Effectiveness of ABC Waters Design Features for Runoff Quantity Control in Urban Singapore

Abstract

Active, Beautiful, Clean Waters (ABC Waters) design features—natural systems consisting of plants and soil that detain and treat rainwater runoff—comprise a major part of Sustainable urban Drainage Systems (SuDS) in Singapore. Although it is generally accepted that ABC Waters design features are able to detain runoff and reduce peak flow, their effectiveness in doing so has not been studied or documented locally. This research aims to determine their effectiveness in reducing peak flow based on a newly constructed pilot precinct named Waterway Ridges. Four types of ABC Waters features have been integrated holistically within the development, and designed innovatively to allow the precinct to achieve an effective C-value of 0.55 for the 10-year design storm; the precinct-wide integration and implemented design with the aim of substantially reducing peak flow are firsts in Singapore. The study is based on results from an uncalibrated 1D hydraulic model developed using the Storm Water Management Model (SWMM). Identification of key design elements and performance enhancement of the features via optimisation were also studied. Results show that the features are effective in reducing peak flow for the 10-year design storm, by 33%, and allowed the precinct to achieve an effective C-value of 0.60.

1. Introduction

Cities are being exposed to the threat of increasing flood risk due to climate-change-induced sea level rise and the increased frequency of extreme rain events, as well as increased urbanisation resulting in more impervious areas and runoff [1]. This is especially true of Singapore, a tropical island that has undergone rapid urbanisation over the past few decades despite its land constraints; the city state has an area of about 719 km², is the third most densely populated country in the world, and has an urban population of 100% [2,3]. The historical rainfall data from 1980 to 2012 also indicate a trend of higher rainfall intensities (Figure 1). The conventional approach of building wider and deeper drains to quickly collect and channel rainwater runoff away from the urban catchments has been accepted to be no longer adequate or sustainable to cope with the challenge of climate change, limited land and increasing urbanisation in Singapore. There was a paradigm shift on storm water management in 2006 when PUB Singapore’s national water agency, launched the Active, Beautiful, Clean Waters (ABC Waters) Programme [4]. The ABC Waters Programme is an initiative to transform Singapore’s drains, canals and reservoirs beyond their traditional functions of drainage, flood control and water supply storage into beautiful streams, rivers and lakes that are well integrated with the surrounding landscapes. The programme is part of PUB’s strategic objective to bring Singaporeans closer to water so that they can better appreciate and cherish this precious resource, and in turn keep the waters of Singapore clean and free of litter.

Under the programme, storm water runoff is proposed to be managed in a more sustainable manner via the implementation of ABC Waters design features, natural systems consisting of plants and soil that are able to detain and treat rainwater runoff before discharging the cleansed runoff into the downstream drainage system [4]. There are various types of ABC Waters design features including the bioretention basins or rain gardens, bioretention swales, vegetated swales, constructed wetlands (whose types consist of surface flow, floating and subsurface) and sedimentation basins. While ABC Waters design features are similar to Sustainable urban Drainage Systems (SuDS), Water sensitive Urban Design (WSUD), Low Impact Development (LID) and Green Infrastructure (GI) that are prevalent in other countries [5], ABC Waters design features generally have a stronger focus on the cleansing function and represent a subset of the features represented by the other terms; there are features such as green roofs, detention tanks and pervious pavement which are not considered as ABC Waters design features but are part of the features under SuDS/WSUD/LID/GI.

The Source-Pathway-Receptor approach in drainage management is also advocated in Singapore and ABC Waters design features are recommended as one of the source control measures to manage storm water runoff [6]. A runoff control legislation, imposed from 2014 onwards via an addendum to the Code of Practice (COP) on Surface Water Drainage, requires “industrial, commercial, institutional and residential developments greater than or equal to 0.2 ha in size” to “control the peak runoff discharged from the development sites” [7]. It is stated that “the maximum allowable peak runoff to be discharged to the public drains will be calculated based on a runoff coefficient of 0.55, and for design storms with a return period of 10 years and for various storm durations of up to 4 h (inclusive)”. In addition to detention tanks, ABC Waters design features are also among the measures recommended in the COP on Surface Water Drainage to achieve the peak runoff reduction [7].

