Advertisement

Water Resources Management

, Volume 32, Issue 15, pp 4933–4951 | Cite as

Optimal Design of LIDs in Urban Stormwater Systems Using a Harmony-Search Decision Support System

  • F. De PaolaEmail author
  • M. Giugni
  • F. Pugliese
  • P. Romano
Article

Abstract

During the last years, climate changes and urbanization are causing huge urban pluvial flood events in many countries in the world, driving to both develop and apply effective and innovative approaches for the design and management of urban stormwater systems. The gradual urbanization is provoking the increase of impervious surfaces and, consequently, of surface runoff and velocity and the reduction of concentration times of watersheds, both increasing soil erosion and worsening the water quality as a consequence of the intensive contamination. In this field, Low Impact Development (LID) practices for urban runoff control can be intended as an effective approach to both improve the urban resilience against the flooding risk and assure environmental interventions to adequate the urban stormwater systems to both climate and land use changes. In this paper, a Decision Support System (DSS) for the optimal design of LIDs in urban watershed is presented and discussed. The procedure is tested on Fuorigrotta (IT) and Ponticelli (IT) urban watersheds, with the aim of assessing the effectiveness of LIDs application in reducing both the flooded and conveyed volumes, at the expense of cost-effective solutions.

Keywords

Low impact development Best management practices Meta-heuristic model Harmony search Urban stormwater systems Decision support system 

1 Introduction

During the last years, climate and land use changes are provoking the worsening of urban flash floods. Having rapid onset causing building and businesses collapses and damages with harm to people and, in the most dangerous cases, fatalities (Hapurachchi et al. 2011), there is only low opportunity for effective response.

Many models in the literature are based upon the estimation of the flash flood risk and the spatio-temporal detection of flash flood occurrences (Špitalar et al. 2014), taking into account the use of Flood Forecasting and Early Warning Systems (Alfieri and Thielen 2015), social vulnerability (Terti et al. 2015) and risk communication (Lazrus et al. 2016).

In this field, the development of mitigation and adaptive approaches to face the ongoing urban flash floods represents one of the most challenging issue in the management of urban stormwater systems.

Many studies highlighted how the modification of land use and the increase of impervious surfaces (due to building roofs, parking areas and roads) is provoking extreme flooding in urban areas, undermining the water quality in receiving water bodies (Khan et al. 2006; Dietz 2007). Nevertheless, the urbanization is a worldwide process involving many countries, included the low-income ones (UN 2017). With reference to upper-middle and high-income countries, several studies (DESA 2015) predicted the increase of urbanization during the next thirty years up to percentages of 80 and 75% of areas having urbanization rates (percentage of urban population) higher than 50 and 75%, respectively. This trend is significantly affecting the natural ecosystems, causing severe future forcing (Du et al. 2012; Hager et al. 2013; Jia et al. 2013), due to the increase of the runoff coefficient and the modification of the hydrological cycle, significantly mining the human safety and fruition of water resources.

Moreover, the ongoing climate change represents a preeminent cause of increasing severe storm events, with significant consequences for infrastructures and flood protection projects (Yilmaz et al. 2014; Field et al. 2014). However, in spite of many studies focused on the influence of climate changes and, basically, of extreme rainfalls, to both design and operate urban stormwater infrastructures, to date there are still many limitations in their effective applications (Arnbjerg-Nielsen et al. 2013).

Significant changes are founding in the design and management of stormwater systems because the current approaches are mostly devoted to environmental, economic and social perspectives (Fletcher et al. 2015). With the aim of mitigating the impacts of both urbanization and climate changes, the application of Low Impact Development (LID) approaches is meaningfully taken into consideration (US EPA 2004; Davis 2005; De Martino et al. 2012; De Paola et al. 2017a), because able to combine environmental and nature-based aspects with technical effectiveness. Nowadays they represent a recognized environmental approach to improve the quali-quantitative measurement, control and management of stormwater runoff, assuring cost-effective responses (US EPA 2004).

Several LIDs are presently available. Among them vegetative swales, bioretention cells, rain gardens, green roofs, infiltration trenches, porous pavements, rain barrels and rooftop disconnection are managed and modeled by the EPA SWMM 5.1 hydraulic solver (Rossman 2017).

Each technique owns technical (removal efficiency, size, etc.) and non-technical (cost, environmental impacts, disposal capacities, etc.) properties with high rate of uncertainties (Li et al. 2013). The selection of suitable LIDs thus represents a complex but preeminent step for effective and sustainable management of urban stormwater systems.

Moreover, their planning management is strictly related to the analysis of relevant urban development planning schemes, such as urban master planning, land use, landscape, drainage system, water pollution control planning, etc. In this field, the LIDs implementation in urban watersheds requires the detection of the most technical and cost-effective selection, design and location, able to accomplish the whole set of required goals and objectives (Cheng et al. 2009; Jia et al. 2012, 2015).

Aimed at providing effective tools to estimate the LIDs reliability, several models and optimization tools are available in the literature (Elliott and Trowsdale 2007). Among them, Zhang et al. (2013) implemented a multi-objective optimization framework for peak flow management in the Fox Hollow (USA) Watershed. The EPA SWMM model was coupled with the ε-NSGAII algorithm for developing 37 LID scenarios. The bioretention application resulted the most frequently selected, because able to treat the runoff from the surrounding areas, providing significant hydrologic benefits.

