200 L, 300 L, 400 L and
600 L. In the single story house, two tanks are employed with the upper tank of 1 m3
capacity located at the eve level, just below the roof collection area. Three tanks are
employed in the 2 story house with the upper tanks located at eve and first floor levels
and the parent tank at ground level. In the two story house, the demand is taken as

104
equally divided between the two floors. It should be noted that a single pump is used to
lift the collected rain water at ground level parent tank to the top most upper level tank,
making the static head the same for both multi-tank and the comparative single ground
level tank situation. Hence, energy requirement percentage can be calculated using the
quantities of rain water that can be pumped up in multi-story situation Q (Equation 3.7)
and that of single tank situation QO. Therefore, E% in this case can be represented by
Q/ QO. The energy required is calculated using equations 3.7 and 3.8 and is shown as a
percentage of energy required to pump collected rain water from a single tank at ground
level against D/AR, where A is the collector area in m2 and R is the annual average
rainfall in m for a particular geographical region. Use of the parameter D/AR will give
more flexibility to use any combination of A and R, for a given constant AR value.
Fewkes (1999) generalized curves validated for Sri Lanka (Sendanayake & Jayasinghe,
2007), is used to determine WSE for a given demand and storage volume. All storage
tanks located at upper levels are of 1 m3 capacity.

The roof collection area is taken as 50 m2 in the wet climatic region of Sri Lanka, where
the annual average rainfall is 2500 mm (Meteorological Department of Sri Lanka).
Therefore, AR is calculated as 125 m3 and for maximum WSE, Sp is taken as 0.1 AR,
i.e. 12.5 m3. As the generic curves for WSE is valid for 0.25(AR)i ≤ Di ≤ 2.00(AR)i, the
maximum possible demand is calculated as 600 L/day. The amount of rain water that
can be pumped up when only the parent tank is employed is denoted as Qo. The value
Q/Qo is representative of the energy requirement in pumping as a percentage (Equation
3.11). Q/ Qo values are plotted against D/AR to determine the operating characteristics
of CMTRWH systems, where D is the total daily demand. This will effectively
compare the CMTRWH situations for two and three tank models with conventional
single tank RWH systems under the same A, R and D. Tables 3.1 and 3.2 give the
energy requirement percentages Q/Qo for 2 and 3 story houses respectively. η1 and η2
are WSE values for 1st and 2nd floors. Chart 3.4 graphically present the values obtained
in Tables 3.1 and 3.2.

105
It is important to note that for an annual rainfall of a lesser value to select a larger roof
collection area thereby obtaining a AR value which could satisfy the operating
conditions for a given demand D.

A sample calculation is presented in Appendix 1.3 indicating the method of calculation

of Q/ Qo for D=300 L/day, A=50 m2 and R=2500 mm/yr for a 3 Tank CMTRWH
model.

Table 3.1: Energy requirement % vs. Demand in Two Tank model

D L/day D m3/yr D/AR η0 η1 Q (m3) Qo(m3) Q/Qo%
200 73 0.58 100 67.5 23.73 73 33
300 109.5 0.87 90 50 49.28 101.28 48
400 146 1.17 65 45 52.2 116.8 44
600 219 1.74 35 32 44 122.64 36

Table 3.2: Energy requirement % vs. Demand in Three Tank model

D
L/day D m3/yr D/AR η0 η1 η2 Q (m3) Qo(m3) Q/Qo%
200 73 0.58 100 92.5 77.5 10.95 73 15
300 109.5 0.87 92.5 77.5 55 36.96 101.29 36
400 146 1.17 80 67.5 52.5 35.04 116.8 26
600 219 1.74 56 50 42.5 28.06 122.64 23

Chart 3.4: Energy requirement % vs. Demand in Two and Three Tank models

106
3.10.2 Energy required in pumping rainwater with make-up water

When make-up water is available, the pumping energy required in CMTRWH situations
is compared with the energy required in pumping the total demand as a percentage of
the latter.