Although it is generally accepted that ABC Waters design features are able to detain runoff and help to reduce peak flow, their effectiveness in doing so has not been studied or documented locally in Singapore. Modelling of the runoff control performance of ABC Waters design features is almost non-existent in Singapore and while modelling is carried out, they are done to assess the performance of the features in terms of water quality improvement [8]. This could be inherently due to the fact that ABC Waters design features are not widely used for runoff control in Singapore at the time of writing, as the general perception is that ABC Waters design features are not effective enough to lower peak flow to the standards (C = 0.55) stipulated by the runoff control legislation in Singapore. Instead, the features are normally implemented in Singapore for improving water quality and other ecosystem services such as providing social and recreational spaces for the public or residents in a development [4]. This research thus aims to address the gap of a lack of study on the effectiveness of ABC Waters design features for runoff quantity control in Singapore by performing such a study, and to do so via a demonstration on the use of modelling and optimisation tools for such an assessment.

There are modelling as well as field monitoring studies on the effectiveness of SuDS/LID/WSUD/GI in reducing peak flow within temperate catchments internationally [9,10,11,12,13,14,15,16,17,18]. However, the studies on SuDS application in tropical climates are limited. Taking the example of bioretention systems (i.e., vegetated land depressions that are designed to detain and treat storm water runoff via sedimentation, filtration and biological uptake [4], which is a type of ABC Waters design feature), they have been reported to be able to achieve runoff peak flow reduction in the range of 24–99% in the United States of America (USA) [15,17,19,20,21] and 80–90% in Australia [22,23] based on field monitoring studies. The peak flow reductions depend on the ratio of catchment area to bioretention area, the drainage configuration and the rainfall magnitude, with smaller peak flow reductions being achieved for higher rainfall intensities as well as increased rainfall depths and longer rainfall durations [16,17,18,21]. Hence there is a need for region specific study due to the large variations in the peak flow reduction effectiveness of the bioretention systems [20].

This research directly addresses the gap of lack of study on the peak flow reduction performance of ABC Waters design features/SuDS/LID/WSUD/GI in the tropical climate setting. In addition, the research aims to identify key design elements that could help to enhance the effectiveness of ABC Waters design features in reducing peak flow. Performance enhancement of the features will also be studied via optimisation.

The relevance of the study results will not be limited to Singapore as the study results will also provide inputs towards bioretention design and practice in many other countries with similar tropical climates, such as countries in South East Asia, Sri Lanka, Uganda, Honduras, El Salvador etc. It is hoped that the study results would help to guide the design and implementation of SuDS/LID/WSUD/GI in the other countries with tropical climates.

2. Details of the Site and ABC Waters Design Features

2.1. The Study Site

The research was based on a newly constructed four hectare pilot project named Waterway Ridges, which is a joint collaboration between PUB and the Housing Development Board of Singapore. The construction of this pioneer precinct began in April 2012 and was completed in April 2017. This public residential development contains four different types of ABC Waters design features that are integrated in the design at precinct level, namely bioretention basins/rain gardens, bioretention lawns, vegetated swales and vegetated swales with gravel layer (termed as “gravel swales” in the rest of this paper). Such integrated application has been adopted at precinct level in Singapore for the first time. The ABC Waters design features are also installed with the aim of substantial peak flow reduction, which is another first in Singapore.

The ABC Waters design features at the site have been innovatively designed to reduce the peak flow of the precinct to correspond to a reduced runoff coefficient of 0.55 for all storm events up to the 10-year return period. The innovative design, which is implemented for the very first time in Singapore, comprises 400–750 mm thick detention gravel storage layers which are incorporated below the rain gardens and gravel swales, and orifice outlets that restrict the outflow from the features. Normally dry, the network of ABC Waters design features (Figure 2) receives runoff from about 60% of the total site area during a rain event [24].

The climate in Singapore is characterised as tropical and coastal. The country experiences high and uniform temperatures (the minimum and maximum temperatures are 23–25 °C and 31–33 °C during the night and day, respectively), high humidity (the mean annual relative humidity is 83.9%), and abundant rainfall throughout the year. The average annual rainfall is 2328.7 mm (based on long-term records from 1869 to 2016) and it rains 178 days a year on average [25]. There are two monsoon seasons in the country, namely the Northeast Monsoon occurring from December to early March, and the Southwest Monsoon from June to September. Though there are no distinct wet or dry seasons in Singapore, there is a monthly variation of rainfall in Singapore. Generally, there is increased rainfall from November to January during the Northeast Monsoon season. On average, the wettest and driest months are December (318.6 mm) and February (112.8 mm), respectively [26].

