See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/318465795

Enhancing future resilience in urban drainage

system: Green versus grey infrastructure

Article in Water Research · July 2017

DOI: 10.1016/j.watres.2017.07.038

CITATIONS READS

0 60

3 authors, including:

Xin Dong
Tsinghua University
26 PUBLICATIONS 65 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

measuring and explaining the eco-efficiencies of WWTPs in China: an uncertainty perspective View
project

assessing the environmental impacts of wastewater treatment sectors in China: a life-cycle and
economy-wide perspective View project

All content following this page was uploaded by Xin Dong on 24 July 2017.

The user has requested enhancement of the downloaded file.

Accepted Manuscript

Enhancing future resilience in urban drainage system: Green versus grey

infrastructure

Xin Dong, Hao Guo, Siyu Zeng

PII: S0043-1354(17)30611-5
DOI: 10.1016/j.watres.2017.07.038
Reference: WR 13081

To appear in: Water Research

Received Date: 24 February 2017

Revised Date: 19 June 2017
Accepted Date: 16 July 2017

Please cite this article as: Dong, X., Guo, H., Zeng, S., Enhancing future resilience in urban drainage
system: Green versus grey infrastructure, Water Research (2017), doi: 10.1016/j.watres.2017.07.038.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

1 Enhancing future resilience in urban drainage system: green versus grey

2 infrastructure

PT
3 Xin Dong 1,2, Hao Guo1, Siyu Zeng 1,2*

RI
4 1. School of Environment, Tsinghua University, Beijing, 100084, China

5 2. Environmental Simulation and Pollution Control State Key Joint Laboratory, School of

SC
6 Environment, Tsinghua University, Beijing 100084, China

U
7 Corresponding author: szeng@tsinghua.edu.cn
AN
8 Abstract:
M

9 In recent years, the concept transition from fail-safe to safe-to-fail makes the application of

10 resilience analysis popular in urban drainage systems (UDSs) with various implications and
D

11 quantifications. However, most existing definitions of UDSs resilience are confined to the
TE

12 severity of flooding, while uncertainties from climate change and urbanization are not
EP

13 considered. In this research, we take into account the functional variety, topological complexity,

14 and disturbance randomness of UDSs and define a new formula of resilience based on three parts
C

15 of system severity, i.e. social severity affected by urban flooding, environmental severity caused
AC

16 by sewer overflow, and technological severity considering the safe operation of downstream

17 facilities. A case study in Kunming, China is designed to compare the effect of green and grey

18 infrastructure strategies on the enhancement of system resilience together with their costs.

19 Different system configurations with green roofs, permeable pavement and storage tanks are

20 compared by scenario analysis with full consideration of future uncertainties induced by

1
 ACCEPTED MANUSCRIPT

21 urbanization and climate change. The research contributes to the development of sustainability

22 assessment of urban drainage system with consideration of the resilience of green and grey

23 infrastructure under future change. Finding the response measures with high adaptation across a

24 variety of future scenarios is crucial to establish sustainable urban drainage system in a long

PT
25 term.

RI
26 Key words:

SC
27 Urban drainage system, resilience, grey infrastructure, green infrastructure, climate change

28 1. Introduction

U
AN
29 A sustainable urban infrastructure system ought to be functional now and adaptive in uncertain

30 future. It means the infrastructures not only need to be reliable subjected to the standard loading,
M

31 but also are able to minimize the magnitude and duration of failures under exceptional conditions.

32 Resilience is a concept that can describe the aforementioned requirement on the urban
D

33 infrastructure perfectly and has become an important goal for the system. Furthermore, resilience
TE

34 in infrastructure system is regarded as engineering resilience originating from ecological concept

35 which represents the persistence of natural systems in the face of changes in ecosystem variables
EP

36 due to natural or anthropogenic causes (Peterson et al., 1998; Walker et al., 2004). Generally,
C

37 engineering resilience gauges the displacement magnitude of system’s behavior away from
AC

38 designed function and the recovery speed of system’s behavior under the shocks or disturbances

39 (Francis and Bekera, 2014; Henry and Ramirez-Marquez, 2012). The idea of engineering

40 resilience has been used in many infrastructure systems to characterize the ability to reduce the

41 magnitude and/or duration of negative impacts of the disruptive events, such as power

42 transmission grid, road network, and internet etc. (Francis and Bekera, 2014; Henry and

2
 ACCEPTED MANUSCRIPT

43 Ramirez-Marquez, 2012; Cohen et al., 2000; Shafieezadeh and Burden, 2014; Labaka et al.,

44 2015), which has transferred the principle of the planning and building of urban infrastructures

45 from fail-safe to safe-to-fail (Ahern, 2011; Butler et al., 2014). In particular, there is a growing

46 emphasis on enhancing the resilience of urban infrastructures in the face of unprecedented

PT
47 urbanization and climate change (McDaniels et al., 2008; Ouyang et al., 2012; Hossain et al.,

RI
48 2015).

49 Urban drainage systems (UDSs) are critical and complex infrastructures in cites.

SC
50 Notwithstanding all efforts made in improving the sustainability of UDSs on the basis of

51 integrated planning and management, UDSs are subject to the disturbances resulting from both

52
U
system exterior and interior such as climate change, urbanization, sewer collapse, and etc.
AN
53 (Semadeni-Davies et al., 2008; Yazdanfar and Sharma, 2015; Zhou, 2014) It has been widely

54 acknowledged that climate change and urbanization could trigger more unexpected system
M

55 failures and introduce more external uncertainties implied on the system (Zhou, 2014; Vojinovic
D

56 et al., 2014; Notaro et al, 2015). Changes in climatic conditions, such as the increase of rainfall
TE

57 intensities, shifts of precipitation patterns, and more extreme weather events, result in that UDSs

58 expose to heavy rainfall strikes in high frequency (Willems et al., 2012; Grum et al., 2006;
EP

59 Karamouz et al., 2011), which could lead to a significant augmentation in the frequency and

60 magnitude of urban flooding. Intensified aggregation of the population and economic activities
C

61 due to urbanization also enlarge the burden on the existing UDSs in terms of the quantity of
AC

62 wastewater and urban runoff, the amount of nutrients, and the variety of pollutants as well (Hatt

63 et al., 2004; Huong and Pathirana, 2013), which would exert substantial influence on the quality

64 of urban water environment. Facing the destined but unpredictable uncertainties, design and

65 retrofitting of UDSs considering the potential impacts of climate change and urbanization in

3
 ACCEPTED MANUSCRIPT

66 terms of enhancing the system resilience would become ever more important to keep our city

67 habitable into the future.

