© IWA Publishing 2017 Water Practice & Technology Vol 12 No 3

698 doi: 10.2166/wpt.2017.077

Modelling rainwater harvesting for commercial buildings

Swee Wen Fooa,*, Darrien Yau Seng Maha and Bartholomew Emily Ayub

a
Department of Civil Engineering, Faculty of Engineering, Universiti Malaysia Sarawak, km-17 Kuching Samarahan
Expressway, 94300 Kota Samarahan, Sarawak, Malaysia

b
Department of Irrigation & Drainage Ministry of Local Government and Housing, State Government of Sarawak, 9th and 10th
Floor, Wisma Saberkas, 93626 Kuching, Sarawak, Malaysia

*Corresponding author. E-mail: [email protected]

Abstract

Commercial areas are generally fully paved and with more impervious land cover than residential areas. This
paper demonstrates a wall-mounted rainwater harvesting system designed to deal with limited land space. An
arrangement of three tanks in series was used on a commercial shop lot where flat roofs generate large amounts
of runoff. The system is compact, and can be installed and fitted close to any wall, promoting the efficient use of
space. Analytical procedures and computational fluid dynamic modelling were used to explore the system’s
potential. This rainwater harvesting system, with its three water storage tanks, works well, and is suitable for
implementation and can be integrated into urban stormwater management.

Key words: flat roof, flow, roof runoff, shop lot, wall-mounted tank

INTRODUCTION

Rainwater harvesting is the interception of rainwater from roof systems so that it can be collected,
stored, and possibly treated for reuse (Natibu & Mahoo 2000; Petrucci et al. 2012). The rainwater
tank type and size are the major considerations for such a system (Worm & van Hattum 2006). Con-
ventional rainwater tanks have substantial capacity, whether above or below ground (Morey et al.
2016), and occupy a lot of space that could be used for other purposes.
Wall-mounted tanks like those introduced here have smaller capacity than conventional ones and
are installed on building walls (Tang & Mah 2015). Figure 1 presents examples of wall-mounted rain-
water tanks on houses. Generally, sloping roofs are favored and house roofs have several sloping
planes fragmented to varying surface areas. The tank shown transfers parts of the rainwater from
the drainage downpipe. Too large a storage volume may result in a tank that it might not be feasible
to suspend from a wall.
However, this paper deals with commercial buildings, which often have flat or gently-sloping
roofs. They may use less water per capita than residential buildings, but harvesting rainwater
could contribute to reducing rainwater volumes in urban drains, thus helping to mitigate flash
flooding. Figure 2 shows typical shop lots in Malaysia, where rainwater is drained to both the
front and back. The roof is a single plane with a large surface area. Placing rainwater tanks on
the tightly-spaced lots requires a new approach and a series of wall-mounted tanks was thought
suitable to cater for the greater amount of roof runoff. This paper outlines an investigation into
the parameters for the wall-mounted tanks using analytical and computational fluid dynamics
(CFD) methods.
 Water Practice & Technology Vol 12 No 3
699 doi: 10.2166/wpt.2017.077

Figure 1 | Wall-Mounted Rainwater Harvesting System on Residential Buildings (www.synergy-contract.com).

Figure 2 | Typical Shop Lots, (a) Front and (b) Back.

RAINWATER HARVESTING SYSTEM

The basic components of a typical rainwater harvesting system comprise a catchment area, convey-
ance system and water storage devices (DID 2012). Technically, the catchment is the surface
receiving the rainfall and supplying the system, and the conveyance (transfer) system comprises gut-
ters, downpipes and piping (Dwivedi & Bhadauria 2009).
In order to check the effectiveness of a rainwater harvesting system, analytical procedures relate the
flow rate of the associated piping system and tank storage volume over time (Figure 3). Rainwater
enters a building drainage system at a given intensity, and its volume and velocity through the
piping system can be simulated (Liao et al. 2015). When the rainwater reaches the tank, the latter’s
filling rate could be used to determine the optimum tank size. If filling is too quick the system has
been under-designed, if it is too slow this indicates over-design.
CFD provides more detailed simulation than analytical procedures, and enables calculation of the
velocity and pressure fields inside a control volume, over the x, y, and z dimensions at any specific
time. The design flow rates determined in analytical procedures can be taken as the control volume,
and plugged into CFD to visualize how the velocity and pressure fields act across the system, including
all its bends, inlets, outlets, tight and broad spaces, etc. (Alnaimi et al. 2015; An et al. 2015).