2.2. Design of ABC Waters Design Features at the Study Site

Four different types of ABC Waters design features are implemented at the precinct. There are 21 rain gardens (with codes FB1 to FB21), four vegetated swales (with codes VS1 to VS4) and two gravel swales (with codes GS1 to GS2) in the precinct.

2.2.1. Bioretention Basins/Rain Gardens

Rain gardens or bioretention basins are vegetated land depressions that are designed to detain and treat storm water runoff via sedimentation, filtration and biological uptake [4]. A typical cross section of the rain gardens implemented at Waterway Ridges is shown in Figure 3; various elements of the design are indicated by the numbers (1) to (5), and will be referred to in the rest of this paragraph. During a rain event, runoff from the subcatchment would flow into the rain garden and be allowed to pond up to a maximum depth of 200 mm on the surface. Concurrently, the runoff percolates through the filter media and is eventually collected by the subsoil perforated pipes (5). Excess runoff greater than the 200 mm surface detention depth will overflow into the overflow manhole (1) and be directed into the underground gravel layer for detention and storage via the subsoil perforated pipes (5). Meanwhile, the amount of flow leaving the feature and entering the discharge pipe (4) will be regulated through the orifice outlet (3), the opening size of which was predetermined through calculations to correspond to C = 0.55 for the 10-year design storm. When the underground gravel layer is full, the water level in the manhole will rise to the overflow standpipe opening (2) and be discharged via the discharge pipe (4) that connects to the roadside drains within the precinct.

Unlike the typical rain gardens with drainage layers that are 200–250 mm thick, the rain gardens implemented at Waterway Ridges have a much thicker gravel layer ranging from 400–750 mm, and the orifice outlets are not part of the design of typical rain gardens. The overflow manhole is also fitted with gratings and a “filter basket” to minimise debris from flowing into and choking the gravel storage layers. Literature review suggests that such a design is new not only in Singapore but also internationally. Among the designs found, one that is most similar to the rain gardens at Waterway Ridges is the “Cap-orifice Flow regulator” design [27]. In this design, a cap-orifice is proposed to be installed at the exit of the subsoil perforated pipe for the purpose of regulating the infiltration rate as well as the outflow from the rain garden [27]. However, such a design would not be effective in reducing peak flows in tropical climates with extreme rainfall like Singapore, where a large portion of the subcatchment generated runoff (from storm events greater than the 3-month return period) would overflow into the overflow manholes instead of infiltrating through the filter media of the rain gardens.

2.2.2. Bioretention Lawns

Bioretention lawns have the same design as rain gardens, but have lawns extensively as their planting as compared to shrubs in the rain gardens (Figure 4). When it is not raining, the bioretention lawns serve as recreational spaces for the residents of Waterway Ridges.

2.2.3. Vegetated Swales

Vegetated swales are essentially earth drains that convey runoff. In Waterway Ridges, they are gravel lined and vegetated with short grass (Figure 5). Due to their increased roughness, the flow of runoff is slower in the swales as compared to concrete-lined drains. The slower runoff velocities help promote sedimentation, thus resulting in cleaner runoff.

2.2.4. Gravel Swales

A typical cross section of the gravel swales implemented at Waterway Ridges is shown in Figure 6; various elements of the design are indicated by the numbers (1) to (4), and will be referred to in the rest of this paragraph. Gravel swales are essentially vegetated swales (1) that are coupled with underground gravel storage layers (2). Runoff flowing through gravel swales would be channelled directly into the gravel layers via the overflow manholes (3) and subsoil perforated pipes (4). The components in the overflow manholes of gravel swales are the same as that of the rain gardens. Similarly to the rain gardens, flow leaving the gravel swales will be regulated through an orifice outlet located within the overflow manhole.

2.3. Monitoring of the Study Site

There are plans for water quality and quantity monitoring to be carried out at the precinct. The plan includes flow monitoring at each of the four precinct discharge outlets as well as at the inlets and outlets of four specific ABC Waters design features (one for each type) to determine the flows entering and leaving each feature type.

At the time of publishing however, the monitoring has yet to commence and as such, the results presented in this manuscript will be based on uncalibrated models. It is acknowledged that the results presented will be preliminary and that model calibration will need to be performed to yield validated results. Nevertheless, the results do provide an indication on the peak flow reduction effectiveness of the ABC Waters design features (and innovative design adopted for the features) at the precinct, and in a tropical climate setting.

3. Research Methodology

The general workflow that was adopted for the research is as shown in Figure 7. More details on the specific research activities that were carried out for each task are as outlined in the following sub-sections.