68 In recent years, the application of green infrastructure is deemed as an effective and flexible

69 strategy for the mitigation of and adaption to the disturbances generated from climate change and

PT
70 urbanization and thus the promotion of the sustainability of UDSs (Liu et al., 2014; Liu et al.,

RI
71 2015; Pugh et al., 2012). Green infrastructures are vegetated or sustainability-based practices,

72 such as green roofs, porous pavements, bio-retention cells, and swales, which can reduce the

SC
73 amount of stormwater entering UDSs (Tavakol-Davani et al., 2015). On the contrary, gray

74 infrastructures are composed predominantly of concrete and steel (thus the gray part of the term)

75
U
and installed as part of the UDS (Tavakol-Davani et al., 2015). Plenty of studies have evaluated
AN
76 the roles of various types of green infrastructures on storm water management as well as carbon

77 emission control (Liu et al., 2014; Liu et al., 2015; Pugh et al., 2012) and compared the
M

78 performances with those of grey infrastructures. Cohen et al. (2011) carried out a life-cycle cost
D

79 analysis comparing a green (rain gardens) and gray (tunnels) infrastructure combination to a
TE

80 gray-only option to control combined sewer overflow. Their results suggested that the green/gray

81 combined alternative turned out to be more cost-effective than the gray-only option. Similar
EP

82 findings were represented the study by Wang et al. (2013). Casal-Campos et al. (2015) examined

83 the robustness of green and gray drainage strategies in the future using a regret-based approach
C

84 and suggest that the combination of green and gray strategies may still offer further potential for
AC

85 robustness. However, the relationship between the resilience and UDS configuration in which

86 green and grey infrastructures are implemented is still obscure. And thus the subsequent

87 optimization of the UDS configuration considering the resilience cost is still insufficiently

88 discussed.

4
 ACCEPTED MANUSCRIPT

89 Over the past two decades, although more attentions have been paid to the resilience in UDSs

90 (Mugume et al., 2014; Mugume et al., 2015; Golz et al., 2015; Mugume et al., 2015), the

91 quantitative definition of this conenpt is still open to debate (Meerow et al., 2016). When climate

92 change and urbanization is concerned, most studies are limited to the flooding resilience when

PT
93 characterizing the UDSs (Mugume et al., 2014; Mugume et al., 2015; Mugume et al., 2015).

RI
94 Moreover, two groups of methods are reported for the assessment. A comprehensive indicator

95 system established on the basis of a set of measurable surrogate attributes is frequently used. De

SC
96 Bruijn uses reaction threshold, amplitude, graduality, and recovery rate to give an overview of

97 the system’s reaction to flooding and determine the flooding resilience (De Bruijn, 2004).

98
U
Mugume et al. emphasize the flexibility and redundancy, and use flood volume and nodal flood
AN
99 duration to examine the impact of pipe failure on two surrogate attributes of resilience (Mugume

100 et al., 2015). Instead of this multi-attribute-based approach, Mugume et al. value the flooding
M

101 resilience using one integrated indicator with the consideration of both magnitude and duration
D

102 of flooding, based on the modeling of the change process of UDSs’ hydraulic characteristics and
TE

103 utility performance under the flooding condition (Mugume et al., 2014; Mugume et al., 2015).

104 Although the previously reported methods are verified to be feasible for the resilience
EP

105 assessment focusing on the urban flooding control, none could support the further identification

106 of the critical points for the enhancement of system resilience, which is quite meaningful for the
C

107 decision makers.

AC

108 Apparently, the roles required by the sustainable UDS are comprehensive and include

109 avoiding degradation of urban water environment, ensuring the secure operation of wastewater

110 treatment plants (WWTPs) in the combined sewer system, and so on. For the sake of a

111 generalized assessment of the UDS, its function and related resilience should not be confined to

5
 ACCEPTED MANUSCRIPT

112 guard against urban flooding only. Comprehensive resistant capacity of UDSs should be

113 considered in the resilience quantification, especially when the comparison between green and

114 grey infrastructure is conducted. Moreover, the performance of the whole UDSs is a nonlinear

115 superposition of contributions of all system components including nodes, conduits, pumps, and

PT
116 etc. and highly dependent on the complicated topology structure of the system. Due to the

RI
117 considerable spatial variability of resilience in the UDSs, a quantification method that could

118 facilitate the identification of sensitive components and prioritize the critical points in the whole

SC
119 system is needed.

120 Aiming at deeper understanding of the resilience response of the UDSs to future external

121
U
interference in terms of long-term circumstance change, this paper focuses on the comparison of
AN
122 the roles of green and grey infrastructures in enhancing the system resilience, on the premise of

123 the development of a new characterizing method and the application of a dynamic model for
M

124 system behaviour simulation. As for the disturbance to the system with different configuration of
D

125 green and/or grey infrastructures, 25 future scenarios considering uncertainty brought by climate
TE

126 change and urbanization are designed and evaluated. The relationship between the system

127 resilience and infrastructure cost is also analysed.

EP

128 2. Methods
C

129 2.1 Computation of UDSs resilience index

AC

130 In this work, taking the system failure both on magnitude and duration into account, the UDSs

131 resilience is estimated with the area between the normal system performance curve and the

132 observed system performance curve at any time after occurrence of one disturbance, which has

133 been justified by several studies (McDaniels et al., 2008; Ouyang et al., 2012; Mugume et al.,

6
 ACCEPTED MANUSCRIPT

134 2015). Different from the previous researches, with consideration of the system functionality i.e.