Gulbaz and Kazezyilmaz-Alhan (2015) applied the EPA SWMM 5.1 solver to assess the LIDs effectiveness to reduce the surface runoff and the Total Suspended Solids (TSSs) in the Sazlidere Watershed in Turkey, by applying retention basins, vegetative swales and permeable pavements. Jia et al. (2012) applied an optimization procedure, by coupling the EPA SWMM solver with the BMP DSS, to design LIDs in the Beijing Olympic Village (CH) observing the capability of the developed approach to reduce both the runoff volume and the peak rates. Liu et al. (2016) combined the hydrologic/water quality model (L-YHIA-LID 2.1) with an optimization algorithm for the optimal selection and location of LIDs in the Crooked Creek (US) watershed. It was observed that for sites with variable features significantly different benefits could be achieved by using LID practices. Kim et al. (2017) promoted an online survey in Texas (US) to assess the stakeholders’ perception about the application of LIDs and the limits deriving from lack of policies and regulations. Among the considered stakeholders, the urban planners showed the most expertise in LIDs, against the developers who resulted not highly familiar with their application. Nevertheless, the significant interest from experts to develop LIDs was observed, however needing proper policy assistance and education support.

Ahmed et al. (2017) implemented a web-based DSS for the EPA SWMM 5.1 application to design LID practices in the urban catchment of the campus of the University Technology Malaysia. The Water Management Analysis Module (WMAM) was applied to compare results from different scenarios, with and without the LID installation. A sensitivity analysis was performed to detect the best LIDs design. Results showed the LID capability to limit the peak flow, at the same time resulting an interesting approach during the planning and management activities in urban watershed from decision-makers.

In order to overcome the limitation of available models for LIDs design and management, in this paper a DSS for the optimal LIDs design in urban stormwater systems, implemented by De Paola et al. (2017a), was tested on two urban watersheds in the Municipality of Naples (IT). The developed model was based on the interface between the meta-heuristic model Harmony Search (Geem et al. 2001) and the EPA SWMM 5.1 hydraulic solver, aimed at optimally combining the LIDs application in urban stormwater systems. The DSS searched for the technical solutions able to reduce the flooding volumes in the urban stormwater systems by minimizing the Objective cost Function (OF). With respect to the preliminary application in De Paola et al. (2017a), the present work aims to both improve the territorial analysis and refine the modelling of sewer systems, by testing the implemented procedure to more complex configurations.

2 Methodology

The applied meta-heuristic model is the natural-based Harmony Search (HS) algorithm, implemented by Geem et al. (2001) as an optimization procedure able to numerically reproduce the music process of playing a musical harmony through a jazz improvisation. Indeed the musical harmony is specified by a solution vector and schemes of optimization techniques are able to reproduce the musician’s improvisation.

The HS model analytically duplicates the “perfect state” of harmony, similarly to a jazz improvisation, by reproducing an aesthetic estimation, quantified through the calculation of the Objective Function (OF). The final harmony is achievable by playing an enough number of practices, represented by the set number of iterations in the algorithm. The standard procedure of the HS algorithm is composed of five steps (Geem et al. 2001):
  • Step 1: Initialization and setting of HS parameters: both the introduction and initialization of parameters are carried out. They are the Harmony Memory Considering Rate (HMCR), the Pitch Adjusting Rate (PAR), the Harmony Memory Size (HMS: number of solution vectors) and the termination criterion. The latter corresponds to the total number of improvisations (i.e. the number of iterations). HMCR and PAR are set in Step 3 to modify the solution vectors to maximize/minimize the OF. Being a Multi-Objective algorithm, the Pareto Front is defined, composed of the solutions not dominated by any other solution in the population;

  • Step 2: Harmony Memory (HM) initialization: a number of randomly generated solution vectors equal to HMS is introduced into the HM matrix:

$$ HM=\left[\begin{array}{ccc}{x}_1^1& \cdots & {x}_N^1\\ {}\vdots & \vdots & \vdots \\ {}{x}_1^{HMS}& \cdots & {x}_1^{HMS}\end{array}\right]{\displaystyle \begin{array}{c}\to \\ {}\vdots \\ {}\to \end{array}}{\displaystyle \begin{array}{c}f\left({x}^1\right)\\ {}\vdots \\ {}f\left({x}^{HMS}\right)\end{array}} $$
(1)
  • Step 3: New harmony improvisation from HM: the generation of a new harmony vector x′ = (x′1, x′2,..., x′N) is performed as a function of three main rules, namely the Memory Considerations, the Pitch Adjustments and the Randomization. The first decision variable (x′i) can be applied choosing among any value in the HM range (xi1 ̴ xiHMS). The probability of selecting a completely random value for the decision variable x′i is also introduced:

$$ x{\hbox{'}}_i=\left\{\begin{array}{l}x{\hbox{'}}_i\in \left\{{x}_i^1,{x}_i^2,\dots, {x}_i^{HMS}\right\}\kern1.5em \mathrm{with}\ \mathrm{probability}\ HMCR\\ {}x{\hbox{'}}_i\in {\mathbf{X}}_i\kern7.5em \mathrm{with}\ \mathrm{probability}\ \left(1\hbox{-} HMCR\right)\end{array}\right. $$
(2)
where HMCR varies from 0 to 1. It defines the probability of selecting one value from the historical values stored into the HM, whereas the complement (1-HMCR) is the probability of choosing a random feasible value, not limited to those stored into the HM. Further decision variables (x′2,..., x′N) can be chosen by applying the same procedure. Then, after the Memory Considering Operation, the Pitch Adjusting Operation operator is applied to set the evolving rate through the neighboring values starting from the initial setting of the HM;
  • Step 4: HM update: the HM update is performed by substituting the new harmony vector x′ = (x′1, x′2,..., x′N) if better, in terms of the OF, than the worst harmony in the HM. This last is thus deleted;

  • Step 5: Application of the stopping criterion: the optimization is concluded once the stopping criterion (i.e. the number of set iterations) is achieved. If not, the algorithm comes back to Steps 3 and 4.