ANALYTICAL ANALYSIS

A 15-minute, 10-year average recurrent interval design rainfall event is adopted here for design pur-
poses. The storm duration considers the time taken for the rainwater to flow from the flat roof to
 Water Practice & Technology Vol 12 No 3
700 doi: 10.2166/wpt.2017.077

Figure 3 | Proposed Wall-Mounted Rainwater Harvesting System.

an outlet; and 15 minutes are usually taken in normal practice. The design rainfall is taken from the
official release of the Sarawak Department of Irrigation and Drainage (DID), in which an intensity of
180 mm/hr is derived based on the local weather variations.
When a shop lot with a roof catchment of 139.08 m2 is subjected to 180 mm/hr of rainfall for 15
minutes, the catchment could generate 6.3 m3 of runoff (as in Table 1). This scenario is taken as repre-
senting the worst-case. The amount of runoff is huge and therefore two outlets are required, by
dividing them half each to outlets front and back of the shop lot. By placing wall-mounted tanks

Table 1 | Hydrologic analytical parameters

No. Design Parameters Data

1. Roof area for a single shop lot 0.013908 ha/139.08 m2

2. Design storm duration 15-minute, 10-year ARI
3. Rainfall intensity 180 mm/hr
4. Volume of roof runoff generated By using Rational Method,
Q ¼ cIA/360 (c ¼ 1 for 100% imperviousness of roof)
¼(1  118 mm/hr  139.08 m2)/360
¼ 0.006954 m3/s
V ¼ Q  duration of storm
¼ 0.006954 m3/s  (15  60) s
¼ 6.3 m3 (half to front, half to back)
5. Roof runoff using Rational Method Qin ¼ 0.003475 m3/s (to rainwater tanks)
6. Total volume of three rainwater tanks 3(1.5  1.0  0.7 m) ¼ 3.15 m3
7. Velocity at Outfall (m/s) Runoff at Outfall (m3/s)
Using 60 mm pipe v1 ¼ 6.9 Qout1 ¼ 0.012000
Using 90 mm pipe v2 ¼ 1.8 Qout2 ¼ 0.000131
Using 120 mm pipe v3 ¼ 2.0 Qout3 ¼ 0.000205
 Water Practice & Technology Vol 12 No 3
701 doi: 10.2166/wpt.2017.077

on either side of the shop lot’s wall, 50% of the rainwater runoff can be harvested, and the
volume concerned enables the design of water tanks. The other half is directed to urban drainage
for disposal.
The harvested rainwater could be reused, for which a pipe would connect the tank to the intended
appliances, or released to the environment via a downpipe. Neither connection is included in the
analysis, which deals only the critical mass flow into and out of the system. It is also limited in its
detail, such as energy losses in pipe connections.

Figure 4 | Types of Inlet Design, (a) T-Inlet, (b) L-Inlet and (c) I-Inlet.
 Water Practice & Technology Vol 12 No 3
702 doi: 10.2166/wpt.2017.077

CFD SIMULATION

There are three inlet designs. Design 1 has the rainwater runoff distributed through three arms that is
similar to a T intersection, therefore such a design is termed T-Inlet from this point onwards. Design 2
has a sharp 90 ° bend that resembling a L shape, therefore it is termed L-Inlet. Design 3 has an
inclined distribution inlet pipe with a 10° slope, which is termed I-Inlet. The 10° slope is chosen so
that to have a consistent length from inlet opening to tanks for all three designs.
Generally, T- and L-Inlets are designed with one tank which connects directly to the inlet opening –
the middle tank for T-Inlet and the left tank for L-Inlet (Figure 4(a) and 4(b)). In such a design, the
named tanks will receive the greatest volume of water by gravity. Meanwhile, the I-Inlet has no
direct connection to the inlet opening (Figure 4(c)).
Expansion and contraction occur commonly in pipelines with connections and appliances, and
dimensional changes in flow systems have profound implications on flow patterns. The energy used
to move water through pipelines can neither be created nor destroyed, but must be conserved locally.
Generally, water tends to flow faster in smaller diameter pipes and the resultant pressure is lower;
once it passes to larger diameter pipes, the velocity is lower and the pressure is higher, to conserve
energy. In pipeline-tank networks, dimensional changes occur when a pipeline is connected to a
tank to accommodate inflow or a tank to a pipe for outflow (Liu et al. 2016).