135 maintaining the security of water ecology and human society, the UDSs performance is

136 examined from three aspects besides the urban flooding control. In this proposed formulation of

137 UDSs resilience index (Res, shown in Eq 1), the loss of UDSs functionality caused by the urban

PT
138 flooding, combined sewer overflow (CSO), and shock loading to the WWTPs downstream are

RI
139 combined, which represents the social severity (Sev-s), environmental severity (Sev-e), and

140 technological severity (Sev-t), respectively. The advantage is obvious in that involving manifold

SC
141 functions of the system in the index leads to a more elaborated and structural quantification of

142 engineering resilience of UDS and facilitates the understanding of the linkage among

143
U
engineering resilience, social resilience, and ecological resilience as well. Furthermore, to
AN
144 average the impacts of rainfall events with different intensities so that the comparison between

145 the green and grey infrastructure is conducted on a sound and systematic base, the mathematical
M

146 expectation of system performance under different precipitation conditions is used to represent
D

147 UDSs resilience (Eq 1).

TE

1 1
Res = ∑ ⋅
T k s Sevs k e Seve kt Sevt
T
1+ + +
Qtw Qtw Qtw
1 1
EP

148 =∑ ⋅ (Eq
Nf
 t fi  No
 toi  Np
 t pi 
( Q fi − Ac fi ) dt  ( Q pi − Cawwtpi ) dt 
T
∑ ∑ ( Qoi − Aloi ) dt  ∑
tn tn tn
∫ ∫ ∫
T
  
i =1  t n − t0  +k i =1  t n − t0  +k i =1  t n − t 0 
t0 t0 t0
1 + ks Nw e Nw t Nw

∑ ∫ Qwi dt ∑∫ ∑∫
tn tn tn
Qwi dt Qwi dt
C

t0 t0 t0
i =1 i =1 i =1
AC

149 1)

150 Where T is the different return periods of rainfalls which are concerned in the Res calculation; ks,

151 ke and kt are the weights of Sev-s, Sev-e and Sev-t respectively which depend on the decision

152 preference of local stakeholders with respect to the impacts elimination of urban flooding, CSO

153 and shock loading to the WWTPs downstream; t0 is the moment when the rainfall begins; tn is

7
 ACCEPTED MANUSCRIPT

154 the moment when the system returns to the state before the external disturbance happen; Qtw is

155 the total volume of wastewater generated between t0 and tn; Nw, Nf, No, and Np are the number of

156 wastewater inflow nodes, flooding nodes, overflow outfalls, and WWTPs respectively; Qwi is the

157 wastewater flow at the ith inflow node; Qfi is the flooding flow at the ith flooding node; Acfi is the

PT
158 maximum acceptable flooding flow at the ith flooding node; tfi is the duration when Qfi is larger

RI
159 than Acfi at the ith flooding node; Qoi is the CSO flow at the ith overflow outfall; Aloi is the

160 maximum allowable CSO flow at the ith overflow outfall; toi is the duration when Qoi is larger

SC
161 than Aloi at the ith overflow outfall; Qpi is the ith WWTP influent flow, Cawwtpi is the maximum

162 operational capability of the ith WWTP with qualified effluent; tpi is the duration when Qpi

163 exceeds Cawwtpi at the ith WWTP.

U
AN
164 The proposed index Res quantifies the resilience as a function of the failure magnitude (i.e.

165 Qfi-Acfi, Qoi -Aloi, and Qpi-Cawwtpi) and failure duration (i.e. tfi, toi, and tpi), which is contributed by
M

166 all system components in a nonlinear, dynamic, and spatially heterogeneous way. The bigger the
D

167 Res, the higher the system resilience, which indicates the system is more capable of handling the
TE

168 considered external threat. For the calculation of Res, a physically based model of the whole

169 UDSs is required to provide the detailed description of all relevant processes occurring in UDSs
EP

170 and their interactions.

171 With the new definition, it can be determined that the upper limit value of the system
C

172 resilience is ∑T 1/T, which is achieved when no CSO, flooding, or WWTPs shock inflow occurs
AC

173 under all rainfall conditions considered in the resilience calculation. Apparently, this upper limit

174 value is only a specified threshold instead of a theoretical extreme value.

175 Of course, the social, environmental and technological severity of UDSs involve many aspect,

176 such as urban flooding can cause economic losses and inconvenience to transportation, and it

8
 ACCEPTED MANUSCRIPT

177 will also threaten the public life etc. But in this work, the system boundary is limited to the

178 drainage system itself, including drainage pipe network and the related green and gray

179 infrastructures. We only use the most intuitive performance of the system i.e. the magnitude and

180 duration of urban flooding, CSO, and shocks to the WWTPs to represent the terminal hazards

PT
181 such as the property damage and casualties triggered by urban flooding, the deterioration of

RI
182 water quality caused by CSO and the abnormal operation of WWTPs initiated by inflow shocks.

183 2.2 UDSs modeling

SC
184 The Storm Water Management Model (SWMM) developed by United States Environmental

U
185 Protection Agency (Rossman, 2009) is used to dynamically simulate UDSs in this work, which
AN
186 can be used to simulate UDSs for single events and in the long term, in dry and wet weather.

187 SWMM describes the precipitation received in the sub-catchment areas, the generation of runoff,
M

188 and the transportation of runoff through the whole sewer system including pipes, channels,

189 storage tanks, pumps, devices, and regulators. In this paper, the variables required for Res metric
D

190 calculation would be valued by SWMM, such as the number of overflow outfalls i.e. No, the
TE

191 flooding flow at the ith flooding node i.e. Qfi and so on.

192 Hornberger–Spear–Young (HSY) algorithm, based on Monte Carlo simulation, is applied to

EP

193 identify and verify the sub-models’ parameters for the simulation accuracy. In the HSY
C

194 algorithm, some prescribed conditions such as the limit value of difference between the
AC

195 simulation and observation are defined. If the prediction of the system’s previous behavior can

196 satisfy the conditions, the parameters used for the prediction are taken as ‘acceptable’. For the

197 system future behavior prediction, only the acceptable parameters can be used (Chen and Beck,

198 1997). In this work, 6 crucial parameters in the SWMM model are calibrated by the HSY

199 algorithm, including manning coefficient for impervious area, manning coefficient for pervious

9
 ACCEPTED MANUSCRIPT

200 area, conduit manning coefficient, maximum rate on the Horton infiltration curve, minimum rate

201 on the Horton infiltration curve and decay constant for the Horton infiltration curve. The

202 simulation errors of peak flow, occurrence of peak flow and average flow in monitoring nodes

203 are used to define the acceptable conditions of parameters.