Starting from the abovementioned procedure, during the last years the HS was applied to many fields, such as logistic and programming optimizations (Geem et al. 2005; Al-Betar and Khader 2012), power energy optimization (Coelho and Mariani 2009; Turk and Radeke 2011), pump scheduling (De Paola et al. 2016a, 2017b), location and setting of Pressure Reducing Valves in water networks (De Paola et al. 2017c), optimal design and calibration of urban sewer systems (De Paola et al. 2015, 2016b).

Specifically, De Paola et al. (2017a) proposed an operative procedure for the optimal design of LID practices in urban stormwater systems, according to the flowchart summarized in Fig. 1.
Fig. 1

Flowchart of the applied procedure for LIDs design (De Paola et al. 2017a)

The first step consists in the implementation of a territorial analysis by using Geographic Information System (GIS) tools. The urban and land use properties of the study area are defined, by assessing the topographic properties of the basin, combined with information about the urban stormwater systems and the watershed lines. The percentage range of areas convertible to LIDs is quantified, as a function of the properties of the considered basin. Specifically, the land use map and the related restrictions imposed by the planning governance instruments are analyzed to identify areas prone to LIDs. The attention is thus paid to streets, urban open spaces, roofs, parking lots and abandoned areas, making a distinction between private and public properties.

The second step consists in the generation, through an automatic GIS tool, of the EPA SWMM file (with extension .inp) to simulate the Baseline Scenario (BS), without any LIDs intervention. The conveyed volume (V0) is thus estimated for the BS and the related flooding volume from the stormwater drainage system (V’0). The hydrologic input data can be represented by either a synthetic hyetograph or a continuous rainfall record. In this work the first option is taken into account, with reference to the Intensity-Duration-Frequency (IDF) curves of the urban water catchment location. In the third step, the .inp file is automatically updated by locating the selected LIDs. The LIDs location is employed by assessing the current and potential areas transformable with LIDs. Moreover, with reference to streets, the modification of the roadways is employed by analyzing the vehicular traffic, in order to identify the main urban crossing routs.

Here the HS optimization procedure is applied according to the abovementioned Steps. The OF is set to minimize the realization costs of the whole set of chosen LIDs, in agreement with the fixed constraints (maximum LID extension).

New hydraulic simulations are thus performed by using the EPA SWMM 5.1 hydraulic solver, as a function of the new generated .inp file. New conveyed V and flooded V′ volumes are estimated and the performance indicators r = V/V0 and r’ = V′/V’0 are assessed. The realization costs are set in agreement with Zhang et al. (2013), by accounting for the solely actualized construction costs.

At the last Step, a solution set of the selected LIDs is provided and, by fixing an allowable budget, the choice of the feasible solution able to minimize r and r’ is applied. The operative procedure for the design of LIDs is summarized in Fig. 2.
Fig. 2

Flow-chart of the implemented structure for optimal LIDs design

In this work the proposed procedure is tested on two case-studies referring with urban watersheds in the Municipality of Naples (IT), aimed at assessing the model reliability to optimise the LIDs application in urban contests to reduce the flooded and discharge volumes, at the expense of cost-effective solutions.

3 Study Areas

Two case studies are analyzed: the urban watershed of Fuorigrotta district in Naples (IT), introduced by De Paola et al. (2017a) and the urban watershed of Ponticelli (NA) district.

The Fuorigrotta (NA) urban watershed has area of 1.89 km2 with perimeter of 6975 m (Fig. 3). The boundary limit is represented by Via Terracina in the northern side, by Posillipo Hill in the east and south sides and by Mostra d’Oltremare building in the west side.
Fig. 3

Fuorigrotta (NA) urban watershed with land use

The territorial analyses are developed by using the UTM WGS84 33 N multi-precision topographic geo-database, drawn up by the Regione Campania Authority in 2004 and updated to the current configuration.

The land use, the impervious surface partition, the urban traffic flows and the location of urban areas potentially convertible with LIDs are both detected and analysed. The GIS Hydrologic Tools commands operating on a Digital Terrain Model (DTM) from Regione Campania Authority are applied to assess altitudes, slopes and extension of the considered basin in order to define the subcatchment properties to be implemented in EPA SWMM 5.1, namely the area extension, the equivalent width and the required parameters to define the soil permeability.

The Fuorigrotta (NA) district is composed of an urban stormwater system built in 1950, with following integrations and improvements. The main sewer is the Arena Sant’Antonio which conveys the whole rainwater of the Fuorigrotta and Pianura districts, whereas the Cuma outflowing stream conveys the wastewaters.

In the implemented study, the urban watershed is divided into two hydrographic sub-basins, namely the North and the South Basins, flowing into the Arena Sant’Antonio main sewer and into the Via Diocleziano sewers, respectively. The sub-basins repartition is employed by applying the GIS Hydrologic Tools commands to develop the watershed and the water drop analysis.

This system only manages the stormwater volumes, whereas the wastewaters flow across the Cuma main sewer, located at 12 m depth from the ground level (De Paola et al. 2017a).

The second case-study infers the Ponticelli (NA) urban watershed, a hydrographic basin managed by the Autorità di Bacino (AdB) of Campania Centrale. It has the highest urbanization index in terms of housing density and productive-commercial activities among all the basins of the Campania Region, showing, from the morphological point of view, significant hydro-geological, volcanic and seismic vulnerabilities.