Table 3.3: Energy requirement % vs. Demand in 2 Tank model – with make-up water
D L/day D m3/yr D/AR η0 η1 Q M Q+M (Q + M)/D
200 73 0.58 100 67.5 23.73 0 23.73 32.5
300 109.5 0.87 92.5 50 49.28 5.48 54.76 50
400 146 1.17 80 45 52.2 28.11 80.31 55
600 219 1.74 56 32 44 104.24 148.24 68

Table 3.4: Energy requirement % vs. Demand in 3 Tank model – with make-up water
D
L/day D m3/yr D/AR η0 η1 η2 Q M Q+M (Q + M)/D
200 73 0.58 100 92.5 77.5 10.95 0 10.95 15
300 109.5 0.87 92.5 77.5 55 36.96 3.67 40.63 37
400 146 1.17 80 67.5 52.5 35.04 20.44 55.48 40
600 219 1.74 56 50 42.5 28.06 72.5 100.56 49

In the above case energy requirement percentage E% is given by (Q + M)/D where Q is

the amount of collected rain water that can be pumped up and M is the amount of make-
up water required for the overall system and D is the annual demand. A sample
calculation is presented in Appendix 1.3 indicating the method of calculating M for
given S, D, A and R.

107
Chart 3.5: Energy requirement (E) % vs. Demand in 2 & 3 Tank models
With make-up water

3.11 Energy required in pumping rainwater with unbalanced load

The impact on the energy requirement in pumping rainwater when the load is un-
equally distributed is investigated in a two story house with a cascading three tank
RWH system. When the demand in the ground floor is D1 and the upper floor is D2,
load distribution of D1/ D2 = 0.5, 1 and 2 are studied for total demands of 300 L/day and
600 L/day.

Table 3.5: Energy requirement for rain water pumping-unbalanced load

D=D1+D2 Q/Qo % E%
D1/D2=0.5 D1/D2=1.0 D1/D2=2.0 D1/D2=0.5 D1/D2=1.0 D1/D2=2.0
300 L 38 34 31 32 28 25
600 L 58 54 51 25 17 12

In the Table 3.5, Q/QO% indicates percentage energy required with the system operating
when only the collected rain water and E% indicates the total pumping energy required
as percentage when the system is operating with make-up water. Make-up water is
introduced to the system to meet the inadequacy of collected rain water.

108
Chart 3.6: Percentage pumping energy required for unbalanced load

Chart 3.7: Percentage total energy required for unbalanced load

Table 3.6: WSE for 3 Tank unbalanced load model

D1/D2 300 L/day 600 L/day
η1 η2 η0 η1 η2 η0
0.5 82.5 62.5 92.5 52.5 47.5 56
1 55 77.5 92.5 42.5 50 56
2 50 87.5 92.5 35 67.5 56

109
Table 3.6 gives the WSE values for 3 Tank CMTRWH systems in 2 story buildings
with the demand distributed unequally as indicated. WSE values are calculated for
daily demands of 300 L and 600 L respectively.

It is noted that for both supply tanks to operate at same WSE, η1 = η2 = η*

Since, Q = ηP{D – D1η1 - D2η2}
= ηP{D – (D1 + D2) η*}
= ηPD{1 – η*} Since (D1 + D2) = D

However, Q0 = D η0
Therefore, η* = η0/ ηP (1- Q/Q0) for a given D, A, R, Sp and Si
As Q/Q0 is maximum at D/AR = 1.00, it is seen that η* is minimum when D=AR

Therefore, it implies that when D=AR, the CMTRWH system operates at minimum
WSE though the overall yield is maximized as discussed in paragraph 3.10

3.12 Impact of variation of the storage volume in CMTRWH systems

Since the maximum energy requirement in rain water pumping occurs when D/AR is
1.00, the impact of change in the capacity of parent tank at D = AR is investigated. A is
selected as 50 m2 and R as 2500 mm/year and the capacity of the parent tank is taken as
12.5 m3.

For a single story, cascading two tank model, the volume of the parent tank is reduced
by 20%, 50% and 90% of the original volume of 12.5 m3 for D = 342.5 L/day (at D =
AR) Therefore, SP is selected as 10, 6.25 and 1.25 m3 and Q/QO% and E% values are
calculated as given in Table 3.7

110
Table 3.7: Energy requirement in pumping with variation in parent tank volume
(Two Tank model)
S/AR η0 η1 Q Q/Q0 % E%
0.08 82.5 45 46.88 45 55
0.05 77.5 45 40.63 42 55
0.01 45 45 0 0 55

50
45
40
35
Q/Qo %

30
Two Tank
25
model
20
15
10
5
0
0 0.05 0.1
Sp/AR

Chart 3.8: Energy requirements in pumping with variation in parent tank volume
When D=AR

Chart 3.8 shows the variation of Q/QO% with SP/AR. It can be seen that when Q/QO%
equals zero SP/AR = 0.01. i.e. when both the upper and lower tanks in the CMTRWH
system are of same capacities.