PT
204 2.3 Case study overview

RI
205 The case study area in this investigation is a newly developing urban area in the north of

206 Kunming, a city in the southwest of China. It occupies an area of 10.4 km2 with 2.7×105

SC
207 inhabitants in 2015. This area has a subtropical monsoon climate and the rainy season lasting

U
208 from May to October with an annual average rainfall of 1450 mm. Four subsystems in this case
AN
209 area are included in the study: a representative urban watershed, a hybrid of separated and

210 combined sewer system, a WWTP, and an urban river that flows into the Dianchi Lake (a famous
M

211 eutrophic large-scale plateau lake in China). During the wet weather, CSOs are discharged

212 directly into the urban river with no treatment. This has motivated the local government to put
D

213 forward a drainage system upgrade plan.

TE

214 Three different UDS renovation strategies are compared within this area: two green

215 infrastructure strategies with application of green roofs and permeable pavements and one grey
EP

216 infrastructure strategy with addition of CSOs storage tank. The three strategies are considered to
C

217 be implemented in isolation, which make the pros and cons between green and grey
AC

218 infrastructure be analysed easily and thoroughly. Furthermore, a baseline scenario with nothing-

219 done strategy is used to evaluate the impacts of future uncertainties and different strategies. Five

220 rainfall events with 1, 3, 5, 10, and 30 years return periods are selected for Res calculation (for

221 more information on selected rainfall events refer to SI Fig. S1). Considering the local

222 preference, the k factors in the Res formulation are selected as: ks=1.5, ke=2, and kt=1.

10
 ACCEPTED MANUSCRIPT

223 2.4 Future Scenarios

224 In this work, the change of urban watershed imperviousness and the variation of rainfall

225 intensity are used to quantitatively represent the urbanization and climate change respectively.

226 The future scenarios are defined by the alternations of parameters in the baseline scenario. As a

PT
227 consequence of urbanization, the impervious area would increase. On the basis of the trend

RI
228 analysis for local social-economic development, four urbanization scenarios are designed, in

229 which the impervious area is increased by 5%, 10%, 15%, and 20% (the status quo is 49%).

SC
230 According to the predicted ranges of the precipitation increase due to global climate change

231 (Zhou, 2014; Arnbjerg-Nielsen, 2012; Yin et al., 2016), the impact of climate change is

232
U
simulated by raising the rainfall time series intensity by 5%, 10%, 15%, and 20%. Considering
AN
233 the joint pressure from both urbanization and climate change, 25 combined scenarios are

234 designed to describe the future changes. The parameter values in all scenarios are listed in SI
M

235 Table S1 (Scenario design for future to the case system in terms of climate change and
D

236 urbanization).
TE

237 Under each future scenario, different scales of green and grey infrastructures are set and

238 simulated. For the green infrastructure strategies, different percentages of the total area for green
EP

239 roof or permeable pavement are arranged. Green roof and permeable pavement are distributed

240 with the same percentage area in each of sub-catchments. For the grey infrastructure strategy,
C

241 different numbers of storage tanks are constructed, i.e. different storage volumes applied. For the
AC

242 combined controls, different combinations of green infrastructures and storage tanks are

243 displayed. Details on the layout of green roofs, permeable pavements, and storage tanks are

244 available in the SI Fig. S2, and the equations for cost estimation as well.

11
 ACCEPTED MANUSCRIPT

245 3. Results and discussion

246 3.1 Impacts of urbanization and climate change on resilience of case system

247 The resilience of the case system is calculated under each future scenario (shown in Fig. 1)

PT
248 and the baseline value (status quo with current climate and land use) of the system resilience is

249 0.67. The different degrees of urbanization and climate change would cause different system

RI
250 responses and subsequent resilience changes. Both intensified rainfall accelerated by climate

SC
251 change and growing impervious surface due to urbanization would reduce the system resilience.

252 Only considering the climate change, 20% increase of rainfall intensity (i.e. scenario C20) would

U
253 lead to 16% reduction of system resilience. Similar to climate change, only considering the
AN
254 urbanization, 20% increase of impervious area (i.e. scenario U20) would result in 13% reduction

255 of system resilience. Comparing these two future uncertain disturbances to the system, the effect
M

256 of climate change seems to be more serious, because the climate change gives rise to larger

257 social severity and more rapid growth of environmental severity for UDS (for more information
D

258 to SI Fig. S3 & S4). However, it should be noted that this effect depends on the existing
TE

259 watershed conditions and climate. For the collaborative consequences of urbanization and
EP

260 climate change, the reduction of system resilience would be aggravated, such as, 20% increase of

261 climate change plus 20% improvement of urbanization would cut down the system resilience by
C

262 24%, which depicts the nonlinear response of the relations among climate change, urbanization,
AC

263 and UDS performance in case area.

264

265 Fig. 1. UDS resilience under different urbanization and climate change scenarios

266 Fig. 2 illustrates the dynamic change process of the social, environmental, and technological

267 severity during rainfall under baseline and future scenarios, respectively. For this case drainage

12
 ACCEPTED MANUSCRIPT

268 system, at the first stage of rainfall (about 120-200 min after rainfall begins), SevS is the largest

269 severity for UDS and gains faster, which indicates that the urban flooding is the main

270 contribution for the recession of system resilience. In the middle stage (about 300-350 min after

271 rainfall begins), SevS and SevT tend to be stable in turn. SevE sustains the growth, but the speed of

PT
272 increase begins to slow down. In this stage, CSO is the major cause that leads to the decline of

RI
273 the system resilience. Finally, these three severities are stabilized, and SevE is distinctly larger

274 than the other two. It means that in this case, the environmental severity is the crux for the

SC
275 change of UDS’s resilience, so that the cutting of the CSOs is the most effective measure to hold

276 or increase the system resilience. According to the comparison among SevE, SevS, and SevT, it can

277
U
be found that the environmental benefit of the case UDS is the most sensitive to the future threats
AN
278 from the climate change and urbanization.

279
M

280 Fig. 2. Social, environmental, and technological severity of the system under future scenarios
D
TE

281 3.2 Identification of vulnerable spots based on resilience evaluation

282 According to the definition in Eq 1, the resilience of UDS is the nonlinear superposition in
EP

283 time and space of performances of all nodes composing the whole system. In this paper, an

284 indicator named as node apparent resilience contribution (NARC) is used to describe the impact
C

285 of each node performance on system resilience. The so-called NARC of certain node is defined
AC

286 as a proportion of the system resilience increase, compared with including and excluding this

287 node’s contribution during system resilience calculation using Eq 1. The greater NARC indicates

288 the higher vulnerability of the corresponding node, which means the performance of this node is

289 more relevant to the deterioration of the system resilience. Therefore, to improve the system

13
 ACCEPTED MANUSCRIPT

290 resilience more effectively, the performance of nodes with high NARCs should be preferentially

291 enhanced.