3.1. Data Collection

Prior to setting up the model for the precinct and its suite of ABC Waters design features, design data and general information regarding the precinct and the features were collected. As the final as-built drawings of the precinct and features were not available during the duration of the research, the latest set of construction drawings as obtained from PUB were used for developing the model. Additional information such as rainfall intensity-duration-frequency (IDF) curves and historical rainfall data were also obtained for assessment of the model performance.

3.2. Model Selection

The Storm Water Management Model version 5.1.010 (SWMM v5.1), an open source and freely downloadable hydraulic model, with LID controls developed by the U.S. Environmental Protection Agency (USEPA) was selected and used for the 1D hydrodynamic model study [28]. For Waterway Ridges, backwater effect and reverse flow can occur within the perforated pipes (connecting the gravel storage to the overflow manhole) and orifice outlets of the rain gardens and gravel swales, thus necessitating the use of dynamic-wave hydraulic routing. Due to the small size of the subcatchment and ABC Waters design features (the smallest subcatchment and ABC Waters design feature have areas of 34 m² and 18 m², respectively, while the average areas for the subcatchments and ABC Waters design features are 632 m² and 80 m², respectively), sub-hourly time steps in the order of minutes and seconds would also apply for the variation of runoff and associated detention processes in the precinct. These requirements are met by the SWMM model used.

3.3. Preliminary Analysis, Sensitivity Analysis and Model Parameters

Prior to setting up the model for the precinct, preliminary assessment was carried out to determine the best ways to represent the various ABC Waters features in Waterway Ridges. Based on the SWMM Applications Manual [29] as reference, different ways of modelling the subcatchments and features were tested, and the outflow results from the features were compared and analysed. Sensitivity analysis was also performed to determine the sensitivity of runoff from the subcatchments to the various catchment parameters [7,29]. Figure 8 shows a screenshot of the completed 1D SWMM model for the precinct.

3.4. Hydraulic Routing Model and Time Steps

The dynamic-wave flow routing was necessary as backwater effects and reverse flow would take place within the perforated pipes and orifice outlets of the rain gardens and gravel swales. Time step recommendations as listed in the SWMM Applications Manual [29] were adhered to, and model stability was also verified for the time steps used. The time steps for dry weather flow, wet weather flow and routing were set at 1 h, 1 min, and 10 s, respectively. Typically, the wet weather time step and flow routing time step should not exceed the precipitation recording interval and wet weather time step, respectively [29]. The flow routing time step used is less than 30 s due to the dynamic-wave flow routing that is employed in the model, and was adjusted until the stability of the model was satisfactory.

3.5. Scenarios

In addition to the base scenario of the precinct model (i.e., base precinct model with its suite of ABC Waters design features that are coupled with gravel storage and orifice outlets), additional scenarios as summarised in

Table 1. Summary of base and alternative scenarios of the precinct which were modelled and analysed.

S/N	Scenario Description	Remarks
1	Precinct without any detention (i.e., conventional design without ABC Waters features, gravel storage or orifice outlets)	To assess performance of the conventional design
2	Precinct without ABC Waters features but with orifice outlets from the respective subcatchments	To assess effectiveness of orifice outlets alone in reducing peak flow
3	Precinct with typical design of ABC Waters features implemented in Singapore (i.e., rain gardens and swales without orifice outlets and gravel storage layer)	To assess performance of the precinct if typical design of ABC Waters features was used
4	Precinct with ABC Waters features coupled with orifice outlets but without gravel storage	To determine importance of gravel storage in reducing peak flow
5	Precinct with ABC Waters features coupled with gravel storage but without orifice outlets	To determine importance of orifice outlets in reducing peak flow
6	Precinct with ABC Waters features coupled with gravel storage and orifice outlets	Base model that corresponds to the actual constructed precinct

were also modelled and analysed. The rationale for modelling the various scenarios is also elaborated in the table.

3.6. Assessments

3.6.1. Peak Flow Reduction and Compliance with Peak Flow Reduction Objectives

Symmetrical 4 h 10-year and 3-month design rainfall hyetographs in 5-min intervals (Figure 9) derived from Singapore’s rainfall IDF curves were routed through each of the models for the respective scenarios. The peak flow from the precinct for each of the scenarios were compared and the corresponding runoff coefficients or C-values were back-calculated using Rational Formula (Equation (1)) [30]. Specifically for the base precinct model, the C-value was compared against the design objective of having an effective runoff coefficient of 0.55. As there are four discharge outlets from the precinct, the peak flow from the precinct was obtained by taking the maximum of the sum of the time series flow results from each of the outlets.

$$Q_r=\frac{1}{360}CIA$$

(1)

where

• Q_r = peak runoff (m³/s)
• C = runoff coefficient (dimensionless)
• I = average rainfall intensity (mm/h)
• A = catchment area (ha).