Therefore, when the capacity of the parent tank, Sp is varied at D = AR, i.e. at the
maximum Q/Q0 percentage point, the impact is pronounced only when SP/AR < 0.05. It
is noted that, while the total quantity of rain water required to pump up remain the same
as indicated by same E% value of 55%, the quantity of make-up water required is
increased. In fact, it can be deduced that this is true for any D/AR, 0.25 ≤ D/AR ≤ 2,
since the energy percentage curve (Q/QO) follows the characteristic of the WSE curve.
Therefore, the capital cost of the system can be significantly reduced, with minimum
impact on the energy cost, while the cost of make-up water is marginally increased. For

111
example, SP can be reduced from 10 m3 to 6.25 m3 with Q/Q0% dropping only by 3%
from 45% to 42% while the quantity of make-up water required M is increased only by
6.25 m3.

3.13 Performance of a Two Tank cascading model – case study

Most of the housing units are of either single or two storey type. Though ideally a three
tank model is suitable for a two storey house, a two tank model will adequately perform
while cutting down the capital outlay by eliminating the eve level tank. Hence it is
appropriate to analyze the performance characteristics of the model along with the
pumping requirements for different scenarios.

In the proposed TTCRWH model, two storage tanks are utilized. A smaller capacity
tank is positioned at a higher elevation (possibly at the eve level) into which the
captured rain water be directed. This upper tank (SU) will supply the utility points and
feed a bigger tank (SL) at ground level via the overflow. As such when a rain event
occurs, captured rainwater will flow into the upper tank and then cascade down into the
lower tank and any excess water to be disposed through the overflow of the lower tank.
The total storage capacity of the system consists of the combined capacities of the two
tanks and a pump is utilized to transfer collected rainwater from the lower tank to the
upper tank when the water level in the latter drops. A schematic diagram of a
TTCRWH model is shown in Figure 3.19.

112
 Figure 3.19: Schematic drawing of a CTTRWH model

3.13.1 System dynamics – Two Tank Model

As SL> SU for the same A, R and D ηL > ηU

Since for a given demand D,

The shortfall in the upper tank (SU) is given by D(1 - ηU) and
The shortfall in the lower tank (SL) is given by D(1 - ηL)

The amount of water that can be pumped up is given by Q;

Q = D(1 - ηU) - D(1 - ηL) which simplifies to

Q = D(ηL - ηU) [3.12]

Similarly, if the total demand for water is DT, then the amount of water required from
the mains is given by M;

M = D(1 - ηL) + (DT – D) which simplifies to

M = DT - D ηL [3.13]

113
3.13.2 System performance

The performance of the TTCRWH model can be studied using the equations 3.12, 3.13
and Fewkes generic curves varying the parameters A, R, D and SU

3.13.2.1 System performance with change in capture area (A)

It can be observed that by increasing the capture area A, for a given R, D and SU that the
dimensionless ratio, D/AR, decrease and as a result achieving higher values for ηL.
However since S/AR decrease with the increase of A, the difference between the water
saving efficiencies of lower and upper tanks, (ηL - ηU), tends to rise, increasing the
quantity of water that has to be pumped up.

3.13.2.2 System performance with change in demand (D)

If the demand is reduced by, for example, using water saving devices the water saving
efficiencies ηL and ηU increases rapidly for D/AR > 1 and slightly for D/AR < 1
This is due to the under-performing of the system for D/AR > 1

3.13.2.3 System performance with change in rainfall (R)

It can be noted that moving from wet to dry climatic zones, where the minimum annual
rainfall (Rmin) drops, both ηL and ηU dropping and as a result, the dropping of pumping
requirement due to lower value for (ηL - ηU)

3.13.2.4 System performance with change in upper tank capacity (SU)

By increasing the size of SU for a given set of parameters A, R and D, ηU increases

reducing the quantity of water required to be pumped up Q, and as a result negating the
purpose of two tank system. It also implies that greater the difference in capacity of the
two tanks the higher the pumping requirement.