292 Fig. 3 illustrates the results of NARC analysis and the layout of the vulnerable spots in case

293 area. There are 7 vulnerable spots whose NARCs are larger than 0.01. Spot , , and are

PT
294 CSO outfalls in the system. At the three locations, the generation of CSO would dramatically

RI
295 lead to the reduction of the system resilience, especially at spot . The NARC of spot nearly

296 reaches 25%, which means that if the CSO at spot is under control, the system resilience can

SC
297 be increased by 25%. For these three locations, the construction of CSO storage tanks would be

U
298 an option for the enhancement of the system resilience through the direct mitigation of CSOs
AN
299 impact. The wastewater inlet of the WWTP is located at Spot . Excessive flow exceeding the

300 treatment capacity of WWTP is the reason for which the system resilience goes down. Facing
M

301 such a type of resilience reduction, the installation of facilities that are specially operated for

302 coping with wet weather loading and enlarging WWTP’s capacity can increase UDS’s resilience
D

303 up to 12% in this case. Spot , and are located in the main trunks of the system. Urban
TE

304 flooding overflow from these nodes results in the decrease of the system resilience. Both the
EP

305 application of green infrastructure and the renovation of relevant pipes can diminish this

306 influence. If the urban flooding in these three locations can be wiped off completely, the system
C

307 resilience can be upgraded by 10%.

AC

308

309 Fig. 3. NARC analysis results and the vulnerable spots in the case area

14
 ACCEPTED MANUSCRIPT

310 3.3 Effect comparison of green and grey infrastructures for resilience enhancement

311 As shown in Fig.4, averaging all the future scenario, 20% of the total area built as green roofs

312 and permeable pavement would lead to about 30% and 33% increase of system resilience

313 respectively, while the operation of four storage tanks (52800m3 of storage volume) would give

PT
314 rise to almost 17% improvement of system resilience. By contrast, in this case area, the green

RI
315 roof and the permeable pavement as typical green infrastructure controls are much more

316 effective for the amelioration of the system resilience.

SC
317

U
318
AN
319 Fig. 4. System resilience improvement by green (green roof / permeable pavement) and grey (storage tank)

320 infrastructure.
M

321 Green infrastructures improve the performance of urban drainage system in multiple ways,
D

322 which controls the urban runoff at source and can reduce the system inputs both on total volume

323 and peak flow. Therefore, they can cut down the volume of urban flooding, CSOs, and WWTP
TE

324 influent during rainfall and effectively help against the resilience recession. Illustrated in Fig. 5,
EP

325 20% of the total area built as green roofs can respectively reduce 38%, 33%, and 28% of SevE,

326 SevS, and SevT on average under future scenarios; 20% permeable pavement can reduce 39%,
C

327 41%, and 30% of SevE, SevS, and SevT. Different from the green infrastructures, a storage tank is
AC

328 constructed at the outfall of drainage system and the main function is to intercept CSOs. Hence,

329 storage tank mainly reduces SevE to improve the system resilience. Facing different future

330 scenarios, building 4 storage tanks (i.e. 52800m3 storage volume) can lessen 30% of SevE (Fig. 5).

331 Due to the complicated spatial topological structure of the UDS, the operation of storage tanks

332 would influence the hydraulic process of the whole system, which results in a slight reduction of

15
 ACCEPTED MANUSCRIPT

333 urban flooding in this case. Shown in Fig. 5, 4 storage tanks also can lead to a decrease of SevS

334 by no more than 15%. Furthermore, comparing Fig. 5 (a), (b), (c) and (d), green roof and

335 permeable pavement can more effectively shorten the system failure duration (i.e. less time to

336 reach the turning point) and better adapt to the uncertain disturbance of the future (i.e. smaller

PT
337 shadow areas between severity curves, representing the impact gap between the different degrees

RI
338 of future change) in the case study area.

339

SC
340 Fig. 5. Impact of green (green roof / permeable pavement) and grey (storage tank) infrastructure on social,

U
341 environmental and technological severity of the system.
AN
342 Fig. 6 describes the capacity of green and grey infrastructures of the resilience improvement in

343 the case of urbanization and climate change. The ratio of the system resilience with a specific
M

344 green and grey infrastructure configuration under a future scenario (ResCiUi+G/S) to the baseline

345 resilience (ResC0U0) is used to evaluate the effectiveness of the corresponding measure. If the
D

346 ratio is equal to 1, it means that the green roof, the permeable pavement, or the storage tank can
TE

347 offset the influence of urbanization and climate change in the future. If the ratio is larger than 1,

348 it indicates the controls not only can remove the negative effects, but also can improve the
EP

349 system performance. As shown in Fig. 6, 5% of the total area built as green roofs or permeable

350 pavements can deal with the climate change in scenario C10 or the urbanization in scenario U10
C

351 or the combined scenario C5U5. Similarly, the application of one storage tank can
AC

352 counterbalance the climate change in scenario C5 or the urbanization in scenario U15 or the

353 combined scenario C5U5. However, in this case area, with the increase of the infrastructure scale,

354 green roof and permeable pavement manifests the distinctive advantages. 20% of the total area

355 built as green roofs or permeable pavements can almost offset the impacts of climate change and

16
 ACCEPTED MANUSCRIPT

356 urbanization under all future scenarios, and improve the system resilience synchronously. For

357 this case, the green infrastructures has better adaptability than the CSOs storage tank to deal with

358 the future uncertainty and higher effectiveness to enhance the resilience in the UDS.

359

PT
360 Fig. 6. Ratios of system resilience with different green or grey infrastructure configurations to baseline

RI
361 resilience.

SC
362 3.4 Trade-off between resilience enhancement and cost

363 As shown in Fig. 7, for the relationship between the system resilience and cost, the green roof

U
364 and the permeable pavement perform approximately linearly. With the increase of the
AN
365 construction area/cost of green roofs and permeable pavements, the system resilience would be

366 enhanced. Similarly, the more storage volume i.e. the more cost, the higher system resilience.
M

367 But for the storage tank control, there is a nearly logarithmic relationship between the system

368 resilience and cost. In this case area, if only the grey facility is selected, the improvement of
D

369 system resilience would eventually enter a stagnant phase after two storage tanks are introduced
TE

370 into the system. More storage volumes are not able to boost higher system resilience and the
EP

371 option of two tanks is the most cost-effective for enhancing the resilience in the case system.