The current urbanization asset is due to the transformation processes which, starting from the ‘80 years, have provoked significant modifications of territory and ecological equilibriums. These factors imply the presence of diversified housing structures, having areas with huge anthropization and industrial settlements, combined with historic, landscape and natural zones (AdB Campania Centrale 2013).

The studied area is a portion of the Volla Basin, having extension of about 20 km2 of the drained area of Sebeto river. It is composed of a complex system of artificial drains, conveying rainfall waters from Monte Somma and Volla and Poggioreale plane. Two main systems are detected: the first, served by Sbauzone sewer (about 5 km), includes the Pollena, Trocchia and Zazzera basins, whereas the second one drains the Volla plane, through the Cuozzo and Reale channels, flowing the rainfall waters into the Napoli harbour (Fig. 4).
Fig. 4

Ponticelli (NA) urban watershed detection (Comune di Napoli 2005)

During the modelling the studied area is divided into 17 sub-basins, estimating the main parameters, namely the area, the width, the average slope, the imperviousness and the Manning coefficient. The BS is also reproduced by GIS applications, by both extrapolating physical parameters from the Digital Terrain Model (DTM) and elaborating the natural watershed directions using Hydrology Tool commands from the DTM (Fig. 5).
Fig. 5

Ponticelli (NA) basin with natural watersheds

The land use of each basin is thus derived (Fig. 6) to both specify the intended use of the studied area and detect the portions potentially convertible to LIDs. In Table 1 properties of each sub-basin are summarized and detected in Fig. 7, with reference to the BS.
Fig. 6

Ponticelli (NA) land use map - Baseline Scenario (BS)

Table 1

Main properties of the detected Ponticelli (NA) sub-basins - Baseline Scenario (BS)

Basin

Area

Average Slope

Impervious Rate

(m2)

(%)

(%)

1

201,408

2.43

76.62

2

37,449

1.35

79.05

3

69,854

1.68

80.34

4

24,524

1.75

81.85

5

58,588

1.33

88.98

6

20,433

1.27

66.81

7

81,558

0.99

70.32

8

49,127

1.45

84.27

9

36,452

0.80

83.67

10

95,087

1.21

88.39

11

55,279

0.86

84.08

12

32,407

1.03

94.95

13

118,332

1.36

90.15

14

46,453

1.12

80.02

15

56,745

1.14

86.14

16

84,971

1.33

61.86

17

82,716

1.90

84.67

Fig. 7

Detection of 17 sub-basins of the Ponticelli (NA) urban watershed

4 Data

With reference to the Fuorigrotta (NA) urban watershed, a simplified sewer model is implemented in EPA SWMM 5.1; indeed the solely main sewers, located downstream the North and South Basins, were introduced, joining in the J2 node, as depicted in Fig. 8. The simplified model allows the preliminary assessment of the implemented procedure reliability to provide effective solutions with affordable computational times.
Fig. 8

EPA SWMM 5.1 model of Fuorigrotta (NA) urban watershed

This configuration comprises the BS, corresponding to the rainwater system without LIDs application. 18 rainfall events are taken into account, provided by the Basin Authority of Campania Centrale, with reference to 6 durations, equal to 0.2, 0.5, 0.75, 1, 3 and 6 h, respectively. Moreover, the historical rainfall event occurred in the Fuorigrotta (NA) district on 15/09/2011, having rainfall intensity higher than 24 mm/h, is also simulated in order to assess the model capability to provide reliable solutions also when extreme events occur. It corresponds to a 500 years return period T, causing estimated damages of about 240 M€.

The tri-parametric equation derived from VAPI Campania Project (Rossi and Villani 1995), as a function of the TCEV (Two Component Extreme Value) distribution model is applied to calculate the rainfall intensity id,T related to duration d and return period T:
$$ {\mathrm{i}}_{\mathrm{d},\mathrm{T}}=\frac{{\mathrm{K}}_{\mathrm{T}}\cdot {\mathrm{I}}_0}{{\left(1+\frac{\mathrm{d}}{{\mathrm{d}}_{\mathrm{c}}}\right)}^{\upbeta}} $$
(3)
with I0 = 77.08 mm/h, dc = 0.3661 h, β = 0.7995–8.6077·10−5·Z where Z is the average height of the urban sub-catchment (m a.s.l.). With reference to return periods T of 2, 10 and 20 years, the growing factor KT is estimated of 0.92, 1.43 and 1.65, respectively.

The spilled volume V0 of the BS is considered to assess the benefits from the LIDs application, in terms of reduction of spilled volume through the performance indicator r = V/V0.

Porous pavements, bioretention and green roofs LIDs are selected among those available, combined with storage tanks, having maximum potential capacity of 1000 m2, located at the final nodes of the two sub-basins (Fig. 8). The latter are intended as a structural measure to be combined with the selected LIDs application whether these lasts are not able to prevent flooding for a specific sub-catchment. Setting LIDs parameters are summarized in the following Table 2, whereas in Table 3 the input properties to assess the allowable LIDs sizing of each sub-basin are defined.
Table 2

Setting parameters of LIDs for Fuorigrotta (NA) urban watershed

 

Porous Pavements

Bioretention

Green Roofs

Surface Berm Height (mm)

200

200

200

Surface Manning Roughness (−)

0.1

0.4

0.15

Surface Slope (%)

1.0

1.0

1.0

Soil Thickness (mm)