3.6.2. Effectiveness of Various ABC Waters Design Feature Types in Reducing Peak Flow

The maximum flows entering and leaving each feature were obtained from the base precinct model for the 10-year design storm, and the corresponding peak flow reductions were calculated using Equation (2). Averages of the peak flow reductions for the various design feature types were then compared to determine if any of the feature types are more effective in reducing peak flow as compared to others.

$$Peakflowreduction(\%)=\frac{maximuminflow-maximumoutflow}{maximuminflow}\times 100\%$$

(2)

However, it was also noted that the peak flow reduction alone would not be a fair comparison across the different features and even within the same type, as the features have different catchment areas and storage volumes. As such, the peak flow reductions were adjusted using the ratio of the storage volume (both surface detention and gravel storage for the rain gardens, and gravel storage alone for the gravel swales which do not have any surface detention) to the effective impervious catchment area of each feature (using Equations (3)–(5), and then normalised (using Equation (6)). Averages of the normalised peak flow reductions for the various feature types were compared to assess their peak flow reduction effectiveness.

$$\begin{array}{c}Effectivecatchmentarea\hfill \\ =1\times imperviouscatchmentarea+0.45\hfill \\ \times perviouscatchmentarea\hfill \end{array}$$

(3)

$$Adjustmentratio\left(\mathrm{m}\right)=\frac{Totalstoragevolumeoffeature\left({\mathrm{m}}^3\right)}{Effectivecatchmentareaoffeature\left({\mathrm{m}}^2\right)}$$

(4)

$$\begin{array}{c}Adjustedpeakflowreductionfor{i}^{th}feature(\%)\hfill \\ =\frac{Largestadjustmentratioamongallfeatures}{Adjustmentratiofor{i}^{th}feature}\hfill \\ \times Peakflowreductionfor{i}^{th}feature\hfill \end{array}$$

(5)

$$\begin{array}{c}\mathit{Normalised}\mathit{peak}flow\mathit{reduction}\mathit{for}{i}^{th}\mathit{feature}(\%)\hfill \\ =\frac{Adjusted\mathit{peak}flow\mathit{reduction}\mathit{for}{i}^{th}feature-Smallestadjustedpeakflowreduction}{Largestadjustedpeakflowreduction-Smallestadjustedpeakflowreduction}\hfill \end{array}$$

(6)

4. Results

4.1. Peak Flow Reduction for the 10-Year Design Storm

The outflows and effective C-values (of the base precinct and the various alternative scenarios) for the 10-year design storm are as shown in Figure 10 and

Table 2. Outflow and effective C-value results of the precinct and alternative scenarios for the 10-year design storm.

S/N	Label	Scenario	Qmax (m3/s)	Reduction * in Qmax	Effective C-Value
1	No treatment/detention	Precinct without any treatment or detention	2.13	0%	0.89
2	Orifice only	Precinct with orifice outlets only	2.13	0%	0.89
3	Typical ABC	Precinct with typical design of ABC Waters features implemented in Singapore	1.85	13.0%	0.77
4	ABC + Orifice	Precinct with ABC Waters features and orifice outlets but without gravel storage	1.86	12.6%	0.77
5	ABC + Storage	Precinct with ABC Waters features and gravel storage but without orifice outlets	1.75	17.7%	0.73
6	ABC + Storage + Orifice	Precinct with ABC Waters features coupled with gravel storage and orifice outlet	1.44	32.6%	0.60

Note: * Relative to Scenario 1 of the precinct without any treatment or detention.

, respectively. From the results, we can see that the best C-value of 0.60 with a corresponding reduction of 32.6% in peak runoff (compared to Scenario 1 where there are no detention or treatment present in the precinct) is obtained in Scenario 6, i.e., when the base precinct with ABC Waters features are coupled with gravel storage and orifice outlets. Although the reduced C-value and peak runoff achieved do not meet the runoff control requirements of Singapore’s COP on Surface Water Drainage, the results show that ABC Waters design features are effective in reducing peak runoff when coupled with storage and orifice outlets, and designed appropriately. It is also noted that the peak flow reduction obtained for the base precinct is generally lower as compared to the reductions obtained by the various bioretention systems that have been monitored in the USA [15,17,19,20,21] and Australia [22,23].