Figure 5 | Simulated Streamlines in Rainwater Harvesting Tanks.

 Water Practice & Technology Vol 12 No 3
703 doi: 10.2166/wpt.2017.077

Three modelling scenarios with common tank sizes are represented in CFD. Scenario 1 involves
60 mm pipe throughout the system, while scenarios 2 and 3 use 90 and 120 mm pipes, respectively
(Figure 5). There are thus nine (9) sub-models.
In an enlarging joint, the water velocity changes from relatively high in the pipe to low when entering
a comparatively large tank, and the velocity reduction causes a sharp rise in pressure. In a contracting
joint, the opposite happens. Generally, flow equilibrium tends to occur in the lower pressure and high
velocity zone first – i.e., at the inlets and outlets, and within the pipes – but not in the tanks. This suggests
that the tanks’ velocity/pressure could be used to determine the optimum design.
Different inlet designs distribute water differently, creating various flow patterns in the tanks. Repre-
senting the flow pattern in streamlines shows how the water travels. All scenarios portray low velocity
and high pressure in the tanks (Figure 6). The streamlines appear to be less intense when flow is fast
(Figure 5(g)), but chaotic when slow (Figure 5(a)–5(c)). Further CFD analysis enables determination of
the design offering the most balanced flow in the tanks.
For Scenario 1, the CFD estimate of the maximum velocity at the outfall was in the range 4.4 to 4.7 m/s.
The analytical procedure yielded an estimate of 6.9 m/s. There were intense streamlines in all tanks indi-
cating slow flows. For T- and L-Inlets, the tank with direct connection to the inlet opening shows the
highest pressure among the three in each design (Figure 6), indicating that the direct connection has
forced the water to drain faster than the other tanks, increasing the risk of choking there.

Figure 6 | Simulated Pressures in Rainwater Harvesting Tanks.

 Water Practice & Technology Vol 12 No 3
704 doi: 10.2166/wpt.2017.077

Table 2 | Summary of modelling scenarios

Maximum Velocity at
Maximum Velocity at Tank Outlet (m/s) Outfall (m/s)

Design Pipe Size (mm) Left Tank Middle Tank Right Tank Indicator Analytical CFD

Design 1-1 0.256 1.163 0.974 Unbalanced 6.9 4.717

Design 2-1 60 1.076 0.254 0.983 Unbalanced 4.717
Design 3-1 0.492 0.619 1.208 Unbalanced 4.420
Design 1-2 0.152 0.497 0.390 Unbalanced 1.8 2.026
Design 2-2 90 0.507 0.152 0.397 Unbalanced 2.029
Design 3-2 0.211 0.320 0.506 Unbalanced 1.998
Design 1-3 0.088 0.279 0.210 Unbalanced 2.0 1.241
Design 2-3 120 0.277 0.080 0.224 Unbalanced 1.237
Design 3-3 0.118 0.182 0.296 Balanced 1.449

For Scenario 2, the analytical procedure and CFD yield maximum outfall velocity estimates
between 1.8 and 2.0 m/s. The streamlines exhibit extremes, however, in which one tank always has
low intensity streamlines while the other two are chaotic. This is an explicit demonstration of unba-
lanced flow/pressure in the tanks.
The analytical procedure for Scenario 3 yielded a maximum outfall velocity estimate of 2 m/s. CFD,
on the other hand, gave about 1.2 to 1.4 m/s for all designs. T-Inlet has low intensity streamlines, indi-
cating fast flow/low pressure; with both L- and I-Inlet, however, one tank is significantly more chaotic
than the others, the I-Inlet design having the most nearly balanced velocities between the three tanks
(Table 2).