0

600

100

Soil Volume Fraction (−)

0.5

0.2

0.5

Suction Head (mm)

3.5

3.5

3.5

Storage Thickness (mm)

300

300

Seepage Rate (mm/h)

0.5

0.5

Table 3

Setting parameters of LIDs for sub-basins in Fuorigrotta (NA) urban watershed

 

North Basin

South Basin

Area (m2)

821,400

1,072,700

Width (m)

4086.6

6476.4

Impervious Rate (%)

92.90

81.40

Porous Pavement Max Area (m2)

383,629

419,553

Bioretention Max Area (m2)

66,184

199,955

Green Roof Max Area (m2)

11,505

141,852

For each LID, by GIS investigation the maximum potential area to be converted to LIDs is set and 10,000 iterations are run for simulations. A total allowable cost is fixed, as a function of the LID unitary costs from Zhang et al. (2013), converted into actualized Euros currency (Table 4). The unitary cost of the storage tanks is estimated by market surveys as a function of the surface A, instead.
Table 4

Unitary costs of applied LIDs (Zhang et al. 2013) and storage tanks

LID

Unitary Cost (€/m2)

Porous Pavements

92.70

Bioretention

231.82

Green Roofs

218.05

Storage Tanks

820.41*A - 856.46

Three levels of simulations are applied, at varying the volume rate V/V0 in the ranges (a) 100–77%; (b) 77–47% and (c) 47–34%. 30,000 solutions are performed, generating the Pareto Front, composed of the dominant solutions able, for a fixed cost, to maximize the volume reduction or, similarly, for a fixed volume rate V/V0, to minimize the construction costs.

With reference to the Ponticelli (NA) urban watershed, rainfall data from AdB of Campania Centrale are applied to derive the rainfall intensity for 0.2, 0.5, 1 and 3 h referring to 2, 10 and 20 years return periods, respectively. As for the first case-study, the historical event occurred on 15/09/2001 (Fig. 9) is also simulated to assess the reliability of the proposed solution to assure effective responses against extreme rainfall events. 12 theoretical hyetographs and a historical one were simulated, aimed at comparing the flooded volume, between the BS and the LIDs scenarios.
Fig. 9

Hyetograph of the historical event dated 15/09/2001

Porous pavements, bioretention and green roofs are selected, because of their close adaptability to the investigated area. As a function of the analysis about the land-use, the maximum extension of sub-basins area convertible with LIDs is estimated, as summarized in the following Table 5.
Table 5

Maximum area convertible with LIDs for 17 Ponticelli (NA) sub-basins

Sub-Basin

Porous Pavements Max Area

Bioretention Max Area

Green Roofs Max Area

(m2)

(m2)

(m2)

1

46,310.9

91,223.8

5793.7

2

2218.0

5095.5

3

14,345.6

33,243.6

783.3

4

1370.2

7544.0

5

13,250.5

32,134.8

5960.4

6

3027.5

5343.9

7

5158.7

17,587.6

8

9894.6

23,033.8

6192.6

9

5115.9

15,958.0

10

14,463.5

35,954.7

8742.7

11

12,360.8

20,702.2

5123.8

12

5121.7

15,140.3

13

19,589.4

58,773.9

14

4685.3

9089.3

15

5942.0

9191.5

16

7779.1

5243.4

17

7373.3

3676.7

TOT

178,007.0

388,937.0

32,596.5

20,000 iterations are run, by estimating for each solution the performance indicator r’ = V′/V’0 and the related required budget.

5 Results and Discussion

In the Fuorigrotta (NA) case-study, the optimal solution is achieved as a function of an available budget of 50–55 M€ with allowable volume rate V/V0 equal to about 70.1%, in agreement with De Paola et al. (2017a). Thus, by fixing the available budget, the optimal design of LIDs is achievable, so as to intend the implemented procedure as a DSS for LIDs design in urban stormwater systems.

In greater detail, in the North Basin, the porous pavements are designed with total area of 269,155.5 m2, bioretention with area of 65.21 m2, whereas green roofs are not included. In the South Basin, porous pavements have total area of 294,291.2 m2, resulting the lone applied LID application. Neither bioretention nor green roofs are in fact comprised, introducing a storage unit of 113.26 m2 to mitigate the runoff peak flows, instead. The required volume rate V/V0 is also in compliance with the geological properties of the studied area, mainly composed of pyroclasts, subjected to strong compaction if highly imbibed.

Comparison between the BS and LID scenarios is carried out. A volume rate V/V0 of 70.1% is quantified, pointing out the technical, environmental and economic benefits derivable from the use of the considered LIDs application (Fig. 10).
Fig. 10

Cost-volume solutions and Pareto Front for Fuorigrotta (NA) urban watershed

From Fig. 10 the potentiality of the implemented approach are observed because from the Pareto Front the conveyed volume reduction is assessed so as to estimate the achievable technical benefits as a function of the allowable budget.