Meanwhile, it can be observed that the performance of Scenario 2 (precinct with orifice outlets only) is as poor as that of Scenario 1, indicating that there is no benefit in implementing orifice outlets that are not coupled with storage or any ABC Waters design feature.

While the performance of Scenario 3 (precinct with typical design of ABC Waters features implemented in Singapore) is not as good as that of Scenario 6, it still yielded a C-value of 0.77 and a peak runoff reduction of 13% (as compared to Scenario 1). This observation shows that the typical ABC Waters features implemented in Singapore (i.e., without storage and orifice outlets) could still contribute in helping to reduce peak runoff in a development even though they are not able to meet the runoff control requirements of the COP on Surface Water Drainage and will need to be implemented together with other features.

4.2. Effectiveness of the Various ABC Waters Design Feature Types in Reducing Peak Flow for the 10-Year Design Storm

Table 3. Peak flow reduction performance of the various feature types for the 10-year design storm.

Design Feature Type	Average Peak Flow Reduction (%)	Average Normalised Peak Flow Reduction (%)
Rain gardens *	39%	47%
Gravel swales	9%	25%
Vegetated swales	0%	0%

Note: * Excluding rain gardens without orifice outlets (i.e., rain gardens FB6, FB10, FB15 and FB17).

summarises the performance of the various feature types in reducing peak flow. From the results, we can see that the rain garden feature type is most effective in reducing peak flow (with average and normalised reductions of 39% and 47%, respectively), followed by the gravel swales (with average and normalised reductions of 9% and 25%, respectively). This result could be attributed to the presence of the 0.4 m thick filter media and extended surface detention in rain gardens but not the gravel swales. The results also show that the vegetated swales do not offer any reduction in peak flow, which is attributed to the lack of gravel storage or any form of surface detention due to the absence of flow regulators (such as weirs) along the vegetated swales.

However, this does not conclude that vegetated swales should not be implemented. While peak flow reduction is the focus of this study, it is noted that ABC Waters design features including vegetated swales also offer other ecosystem services such as improving runoff water quality, enhancing biodiversity [4] and regulating micro-climate via the urban heat island mitigation effect [32].

4.3. Identification of Key Design Elements

It can be observed from the peak flow reduction results from Section 4.1 that the performance of Scenario 5 (precinct with ABC Waters features and gravel storage but without orifice outlets) with a C-value of 0.73 (and peak runoff reduction of 17.7%) is slightly better than that of Scenario 4 (precinct with ABC Waters features and orifice outlets but without gravel storage) with a C-value of 0.77 (and peak runoff reduction of 12.6%). When the orifice outlets and gravel storage are coupled together however, the performance improves to yield a C-value of 0.60, which is an improvement of 18.2% and 22.9% for Scenarios 5 and 4, respectively. It can be concluded from this observation that the orifice outlets and gravel storage are important components of the ABC Waters features (specifically the rain gardens and gravel swales) that will dictate their performance in reducing peak runoff effectively.

4.4. Peak Flow Reduction for the 3-Month Design Storm

The outflows and effective C-values (of the base precinct and the various alternative scenarios) for the 3-month design storm are as shown in Figure 11 and

Table 4. Comparison of the outflow and effective C-value results of the precinct and alternative scenarios for the 3-month and 10-year design storms.

S/N	Scenario	3-Month Design Storm			10-Year Design Storm
		Qmax (m3/s)	Reduction * in Qmax (%)	C-Value	Qmax (m3/s)	Reduction * in Qmax (%)	C-Value
1	No treatment/detention	0.78	0%	0.89	2.13	0%	0.89
2	Orifice only	0.78	0%	0.89	2.13	0%	0.89
3	Typical ABC	0.49	36.8%	0.56	1.85	13.0%	0.77
4	ABC + Orifice	0.49	37.3%	0.56	1.86	12.6%	0.77
5	ABC + Storage	0.44	43.2%	0.51	1.75	17.7%	0.73
6	ABC + Storage + orifice	0.42	46.7%	0.47	1.44	32.6%	0.60

Note: * Relative to Scenario 1 of the precinct without any treatment or detention.

, respectively. Table 4 also includes the results of the 10-year design storm for comparison. Similar to the results of the 10-year design storm, the constructed base precinct with ABC Waters features that are coupled with gravel storage and orifice outlets achieved the lowest C-value of 0.47. The other trends as earlier observed for the 10-year design storm also apply for the 3-month design storm. In addition (and as expected), the peak flow for all the scenarios are below that of the maximum allowable peak runoff (which is calculated based on the 10-year design storm).