114
The operating domain of the Fewkes generic curves dictates that a performing
TTCRWH model can be designed only for 0.25 ≤ D/AR ≥ 2. For values of D/AR
beyond this range the behavior of the curves are found to be unreliable, particularly in
the critical zone of S/AR ≤ 0.05
Further, it is noted that for the system to achieve a water saving efficiency (WSE) of
over 80% (i.e. ηL ≥ 80%), D/AR < 1

Therefore it can be deduced that, for

ηL ≥ 80% D < AR [3.14]

It can also be observed that when the system parameters are selected so that D/AR > 1,
when either A or R is increased or the demand D reduced, ηL increases rapidly while the
increase in ηU moderate due to the fixed nature of the upper tank capacity (SU)

The implications of the above behavior becomes apparent when R > Rmin, which is a
usual occurrence since for reliability of delivery, the minimum annual rainfall, Rmin is
selected in design calculations. It can be shown that when R > Rmin, due to the increase
in (ηL - ηU), the quantity of water to be pumped up Q increases which in turn will
increase the demand on the power source. The effect will be more profound if a stand
alone power source is employed to operate the pump.

However when D/AR < 1, for R > Rmin the value (ηL - ηU) actually reduces, preventing
excess loading on the power source.

It can be shown that for tank capacities SU, SL and annual demand D, the maximum
number of days the system can supply without rain water input is given by,

ddry = (SU + SL)365 [3.15]

D

From historical data, the average maximum number of non-rainy days (rainfall ≤ 0.5
mm) can be taken as, 10, 24 and 45 days for the wet, intermediate and dry zones

115
respectively (Meteorological Department of Sri Lana). Hence, when selecting a value
for SL, it should satisfy Equation 3.15 for system reliability.

Therefore, from Equation 3.15,

ddry ≥ 10, 24 and 45, For the wet, intermediate and dry zones.

3.13.2.5 Pumping requirements for water security

Considering the upper tank SU, the maximum number of days for which it can supply
without an input from pumping is given by dU(max),

dU(max) = 365SU
D

From Equation 3.12, Q = D(ηL - ηU)

If the pumping frequency is taken as NP per year, then the number of days between
consecutive pumping events is given by 365/ NP

It can be deduced therefore, for supply reliability,

dU(max) > 365/ NP

i.e. 365SU > 365

D NP

Hence, NP > D/SU

To compensate for sudden demand loadings, a safety factor K1 can be used,

where K1 > 1.5, thus,
NP = K1D [3.16]
SU

For a pumping frequency of NP, the pumping volume required at a time is Q/ NP

Substituting Equation 3.16 in 3.12 gives,

Q = SU(ηL - ηU) [3.17]

NP K1
Therefore, when the water level in the upper tank SU drops by a quantity equivalent to

116
Q/ NP, a floater switch arrangement can be made to cut-in to activate the pump.

3.13.2.6 Make-up water requirement for water security

From the Equation 3.13, mains water requirement, when the total demand is DT is given
by, M = DT - D ηL
However, the mains water requirement for the RWH system,
ML (i.e. to the lower tank, SL) is ML = D(1 - ηL)

If the number of days the system can supply the demand without mains water is dsup
Then, dsup = 365(SL + SU)
D

If the frequency of supplying mains water is NM, then the number of days between
consecutive supply events is given by; 365/ NM
Since, for system supply reliability,

365 < 365(SL + SU)

NM D

NM > D
(SL + SU)

To compensate for demand surges, a safety factor K2 can be used, where K2 > 1.5.
Thus,

NM = K2 D [3.18]
(SL + SU)

Since the quantity of mains water supply required at a time is given by, ML
ML = D(1 - ηL)
NM

Substituting in Equation 3.18

ML = (SL + SU) (1 - ηL) [3.19]

K2

117
3.14 Control of overflow quantities
Controlling of overflow quantities from RWH systems is an area which need attention
as it is directly linked to the Water Saving Efficiency (WSE) of the system as well as
the discharge volumes on the local drainage systems, particularly in built up areas.