With reference to the Ponticelli (NA) case-study, 20,000 iterations are run, by estimating for each solution the performance indicator r’ = V′/V’0 and the related required budget. The Pareto Front, plotted in Fig. 11, is thus derived.
Fig. 11

Cost-volume solutions and Pareto Front for Ponticelli (NA) urban watershed

The chosen solution provides the reduction of the flooded volume V′/V’0 of 95%, at the expense of a total budget of about 33 M€. Sizes of areas devoted to the LIDs application are summarized in the following Table 6. It corresponds to a budget repartition of 73.9%, 15.8%, 6.2 and 4.1% for storage tanks, bioretention, porous pavements and green roofs, respectively.
Table 6

LIDs design of Ponticelli (NA) urban watershed

Sub-Basin

Porous Pavements (m2)

Bioretention (m2)

Green Roofs (m2)

Storage Tanks (m2)

Storage Tanks Volume (m3)

1

3092.51

4560.92

1269.96

5076.85

6078

2

795.80

1164.17

173.07

249

3

2092.70

1288.68

693.55

124.35

186

4

711.17

266.07

1528.99

326

5

1494.97

1501.59

1199.96

2561.46

1254

6

705.68

345.63

1078.56

423

7

1847.75

644.39

5092.15

2052

8

945.05

1387.11

826.74

1514.48

733

9

1226.87

1109.76

75.83

36

10

1326.71

1286.52

1699.26

100.38

45

11

1305.63

1519.41

705.64

70.75

40

12

1209.10

1203.83

2527.17

1421

13

2538.58

2128.37

110.52

73

14

735.77

1143.36

3047.90

1467

15

953.88

587.11

1502.20

721

16

848.81

1221.26

1591.47

2139

17

495.39

1453.65

4016.02

1222

TOT

22,326.37

22,811.83

6395.11

30,192.15

18,465

The maximum area of the storage tanks is not set as a constraint because of the undersizing of the Ponticelli (NA) stormwater system. Hence the storage tanks are considered because capable to significantly reduce the runoff peak flows. The storage tanks have volume lower than 2200 m3 for the whole set of considered sub-basins, less than the sub-basin 1 equal to 6078 m3 (Table 6). Nevertheless, this size is allowable in the sub-basin 1 by integrating the storage tank in a water square, already available in the site. In compliance with De Paola et al. (2015), the insertion of storage tanks is preferred because cheaper than the resizing approach.

As a consequence of the proposed solution, the reduction of pressurized trunks is observed, also during heaviest rainfall events. Moreover, the reduction of the flooded volume deriving from the LIDs application is of about 66%. By combining the proposed LIDs with the storage tanks, a reduction of 98% is potentially achievable. As an example, in Table 7 results from the BS, the only LIDs application and the LIDs+Storage Tanks application are summarized with reference to duration d = 0.20 h and return period T = 2 years.
Table 7

Comparison between BS, LIDs application and LIDs + Storage Tanks of Ponticelli (NA) urban watershed

Scenario

Flooded Volume (·103 m3)

Flooded Volume Reduction (%)

Baseline Scenario (BS)

10,00

LIDs Application

3,36

66,38

LIDs + Storage Tanks

0,14

98,58

6 Conclusions

In this paper the optimization model proposed by De Paola et al. (2017a) to manage urban stormwater systems is analysed and applied to the two case-studies located in the Municipality of Naples, namely the Fuorigrotta and Ponticelli urban watersheds. The innovative approach to design LID practices in urban stormwater systems is based upon the interface between the Harmony Search meta-heuristic optimization model and the EPA SWMM 5.1 hydraulic solver. The tool can be intended as a DSS for the optimal management of water systems because allows, by fixing the allowable budget, the optimal sizing of LIDs able to minimize the flooded volumes. Conversely, knowing the required reduction of flooded volumes, it allows to estimate the minimum budget required to achieve the fixed goal. It is based upon the preliminary territorial analysis, performed by using GIS software, to evaluate the geological characteristic, the land use and the geodetic properties of the considered site. In the analysed case-studies porous pavements, bioretention and green roofs solutions are chosen among those manageable with EPA SWMM 5.1, combined with storage tanks, useful to lower the runoff peak flows. In both considered case-studies, by comparing the pre and post-intervention scenarios, the model capability to detect LID applications able to assure significant environmental and economic benefits is observed, returning effective responses also when extreme rainfall events occur.

Notes

Acknowledgements

A previous shorter version of the paper has been presented in the 10th World Congress of EWRA “Pantha Rhei”, Athens, Greece, 5-9 July 2017.

The authors would like to thank Eng. Valeria Guerriero and Eng. Giorgia Minale for their support in developing this work.

Compliance with Ethical Standards

Conflict of Interest

None.