CONCLUSIONS

In this project, a rainwater harvesting system with wall-mounted tanks was demonstrated successfully
and tested by computer modeling. Three scenarios are described, and the results presented graphically
in the forms of streamline and pressure zonation, supported by the velocities generated in the tanks.
The modeling shows the performance of the flow entering the system under various inlet designs, in
relation to three pipe sizes.
Of the three inlet designs, the I-Inlet has the most consistent flow patterns/streamlines in the tanks
for all three pipe sizes. The T- and L-Inlets designs have faster flows with increasing pipe size. Despite
that, they exhibit different streamlines/pressure zones corresponding to the three pipe sizes – the
60 mm pipe produces the slowest flow in the tanks.
The I-Inlet rainwater harvesting system should have the most balanced velocity characteristics.
Design 3-3 (I-Inlet with 120 mm pipe) is considered the best for rainwater harvesting in the limited
spaces of a close-knit built environment.

ACKNOWLEDGEMENT

The first author received financial support from the Universiti Malaysia Sarawak’s Special Grant
Scheme F02/SpGS/1405/16/6. The second (project leader) and third authors (collaborator) would
like to express gratitude to the same grant.
 Water Practice & Technology Vol 12 No 3
705 doi: 10.2166/wpt.2017.077

REFERENCES

Alnaimi, F. B. I., Murugasen, N. & Al-Qrimli, H. F. 2015 Design and implementation of rainwater harvesting for domestic usage.
Journal of Energy and Environment 7(1), 22–28.
An, K. J., Lam, Y. F., Hao, S., Morakinyo, T. E. & Furumai, H. 2015 Multi-purpose rainwater harvesting for water resource
recovery and the cooling effect. Water Research 86, 116–121.
Department of Irrigation and Drainage (DID) Malaysia 2012 Urban Stormwater Management Manual for Malaysia, 2nd edn.
Percetakan Nasional, Kuala Lumpur, Malaysia.
Dwivedi, A. K. & Bhadauria, S. S. 2009 Domestic rooftop water harvesting – a case study. APRN Journal of Engineering and
Applied Sciences 4(6), 31–38.
Liao, Z. L., Zhang, G. Q., Wu, Z. H., He, Y. & Chen, H. 2015 Combined sewer overflow control with LID based on SWMM: an
example in Shanghai, China. Water Science and Technology 71(8), 1136–1142.
Liu, K., Yang, M., Zhang, Y. & Shen, C. 2016 Nonlinear characteristics of a sudden expansion followed by sudden contraction
channel. In: ASME 2016 International Mechanical Engineering Congress and Exposition, November 11–17, pp.
V008T10A055-V008T10A055. American Society of Mechanical Engineers, Phoenix, Arizona, USA.
Morey, A., Dhurve, B., Haste, V. & Wasnik, B. 2016 Rainwater harvesting system. International Research Journal of Engineering
and Technology 3(4), 2158–2162.
Natibu, N. & Mahoo, H. F. 2000 Rainwater Harvesting for National Resources Management. A Planning Guide for Tanzania.
ISBN 9966-896-52-X. Regional Land Management Unit (RELMA), Swedish International Development Cooperation
Agency (Sida), Nairobi, Kenya.
Petrucci, G., Deroubaix, J. F., De Gouvello, B., Deutsch, J. C., Bompard, P. & Tassin, B. 2012 Rainwater harvesting to control
stormwater runoff in suburban areas: an experimental case-Study. Urban Water Journal 9(1), 45–55.
Tang, E. E. W. & Mah, D. Y. S. 2015 Rainwater harvesting system in a congested residential estate. Journal of Applied Science &
Process Engineering 2(2), 64–73.
Worm, J. & van Hattum, T. 2006 Rainwater Harvesting for Domestic Use. ISBN 92-9081-330-X. Agromisa Foundation and CTA,
Wageningen, the Netherlands.