3.14.1 Objective
It is noted that a substantial quantity of the roof collection is lost as overflow in the
RWH systems. This is more so in certain months of the year, such as April/May and
October/November, as the established monsoon rainfall peaks for Sri Lanka. If a
percentage of lost over-flow can be retained, it will not only improve the WSE of the
system, but will provide a means of controlling peak loads on the drainage system. This
is particularly useful in built-up areas, where the drainage system can be designed for a
reduced peak flow, whilst improving the annual quantities of harvested rain water.
With high annual rainfall figure in the South-West of the country where most of the
built-up areas are concentrated (Figure 3.20), it is useful to ascertain the overflow
quantities occurring to address the above issues.

Figure 3.20: Annual rainfall distribution in Sri Lanka (in mm)

118
By establishing a relationship between the overflow quantities and the storage volume
of the RWH system for a given demand, a graphical representation can be made to
determine the additional storage volume required for a particular percentage of over-
flow. The graph can be generalized if all the relevant parameters are divided by the
roof area (A), used for harvesting of rainwater, thereby allowing provision to relate to
any given roof area. Once divided by the projected roof area (A), the storage volume
becomes specific storage in L/m2 and the daily demand becomes specific demand in
mm/d for overflow as a percentage of roof collection.

3.14.2 Methodology

A series of CTTRWH systems with the combined storage capacities of 1, 1.5 and 2.5 m3
in Colombo, in the wet climatic region of Sri Lanka (annual average rainfall 2500 mm)
and another set of tanks with similar storage capacities in Anuradhapura in the dry
climatic region (annual average rainfall 1500 mm), were set up and daily yields were
recorded for calendar year (2008). The daily demand was taken as 200 L/d and 100
L/d, representing a household of 4 and 2 people with daily per capita consumption of
200 L of water, of which 25% is used for WC (Sendanayake & Jayasinghe 2006). The
collection area is taken as 25 m2 of clay tiled roof with an inclination of 150 to the
horizontal plane. Daily rainfall data are recorded for the entire period of the research at
both locations, and verified with data collected at the National Meteorological
Laboratories located at close proximity to test sites.

3.14.3 Calculations

The collection coefficient (Cf) of the roof area was calculated as 85%. From daily
rainfall data, the daily roof collection was calculated and the overflow quantities
determined by deducting the daily yield. The annual overflow quantities were
calculated as a percentage of the total roof runoff collection and plotted against specific
storage volume for two specific demands of 8 mm/d and 4 mm/d.

119
Data collected on daily rainfall, yield and calculated values of over-flow quantities are
given in Appendix 2. Calculated values of annual over-flow quantities for given
specific storage (SS) in L/m2 and specific demand (SD) in mm/day are given in Table
3.8

Table 3.8: Annual OF quantities for given SD and SS

SD SS
(mm/d) S (m3) (L/m2) OF/yr
4 1 40 25.4
1.5 60 19.2
2 80 18
2.5 100 17.6
2 1 40 46.1
1.5 60 40.2
2 80 38.9
2.5 100 -

Chart 3.9 represents overflow as a percentage of roof collection versus specific storage
volume in L/m2 Colombo in the wet region of Sri Lanka. In the legend, d/100 and
d/200 represent the daily water demand of 100 and 200 L for WC flushing.

Chart 3.9: Overflow % for different specific storage volumes for Colombo

120
3.15 Summary

Any rain water harvesting system integrated in to a building should be capable of

providing collected rain water to user points at optimum energy efficiency. By
introducing a multi tank rain water harvesting model of cascading roof collection type,
henceforth termed as the cascading multi tank rain water harvesting (CMTRWH)
model, it is found that not only the energy issues, but the spatial, structural and aesthetic
issues can also be overcome. With the underlying concept of effectively distributing the
storage capacity among different floor levels to feed the bulk of the harvested rain water
through gravity, it is found that the energy efficiency of the system can be substantially
improved compared to RWH systems of conventional type, particularly in multi storey
situations.
Considering the importance of generalizing the model for any multi story situation, an
algorithm is developed where the system behaves within the limits of validity defined
for Fewkes generalized curves introduced by Fewkes for water saving efficiency
(WSE). Analyzing the developed algorithm, it is found that for the system to supply
water to user points without the requirement of pumping is defined by an equation
relating D/AR to the number of floor levels for any multi story situation. Further, a
similar equation is developed for system continuity, i.e., to prevent the system running
dry. It is seen from the curves developed, that as the number of floors increase, so does
the maximum possible load D, indicating the inherent energy efficiency of a two story
(3 Tank) housing unit over a single story (2 Tank) one for the same total demand.
However, on the other hand, a single story house with a two tank model is less
vulnerable to system running dry and hence more suitable for dry regions where the
maximum available roof collection can be extracted without system failure.