References

  1. AdB Autorità di Bacino Regionale della Campania Centrale (2013) Piano Gestione Rischio Alluvioni. http://www.adbcampaniacentrale2.it/documenti/psgra/Relazione_F_18.pdf (in Italian). Accessed 20 July 2018
  2. Ahmed K, Chung E-S, Song J-Y, Shahid S (2017) Effective design and planning specification of low impact development practices using water management analysis module (WMAM): case of Malaysia. Water 9:173.  https://doi.org/10.3390/w9030173 CrossRefGoogle Scholar
  3. Al-Betar MA, Khader AT (2012) A harmony search algorithm for university course timetabling. Ann Oper Res 194(1):3–31.  https://doi.org/10.1007/s10479-010-0769-z CrossRefGoogle Scholar
  4. Alfieri L, Thielen J (2015) A European precipitation index for extreme rain-storm and flash flood early warning. Meteorol App 22(1):3–13.  https://doi.org/10.1002/met.1328 CrossRefGoogle Scholar
  5. Arnbjerg-Nielsen K, Willems P, Olsson J, Beecham S, Pathirana A, Bülow Gregersen I, Madsen H, Nguyen VTV (2013) Impacts of climate change on rainfall extremes and urban drainage systems: a review. Water Sci Technol 68(1):16–28.  https://doi.org/10.2166/wst.2013.251 CrossRefGoogle Scholar
  6. Cheng MS, Zhen JX, Shoemaker L (2009) BMP decision support system for evaluating stormwater management alternatives. Front Environ Sci En 3(4):453–463.  https://doi.org/10.1007/s11783-009-0153-x CrossRefGoogle Scholar
  7. Coelho LS, Mariani VC (2009) An improved harmony search algorithm for power economic load dispatch. Energ Convers Manage 50(10):2522–2526.  https://doi.org/10.1016/j.enconman.2009.05.034 CrossRefGoogle Scholar
  8. Comune di Napoli Assessorato alla Vivibilità (2005) Variante al Piano Regolatore Generale. Tavola 5: Zonizzazione. http://www.comune.napoli.it/flex/cm/pages/ServeBLOB.php/L/IT/IDPagina/2166. Accessed 20 July 2018
  9. Davis AP (2005) Green engineering principles promote low-impact development. Environ Sci Technol 39(16):338A–344A.  https://doi.org/10.1021/es053327e CrossRefGoogle Scholar
  10. De Paola F, Galdiero E, Giugni M, Pugliese F (2015) Sustainable development of storm-water systems in African cities considering climate change. Procedia Eng. 119:1181–1191.  https://doi.org/10.1016/j.proeng.2015.08.970 CrossRefGoogle Scholar
  11. De Paola F, Fontana N, Giugni M, Marini G, Pugliese F (2016a) An application of the harmony-search multi-objective (HSMO) optimization algorithm for the solution of pump scheduling problem. Procedia Eng 162:494–502.  https://doi.org/10.1016/j.proeng.2016.11.093 CrossRefGoogle Scholar
  12. De Paola F, Giugni M, Pugliese F (2016b) A harmony-based calibration tool for urban drainage systems. Proc Inst Civ Eng Water Manage 171(1):30–41.  https://doi.org/10.1680/jwama.16.00057 CrossRefGoogle Scholar
  13. De Paola F, Giugni M, Pugliese F, Romano P (2017a) A decision support system for urban stormwater drainage management. Europ Water 57:115–121Google Scholar
  14. De Paola F, Fontana N, Giugni M, Marini G, Pugliese F (2017b) Optimal solving of the pump scheduling problem by using a harmony-search optimization algorithm. J Hydroinf 19(6):879–889.  https://doi.org/10.2166/hydro.2017.132 CrossRefGoogle Scholar
  15. De Paola F, Galdiero E, Giugni M (2017c) Location and setting of valves in water distribution networks using a harmony search approach. J Water Resour Plan Manag 143(6):04017015.  https://doi.org/10.1061/(ASCE)WR.1943-5452.0000760 CrossRefGoogle Scholar
  16. De Martino G, De Paola F, Fontana N, Marini G, Ranucci A (2012) Experimental assessment of level pool routing in preliminary design of floodplain storage. Sci Total Environ 416:142–147CrossRefGoogle Scholar
  17. DESA Department of Economic and Social Affairs of the United Nations (2015) World Urbanization Prospects: The 2014 Revision. (ST/ESA/SER.A/366). ISBN 978-92-1-151517-6Google Scholar
  18. Dietz ME (2007) Low impact development practices: a review of current research and recommendations for future directions. Water Air Soil Pollut 186(1–4):351–363.  https://doi.org/10.1007/s11270-007-9484-z CrossRefGoogle Scholar
  19. Du PF, Jia H, Yu SL (2012) Urban watershed management under rapid urbanization. Front Environ Sci En 6(5):595.  https://doi.org/10.1007/s11783-012-0452-5 CrossRefGoogle Scholar
  20. Elliott AH, Trowsdale SA (2007) A review of models for low impact urban stormwater drainage. Environ Model Softw 22:394–405.  https://doi.org/10.1016/j.envsoft.2005.12.005 CrossRefGoogle Scholar
  21. Field CB, Barros VR, Dokken DJ, Mach KJ, Mastrandrea MD, Bilir TE, Chatterjee M, Ebi KL, Estrada YO, Genova RC, Girma B, Kissel ES, Levy AN, MacCracken S (2014) IPCC: IPCC, 2014: Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, USA, pp. 1132. https://www.ipcc.ch/pdf/assessment-report/ar5/wg2/ar5_wgII_spm_en.pdf. Accessed 20 July 2018
  22. Fletcher TD, Shuster W, Hunt WF, Ashley R, Butler D, Arthur S, Trowsdale S, Barraud S, Semadeni-Davies A, Bertrand-Krajewski JL, Mikkelsen PS, Rivard G, Uhl M, Dagenais D, Viklander M (2015) SUDS, LID, BMPs, WSUD and more – the evolution and application of terminology surrounding urban drainage. Urban Water J 12(3):525–542.  https://doi.org/10.1080/1573062X.2014.916314 CrossRefGoogle Scholar