Further comparisons of the peak flow difference between the worst and best performing scenario (i.e., Scenarios 1 and 6) were also made for the 3-month and 10-year design storms in the zoomed-in plot shown in Figure 12. From the figure, we can see that the peak flow difference is larger for the 3-month design storm (at 46.7%) as compared to the 10-year design storm (at 32.6%), suggesting that the ABC Waters design features implemented are more effective in reducing peak flow for smaller storm events as compared to larger ones. Similar observations were also reported in other bioretention studies [16,17,18,22]. This is generally expected for typical rain gardens (since smaller storm events would result in less surface overflow as compared to larger storm events) and is hence a somewhat surprising observation considering the design of the rain gardens and gravel swales in the precinct which are coupled with underground gravel storage and orifice outlets.

4.5. Performance Enhancement through Optimisation

The original base precinct has an effective C-value of 0.60 and does not meet the requirements of the runoff control legislation in Singapore. This performance can be attributed to the non-optimal design of the orifice outlets in the precinct. To verify this, optimisation was performed for two scenarios, namely (1) Single-objective optimisation of the resultant C-value of the precinct by varying the diameter of the orifice outlets and (2) Multi-objective optimisation of the C-value and cost of gravel layer cum orifice outlets by varying the gravel storage layer depths and diameter of the orifice outlets. More information on the optimisation can be found in Appendix A.

The single objective optimisation showed that the effective C-value of the precinct could be improved to 0.53. The optimised C-value is 11.7% less than the non-optimised C-value of 0.60 with no increase in cost (since the gravel storage layer depths are the same), and is lower than the 0.55 that is stipulated by the runoff control legislation in Singapore. The multi-objective optimisation also resulted in a similar C-value of 0.53 at an additional cost of S$1610 (~$1180).

A separate optimisation study was also performed on an arbitrary precinct model to determine if there is an optimum rain garden area and treated catchment area (in terms of percentage of total site area) that would allow catchments to achieve C = 0.55 for the 10-year design storm. A catchment area of 4 ha was adopted for the study and three land use types were considered; residential, commercial and mixed (50% residential–50% commercial). Runoff from treated parts of the catchment is channelled into a lumped rain garden with gravel storage and orifice outlets based on the design adopted at Waterway Ridges. The study showed that there is an optimum treated catchment area that would result in the lowest possible C-value for each rain garden size, and beyond which the peak flow reduction performance would deteriorate (Figure 13). It was found that a minimum rain garden size of 6% the total site area with treated catchment areas of 45.2–47.2% (depending on land use type) was needed to meet the runoff control legislation in Singapore. More information on the study is provided in Appendix B.

5. Discussion

The results prove that the innovative design implemented for the rain gardens and gravel swales at Waterway Ridges, whereby the features are coupled with underground gravel layers for storage and orifice outlets for restricting flow, is indeed workable. These two key design elements (i.e., the gravel storage and orifice outlets) significantly improve the runoff control effectiveness of rain gardens and gravel swales, as can be seen when the precinct scenario with the typical design of ABC Waters design features implemented in Singapore (i.e., without gravel storage and orifice outlets) was only able to achieve an effective C-value of 0.77. By being able to meet the runoff control requirements in Singapore, the new design allows ABC Waters design features to be considered as another effective option in reducing peak flow, in addition to other structural measures such as underground detention tanks.

Upon comparing the performance of ABC Waters design features between smaller and bigger storm events, it was found that ABC Waters design features were more effective in reducing peak flow for the smaller storm events. This was substantiated by peak flow reduction comparisons of the precinct for the 3-month and 10-year design storms, which showed that there is a larger peak flow reduction for the 3-month design storm (at 46.7%) as compared to the 10-year design storm (at 32.6%). While this was expected for typical rain gardens (since smaller storm events would result in less surface overflow from rain gardens as compared to larger storm events), it was a somewhat surprising observation considering the design of the rain gardens and gravel swales in the precinct which are coupled with underground gravel storage and orifice outlets.