The requirement of pumping is a major issue in any RWH system. Using the algorithm
developed for CMTRWH system, the pumping quantities that is possible in the model is
compared with that of a single tank model. It is found that the pumping energy
requirement in a multi story situation, as a percentage of that required in a single tank
model, is a function of WSE for each floor level and the number of floor levels (n),
when the load distribution is equal for all n floors. It is also found that the optimum

121
performance with regard to the quantity of rain water pumped and the associated
utilization of energy occurs when D = AR. When D > AR, the system tends to drop
WSE at all floor levels resulting in requiring a higher quantity of make-up water, hence
requiring more energy for pumping as demonstrated by the total energy requirement
against D/AR graph.

In many practical situations using CMTRWH models, the occurrence of different water
usages at different floor levels can be expected. Analyzing the algorithm, it can be seen
that the energy requirement is less when the load is biased towards the upper floors.
For a two storey house for example, when D1 and D2 are the demands for the upper and
lower floors, respectively, the percentage difference in energy usage between D1/ D2 =
0.5 and D1/ D2 = 2.0 is only 15%, indicating the impact is not substantial. However, it
also highlights the importance of demand distribution in designing multi storey houses
so that the energy utilization in pumping can be minimized.

If the maximum demand is to be extracted from a given combination of A and R with

the assistance of pumping, it was shown that D should be equal to AR. Considering the
possibility of varying the storage capacity at the optimum conditions when D = AR, it is
observed that the capacity of the parent tank can be reduced so that S/AR = 0.05
without affecting the system efficiency. Hence, the capital outlay required for the
rainwater harvesting model can be cut down by minimizing the tank size for the
optimum conditions.

Overflow quantities in RWH situations

In developing the percentage overflow against specific storage volume chart, if data
from a longer time series is taken, more accurate overflow quantities could be possible.
However, it can be shown that for most tropical climates, the rainfall is seasonal and the
heaviest precipitations occur due to annual monsoons. Studying historical rainfall data
for Sri Lanka, it can be seen that although the annual average rainfall vary by as much
as 20% in the wet region and by approximately 30% in the dry region, the maximum
number of storm events and hence the highest rainfall occurs during April/May and

122
October/November in a calendar year. Further, the average rainfall during the above
periods show close similarity in precipitation amounts (in mm) over a 10 year period
from 1999 to 2009 (National Meteorological Department of Sri Lanka).

Since the maximum overflow occurs during periods of maximum rainfall, it can be
safely assumed that the results obtained from measuring and calculating overflow
quantities in a single year closely resembles a similar data set collected over a longer
period of time. It is clear from historical data, that the average rainfall during peak
rainy months is approximately same with a maximum variation of 15%.

It can be seen from the graph, that for a significant percentage drop in overflow, the
specific storage volume has to be largely enhanced. In any case, practically, overflow
percentage cannot reach zero due to unpredictability of the strength and intensity of rain
events in any particular period of time. However, if a minimum of 50 years of rainfall
data are collected for a particular region and simulated to calculate overflow
percentages, the maximum additional retention volume required for maximum rainfall
occurred as well as average additional retention volume required for annual average
rainfall during peak rainy period can be calculated. Whilst the former can be useful in
flash flood control situations the latter is useful in RWH situations. Further, it can be
seen from the graph that a more pronounced impact can be affected on the overflow
percentages by increasing the specific rain water consumption. Therefore, if harvested
rain water can be used further to WC flushing, a steeper reduction in overflow
quantities can be achieved.

Developing of an overflow percentage chart was not attempted for the dry region of Sri
Lanka due to low overflow quantities as well as the availability of non-built-up land for
natural seepage for excess roof collection. In conclusion, the graph developed can be
used as a design tool in combination with Fewkes generic curves for WSE, in
determining the ideal volume for rain water storage, maximizing the WSE whilst
minimizing the overflow quantities.

123