  23. Geem ZW, Ki JH, Loganathan GV (2001) A new heuristic optimization algorithm: harmony search. Simul 76:60–68.  https://doi.org/10.1177/003754970107600201 CrossRefGoogle Scholar
  24. Geem ZW, Tseng CL, Park Y (2005) Harmony search for generalized orienteering problem: best touring in China. Lect Notes Comput Sci 3612(part III):741–750.  https://doi.org/10.1007/11539902_91 CrossRefGoogle Scholar
  25. Gulbaz S, Kazezyilmaz-Alhan CM (2015) Investigating the effects of low impact development (LID) on surface runoff ad tss in a calibrated hydrodynamic model. J Urban Environ Eng 9(2):91–96.  https://doi.org/10.4090/juee.2015.v9n2.091096 CrossRefGoogle Scholar
  26. Hager GW, Belt KT, Stack W, Burgess K, Grove JM, Caplan B, Hardcastle M, Shelley D, Pickett ST, Groffman PM (2013) Socioecological revitalization of an urban watershed. Front Ecol Environ 11(1):28–36.  https://doi.org/10.1890/120069 CrossRefGoogle Scholar
  27. Hapurachchi HAP, Wang QJ, Pagano TC (2011) A review of advances in flash flood forecasting. Hydrol Process 25:2771–2784.  https://doi.org/10.1002/hyp.8040 CrossRefGoogle Scholar
  28. Jia H, Lu YW, Yu SL, Chen YR (2012) Planning of LID-BMPs for urban runoff control: the case of Beijing Olympic Village. Sep Purif Technol 84:112–119.  https://doi.org/10.1016/j.seppur.2011.04.026 CrossRefGoogle Scholar
  29. Jia H, Yao H, Yu SL (2013) Advances in LID-BMPs research and practice for urban runoff control in China. Front Env Sci Eng 7(5):709–720.  https://doi.org/10.1007/s11783-013-0557-5 CrossRefGoogle Scholar
  30. Jia H, Yao H, Tang Y, Yu SL, Field R, Tafuri AN (2015) LID-BMPs planning for urban runoff control and the case study in China. J Environ Manag 149:65–76.  https://doi.org/10.1016/j.jenvman.2014.10.003 CrossRefGoogle Scholar
  31. Khan S, Lau SL, Kayhanian M, Stenstrom MK (2006) Oil and grease measurement in highway runoff-sampling time and event mean concentrations. J Environ Eng-ASCE 132(3):415–422.  https://doi.org/10.1061/(ASCE)0733-9372(2006)132:3(415) CrossRefGoogle Scholar
  32. Kim J-H, Kim HW, Demarie F (2017) Facilitators and barriers of applying low impact development practices in urban development. Water Res Manage 31(20):1–14.  https://doi.org/10.1007/s11269-017-1707-5 CrossRefGoogle Scholar
  33. Lazrus H, Morss RE, Demuth JL, Lazo JK, Bostrom A (2016) Know what to do if you encounter a flash flood: mental models analysis for improving flash flood risk communication and public decision making. Risk Anal 36(2):411–427.  https://doi.org/10.1111/risa.12480 CrossRefGoogle Scholar
  34. Li H, Harvey JT, Holland TJ, Kayhanian M (2013) The use of reflective and permeable pavement strategies for heat island mitigation and stormwater management. Environ Res Lett 8:015023.  https://doi.org/10.1088/1748-9326/8/1/015023 CrossRefGoogle Scholar
  35. Liu Y, Cibin R, Bralts VF, Chaubey I, Bowling LC, Engel BA (2016) Optimal selection and placement of BMPs and LID practices with a rainfall-runoff model. Environ Model Softw 80:281–296.  https://doi.org/10.1016/j.envsoft.2016.03.005 CrossRefGoogle Scholar
  36. Rossi F, Villani P (1995) Valutazione delle Piene in Campania. In: Grafica Metelliana & C (ed), pubbl. 1472 (in Italian)Google Scholar
  37. Rossman LA (2017) Storm water management model reference manual volume II – hydraulics. U.S. Environmental Protection Agency, Washington, DC EPA/600/R-17/111, 2017. https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P100S9AS.pdf. Accessed on 20 July 2018Google Scholar
  38. Špitalar M, Gourley JJ, Lutoff C, Kirstetter PE, Brilly M, Carr N (2014) Analysis of flash flood parameters and human impacts in the US from 2006–2012. J Hydrol 519:863–870.  https://doi.org/10.1016/j.jhydrol.2014.07.004 CrossRefGoogle Scholar
  39. Terti G, Ruin I, Anquetin S, Gourley JJ (2015) Dynamic vulnerability factors for impact-based flash flood prediction. Nat Hazards 79:1481–1497.  https://doi.org/10.1007/s11069-015-1910-8 CrossRefGoogle Scholar
  40. Turk S, Radeke R (2011) Optimization of energy efficient network migration using harmony search. Lect Notes Comput Sci 6955(LNCS):89–99.  https://doi.org/10.1007/978-3-642-23541-2_11 CrossRefGoogle Scholar
  41. UN United Nations (2017) Department of Economic and Social Affairs. Population Division (2017). World Urbanization Prospects: The 2017 Revision, Key Findings and Advance Tables. Working Paper No. ESA/P/WP/248. https://esa.un.org/unpd/wpp/Publications/Files/WPP2017_KeyFindings.pdf. Accessed 20 July 2018
  42. US EPA (2004) The Use of Best Management Practices (BMPs) in Urban Watersheds. EPA/600/R-04/184, Washington DCGoogle Scholar
  43. Yilmaz AG, Hossain I, Perera BJC (2014) Effect of climate change and variability on extreme rainfall intensity-frequency-duration relationships: a case study of Melbourne. Hydrol Earth Syst Sci 18:4065–4076.  https://doi.org/10.5194/hess-18-4065-2014 CrossRefGoogle Scholar
  44. Zhang G, Hmlett JM, Reed P, Tang Y (2013) Multi-objective optimization of low impact development designs in an urbanizing watershed. Open J Optim 2:95–108.  https://doi.org/10.4236/ojop.2013.24013 CrossRefGoogle Scholar

Copyright information

© Springer Nature B.V. 2018

Authors and Affiliations

  1. 1.Department of Civil, Architectural and Environmental EngineeringUniversity of Naples Federico IINaplesItaly

Personalised recommendations