Meanwhile, the precinct optimisation results show that the effectiveness of the ABC Waters design features in reducing peak flow and in complying with the runoff control requirements for Singapore can be enhanced. The non-optimised ABC Waters design features in Waterway Ridges were able to reduce the effective C-value of the precinct to 0.60 for the 10-year design storm, a reduction of 32.6% in peak flow when compared to the scenario of the precinct without any treatment or detention features. The peak flow reduction performance of the precinct improved considerably when the orifice outlet sizes were optimised using evolutionary algorithm, resulting in an effective C-value of 0.53 which is 11.7% better than the non-optimised C-value of 0.60 obtained from the base precinct model, at no increase in cost. More importantly, this optimised C-value is lower than the 0.55 stipulated by the runoff control legislation in Singapore. The effective C-value can also be lowered slightly below 0.53 when the gravel storage layer depths were optimised together with the orifice outlet sizes, albeit at an additional cost of S$1610 (~$1180) or 3.1% more (as compared to the cost of gravel layer and orifice outlets in the original base precinct).

The study on optimum rain garden area and treated catchment area (that would allow catchments to achieve C = 0.55 for the 10-year design storm) found that a minimum rain garden size of 6% the total site area, with treated catchment areas of 45.2–47.2% depending on land use, would be required. The recommended rain garden size is only slightly higher than the 4–5% (of equivalent impervious catchment area) which is recommended by PUB [8] in order to meet the storm water quality objectives for Singapore [4]. The 6% is also comparable to the results of Waterway Ridges with optimised orifice outlets, whereby a C-value of 0.53 was obtained with an ABC Waters design feature area (excluding batter slopes for the rain gardens) of 5.4% the total precinct area.

Unlike underground detention tanks, ABC Waters design features (which are natural and sustainable) provide other ecosystem services such as the enhancement of biodiversity and improvements in runoff water quality [4]. The combined design of rain gardens and vegetated swales with underground gravel storage layers also reduces the risk of drowning and mosquito nuisance which are present in open detention systems. The incorporation of green infrastructure in urban areas further adds monetary value to a precinct as there is a willingness by the public to pay more for properties in precincts with green infrastructure; for example, a survey of Australian households reveal that buyers are willing to pay an additional $36,000–54,000 per property in streets with rain gardens [33].

The innovative design of the rain gardens and gravel swales (which are coupled with underground gravel layers for storage and orifice outlets to restrict flow) can also be modified to effectively reduce peak flow in countries beyond Singapore. This would apply in particular to countries with the same tropical climate, many of which are still developing. This design is also relevant for temperate regions in Europe, where the usage of Sustainable urban Drainage Systems (SuDS) is prevalent. The findings of this paper can be used to highlight ABC Waters design features as a viable sustainable option to governments of countries where the extensive use of such features have previously not been considered for storm water management.

6. Conclusions

This research was aimed at determining the effectiveness of Singapore’s ABC Waters design features in reducing peak flow during storm events. Hydraulic performance assessment of the innovative design of ABC Waters design features was performed using a 1D SWMM model setup of the Waterway Ridges pilot precinct. The analysis revealed that ABC Waters design features are effective in reducing peak flow and runoff coefficient of the precinct during storm events. The effectiveness is attributed to an innovative design, which comprises the coupling of rain gardens and vegetated swales with underground gravel layers for storage and orifice outlets to restrict outflow. The reduction in peak flow (and effective runoff coefficient i.e., C-value) is 33% and 47% for the 10-year and 3-month design storms, respectively, when compared to the scenario whereby no treatment or detention is performed. When compared with typical designs adopted for the ABC Waters design features, there is also an improvement in peak flow reduction by 23% and 16%, respectively, for the 10-year and 3-month design storms.

The performance of the hybrid ABC Waters design features in terms of reducing the effective C-value of the precinct can also be optimized through design parameters such as orifice outlet diameter and gravel layer thickness; such an optimisation was performed on the precinct which resulted in a further reduction of the C-value by 12%. The reduction of effective C-value via the use of ABC Waters design features can also be improved by optimising the treated catchment area channelled to the feature. In addition to the decrease in peak flow and reduction in runoff coefficient, the innovative design reduces the risk of drowning as well as mosquito breeding when compared against structural detention measures and open detention systems, and also exhibits the other benefits of ABC Waters design features such as in providing ecosystem services and enhancing biodiversity.

It is acknowledged that the results obtained are preliminary and that calibration of the models will need to be carried out to provide validated results; this will be performed when monitoring results become available. Nevertheless, the preliminary results are very encouraging in promoting the use of the modified ABC Waters design features for peak flow and C-value reduction in Singapore and other tropical countries, as well as in the temperate regions of the world where usage of SuDS/LID/WSUD are prevalent.