372 Comparing green infrastructures with CSOs storage tank, the storage tank is even more cost-
C

373 effective with requirements of a small resilience increment, no more than 15%. If there is a
AC

374 higher demand of the resilience restoration, the green infrastructures have to be used and show a

375 dominant role because of the existence of the upper limit of storage tank control. Considering the

376 performance of resilience enhancement and corresponding cost, the efficacy of three storage

377 tanks is equivalent to it of 10% of the total area built as green roofs.

378

17
 ACCEPTED MANUSCRIPT

379 Fig. 7. System resilience and costs with different green/grey infrastructure scales under future scenarios.

380 The green infrastructures and storage tank can be jointly implemented in UDSs and Fig. 8

381 describes the performances and costs of the combined controls. Considering the future

382 uncertainties, 20% of the total area built as green roofs or permeable pavements and 4 storage

PT
383 tanks can realize nearly 50% of system resilience increase on average. Since the green

RI
384 infrastructures are the dominant measures for the resilience enhancement in the case area, the

385 system resilience will be linearly improved with the increase of the facility cost. The input of 125

SC
386 million US dollars would be able to improve 10% of system resilience in the case area.

387 As shown in Fig. 8(a), zone I is an area with the low improvement of the system resilience by

388
U
green and grey infrastructures. In this zone, compared with the baseline scenario, the largest
AN
389 improvement is 12%, and one single type of control can satisfy the requirement. The comparison

390 between green and grey options shows that the CSOs storage tank has a significant advantage.
M

391 Zone II is an area of the medium improvement, and the range of resilience increase is 12~30%.
D

392 In this zone, the improvement can be achieved both by individual control and combined control.
TE

393 Compared with two types of controls, for the similar resilience improvement, the combined

394 measures are more flexible and dominant. For example, although either 3 storage tanks (G0S3)
EP

395 or 10% of roofs built as green roofs (G10S0) can enhance the system resilience nearly by 15%,

396 the combination of 5% of roofs built as green roofs and 1 storage tank (G5S1) can achieve the
C

397 same effect with slightly lower cost. Zone III is a high improvement area, where the
AC

398 improvement is higher than 30%. In this zone, only combined controls can meet the

399 enhancement demand of system resilience. Considering the construction feasibility, maximal

400 application of green roofs and storage tanks can increase 50% of system resilience over baseline

401 scenario. Similar regularity of the three zones can also be found in Fig. 8(b) for permeable

402 pavement instead of green roof. The dots located on the upper envelope in Fig. 8(a) and 8(b)

18
 ACCEPTED MANUSCRIPT

403 represent the most cost-effective configurations for the increase of system resilience. Among

404 them, at least one or two storage tanks are needed to provide the basic capacity for boosting the

405 system resilience, while the scale of green infrastructures determines the overall level of the

406 system resilience.

PT
407

RI
408 Fig. 8. System resilience and costs with different infrastructure configurations. (a) Comparison of green roof,

409 storage tank and the combination of the two controls. (b) Comparison of permeable pavement, storage tank and

SC
410 the combination of the two controls.

U
411 3.5 Limitation
AN
412 This work is based on a specific case. Considering the local conditions, only green roof and

413 permeable pavement are discussed in this paper. Other green infrastructure measurements, such
M

414 as rain barrel and swale, haven`t been involved. Different measurements have different

415 characteristics, therefore the performance of other green infrastructure measurements can not be
D

416 deduced directly from the results of green roof and permeable pavement.
TE

417 Moreover, it should be noted that the comparison between grey infrastructure and green

418 infrastructure without consideration of the impact of spatial layout. Because this work focuses on
EP

419 the resilience performance of different infrastructures in case area. Only the feasible locations are
C

420 chosen for the green roof, permeable pavement and storage tank construction. However, the
AC

421 spatial allocation of measurements is also a very important influence factor for the system

422 resilience.

423 In addition, all the results, including the impact of climate change and urbanization, the

424 comparison between grey infrastructure and green infrastructure, and the cost-effective analysis,

425 are case-based. Different climate, hydrological features, land use, economy and other regional

19
 ACCEPTED MANUSCRIPT

426 characteristics can lead to quite different results. The conclusion may even be the opposite in

427 other areas.

428 4. Conclusion

PT
429 Finding the response solutions with high adaptability to a variety of future uncertain changes

430 and then preparing nowadays is crucial for the establishment of a sustainable urban drainage

RI
431 system. The present work contributes to the advance of sustainability assessment for urban

SC
432 drainage system, focusing on the capability in resilience enhancement of green and grey

433 infrastructure options in the face of potential system disturbance induced by urbanization and

434
U
climate change. The resilience concept used here highlights how the urban drainage system
AN
435 which was perceived as a robust system could now be judged as vulnerable, because of

436 indeterminate climate change, uncertain urbanization, and anticipated sustainable development in
M

437 the future.

D

438 The definition of resilience in urban drainage system presented in this paper considers the
TE

439 spatial and temporal dynamic process of system functionality loss and integrates the impacts of

440 the system on the quality of receiving water, urban public security, and the stability of WWTPs
EP

441 operation. This new formulation of resilience index allows a broader and richer understanding of

442 system resilience and the trade-offs for social, ecological, and infrastructure resilience. It is
C

443 useful in identifying the vulnerable spots in the system quickly. It also permits to recognize
AC

444 where decisions may be particularly relevant. The angle at which we stand and the investigations

445 we made in this paper would contribute to the comprehensive understanding on the UDSs

446 resilience.

447 Although both green and grey infrastructures could improve the system resilience, green ones

448 would have higher adaptability and resistibility to deal with the uncertain future, leading to

20
 ACCEPTED MANUSCRIPT

449 higher system sustainability. This benefit verified in this study from green infrastructure when

450 compared with grey controls seems to agree with those reported in the literatures (Cohen et al.,

451 2011; Wang et al., 2013; Tavakol-Davani et al., 2015). The performance reported for the storage

452 tank also coincides with studies (Tavakol-Davani et al., 2015; Fu et al., 2009), and the tanks

PT
453 improve the system resilience by intercepting CSO spills. Taking the cost into account, green and

RI
454 grey infrastructures have their own advantages when meeting different goals of resilience

455 improvement. Nevertheless, the paper results suggest that a combination of green and grey

SC
456 infrastructures into a “hybrid” control may have a mutually beneficial effect and offer further

457 potential in enhancing the system resilience.

458
U
In general, there is a game relationship among the system configuration in terms of green and
AN
459 grey infrastructure selection, the system resilience, and the additional cost for resilience

460 enhancement, which comes from the complexity on how a sustainable system should be. This
M

461 relationship is of great importance for the newly design and renovation of urban drainage system
D

462 to adapt to the urbanization and climate change in future.

TE

463 Acknowledgment
EP

464 This work was supported by the National Natural Science Foundation of China (51308320 &

465 71473148).
C
AC

466 Supporting information

467 Table S1. Parameters of future scenarios. C0 means no climate change; C5, C10, C15 and C20

468 mean the rainfall intensity time series raise by 5%, 10%, 15% and 20%. U0 means no

469 urbanization; U5, U10, U15 and U20 mean the impervious area increase by 5%, 10%, 15% and

470 20%.

21
 ACCEPTED MANUSCRIPT

Scenario Average value of rainfall intensity (mm/h) Average

name percentage of
1 year 3 years 5 years 10 years 30 years
imperious
C0U0 35.0 45.0 49.6 55.9 65.9 49.0%

PT
C5/C5U0 36.7 47.2 52.1 58.7 69.2 49.0%
C10/C10U0 38.5 49.5 54.6 61.5 72.5 49.0%
C15/C15U0 40.2 51.7 57.1 64.3 75.8 49.0%

RI
C20/C20U0 42.0 54.0 59.5 67.1 79.1 49.0%
U5/C0U5 35.0 45.0 49.6 55.9 65.9 51.5%

SC
C5U5 36.7 47.2 52.1 58.7 69.2 51.5%
C10U5 38.5 49.5 54.6 61.5 72.5 51.5%

U
C15U5 40.2 51.7 57.1 64.3 75.8 51.5%
C20U5 42.0 54.0 59.5 67.1 79.1 51.5%
AN
U10/C0U10 35.0 45.0 49.6 55.9 65.9 53.9%
C5U10 36.7 47.2 52.1 58.7 69.2 53.9%
M

C10U10 38.5 49.5 54.6 61.5 72.5 53.9%

C15U10 40.2 51.7 57.1 64.3 75.8 53.9%
C20U10 42.0 54.0 59.5 67.1 79.1 53.9%
D

U15/C0U15 35.0 45.0 49.6 55.9 65.9 56.4%

TE

C5U15 36.7 47.2 52.1 58.7 69.2 56.4%

C10U15 38.5 49.5 54.6 61.5 72.5 56.4%
C15U15 40.2 51.7 57.1 64.3 75.8 56.4%
EP

C20U15 42.0 54.0 59.5 67.1 79.1 56.4%

U20/C0U20 35.0 45.0 49.6 55.9 65.9 58.8%
C

C5U20 36.7 47.2 52.1 58.7 69.2 58.8%

AC

C10U20 38.5 49.5 54.6 61.5 72.5 58.8%

C15U20 40.2 51.7 57.1 64.3 75.8 58.8%
C20U20 42.0 54.0 59.5 67.1 79.1 58.8%
471

22
 ACCEPTED MANUSCRIPT

472

473 Fig. S1. Process curves of selected rainfalls. The five events are all Chicago pattern rainfalls with

474 duration of 1 hour and peak location r=0.4. The rainfall intensity is calculated based on the local

PT
475 IDF relationship: i=(11.8+7.07lgP)/(t+10)0.708, where i is the rainfall intensity (mm/min), P is the

476 rainfall return period (a) and t is the rainfall duration (min).

RI
477

SC
478 Fig. S2. Basic information of case study area. Four potential positions for storage tanks are

U
479 identified and the less number marked in the figure indicates a prior selection preference. The

appropriate storage volume for each tank is estimated to be 13200m3. Therefore, four scenarios
AN
480

481 for grey infrastructure are selected as 1 storage tank (position 1, total volume of 13200m3), 2
M

482 storage tanks (position 1 & 2, total volume of 26400m3), 3 storage tanks (position 1, 2 & 3, total

483 volume of 39600m3) and 4 storage tanks (position 1, 2, 3 & 4, total volume of 52800m3). The
D

484 area which is available for implementing green roof occupies more than 20% of the total area
TE

485 and it is the same with permeable pavement. For green infrastructure scenarios, different

486 percentages of the total area for green roof and permeable pavement placement are selected as 5,
EP

487 10, 15 and 20%. The 4 storage tank and 4 green infrastructure scenarios are crossed to generate
C

488 16 combined control scenarios for green roof and permeable pavement respectively.
AC

489 The cost of storage tank is estimated according to the cost and volume data of 7 real storage

490 tanks in Kunming. The relationship for a storage tank is: Cost (thousand $) = 1.2*Volume (m3) +

491 16000. The cost of green roof and permeable pavement are selected as 100$/m2 and 80$/m2

492 respectively, which are determined based on the average value of several literatures (see list

493 below).

23
 ACCEPTED MANUSCRIPT

494 Chui, T. F. M., Liu, X., Zhan, W., 2016. Assessing cost-effectiveness of specific LID practice

495 designs in response to large storm events. J. Hydrol. 533, 353-364.

496 Jia, H., Yao, H., Tang, Y., Shaw, L. Y., Field, R., Tafuri, A. N., 2015. LID-BMPs planning for

PT
497 urban runoff control and the case study in China. J. Environ. Manage. 149, 65-76.

498 Joksimovic, D., Alam, Z., 2014. Cost efficiency of low impact development (LID) stormwater

RI
499 management practices. Procedia Engineering 89, 734-741.

SC
500 Montalto, F., Behr, C., Alfredo, K., Wolf, M., Arye, M., Walsh, M., 2007. Rapid assessment of

501 the cost-effectiveness of low impact development for CSO control. Landscape Urban Plan. 82

502 (3), 117-131.

U
AN
503
M

504 Fig. S3. System social, environmental and technological severity under different climate change
D

505 scenarios.
TE

506
EP

507 Fig. S4. System social, environmental and technological severity under different urbanization

508 scenarios.
C

References
AC

509

510 Ahern, J., 2011. From fail-safe to safe-to-fail: Sustainability and resilience in the new urban world. Landscape

511 Urban Plan. 100 (4), 341-343.

512 Arnbjerg-Nielsen, K., 2012. Quantification of climate change effects on extreme precipitation used for high

513 resolution hydrologic design. Urban Water J. 9 (2), 57-65.

514 Butler, D., Farmani, R., Fu, G., Ward, S., Diao, K., Astaraie-Imani, M., 2014. A new approach to urban water

24
 ACCEPTED MANUSCRIPT

515 management: Safe and sure. Procedia Engineering 89, 347-354.

516 Casal-Campos, A., Fu, G., Butler, D., Moore, A., 2015. An integrated environmental assessment of green and

517 gray infrastructure strategies for robust decision making. Environ. Sci. Technol. 49 (14), 8307-8314.

518 Chen, J., Beck, M. B., 1997. Towards designing sustainable urban wastewater infrastructures: a screening

PT
519 analysis. Water Sci. Technol. 35(9), 99-112.

520 Cohen, J. P., Field, R., Tafuri, A. N., Ports, M. A., 2011. Cost Comparison of Conventional Gray Combined

RI
521 Sewer Overflow Control Infrastructure versus a Green/Gray Combination. J. Irrig. Drain. E. – ASCE 138 (6),

522 534-540.

SC
523 Cohen, R., Erez, K., Ben-Avraham, D., Havlin, S., 2000. Resilience of the Internet to random breakdowns.

524 Phys. Rev. Lett. 85 (21), 4626.

U
525 De Bruijn, K. M., 2004. Resilience and flood risk management. Water Policy 6 (1), 53-66.
AN
526 Francis, R., Bekera, B., 2014. A metric and frameworks for resilience analysis of engineered and infrastructure

527 systems. Reliab. Eng. Syst. Safe. 121, 90-103.

M

528 Fu, G., Khu, S. T., Butler, D., 2009. Optimal distribution and control of storage tank to mitigate the impact of

529 new developments on receiving water quality. J. Environ. Eng. - ASCE 136 (3), 335-342.
D

530 Golz, S., Schinke, R., Naumann, T., 2015. Assessing the effects of flood resilience technologies on building
TE

531 scale. Urban Water J. 12 (1), 30-43.

532 Grum, M., Jørgensen, A. T., Johansen, R. M., Linde, J. J., 2006. The effect of climate change on urban
EP

533 drainage: an evaluation based on regional climate model simulations. Water Sci. Technol. 54 (6-7), 9-15.

534 Hatt, B. E., Fletcher, T. D., Walsh, C. J., Taylor, S. L., 2004. The influence of urban density and drainage
C

535 infrastructure on the concentrations and loads of pollutants in small streams. Environ. Manag. 34, 112–124.
AC

536 Henry, D., Ramirez-Marquez, J. E., 2012. Generic metrics and quantitative approaches for system resilience as

537 a function of time. Reliab. Eng. Syst. Safe. 99, 114-122.

538 Hossain, F., Arnold, J., Beighley, E., Brown, C., Burian, S., Chen, J., Tidwell, V., 2015. Local-to-regional

539 landscape drivers of extreme weather and climate: implications for water infrastructure resilience. J. Hydrol.

540 Eng. 20 (7), 02515002.

541 Huong, H. T. L., Pathirana, A., 2013. Urbanization and climate change impacts on future urban flooding in Can

25
 ACCEPTED MANUSCRIPT

542 Tho City, Vietnam. Hydrol. Earth Syst. Sci. 17, 379–394.

543 Karamouz, M., Hosseinpour, A., Nazif, S., 2011. Improvement of urban drainage system performance under

544 climate change impact: Case study. J. Hydrol. Eng. 16(5), 395-412.

545 Labaka, L., Hernantes, J., Sarriegi, J. M., 2015. Resilience framework for critical infrastructures: An empirical

PT
546 study in a nuclear plant. Reliab. Eng. Syst. Safe. 141, 92-105.

547 Liu, W., Chen, W., Peng, C., 2014. Assessing the effectiveness of green infrastructures on urban flooding

RI
548 reduction: A community scale study. Ecol. Model. 291, 6-14.

549 Liu, W., Chen, W., Peng, C., 2015. Influences of setting sizes and combination of green infrastructures on

SC
550 community’s stormwater runoff reduction. Ecol. Model. 318, 236-244.

551 McDaniels, T., Chang, S., Cole, D., Mikawoz, J., Longstaff, H., 2008. Fostering resilience to extreme events

U
552 within infrastructure systems: Characterizing decision contexts for mitigation and adaptation. Global Environ.
AN
553 Chang. 18 (2), 310-318.

554 Meerow, S., Newell, J. P., Stults, M., 2016. Defining urban resilience: A review. Landscape Urban Plan. 147,
M

555 38-49.

556 Mugume, S. N., Diao, K., Astaraie-Imani, M., Fu, G., Farmani, R., Butler, D., 2015. Enhancing resilience in
D

557 urban water systems for future cities. Water Sci. Tech. – W. Sup. 15 (6), 1343-1352.
TE

558 Mugume, S. N., Gomez, D. E., Butler, D., 2014. Quantifying the resilience of urban drainage systems using a

559 hydraulic performance assessment approach. ICUD 2014, Sarawak, Malaysia, September 7–12.
EP

560 Mugume, S. N., Gomez, D. E., Fu, G., Farmani, R., Butler, D., 2015. A global analysis approach for

561 investigating structural resilience in urban drainage systems. Water Res. 81, 15-26.
C

562 Notaro, V., Liuzzo, L., Freni, G., La Loggia, G., 2015. Uncertainty analysis in the evaluation of extreme
AC

563 rainfall trends and its implications on urban drainage system design. Water 7 (12), 6931-6945.

564 Ouyang, M., Dueñas-Osorio, L., Min, X., 2012. A three-stage resilience analysis framework for urban

565 infrastructure systems. Struct. Saf. 36, 23-31.

566 Peterson, G., Allen, C. R., Holling, C. S., 1998. Ecological resilience, biodiversity, and scale. Ecosystems 1

567 (1), 6-18.

568 Pugh, T. A., MacKenzie, A. R., Whyatt, J. D., Hewitt, C. N., 2012. Effectiveness of green infrastructure for

26
 ACCEPTED MANUSCRIPT

569 improvement of air quality in urban street canyons. Environ. Sci. Technol. 46 (14), 7692-7699.

570 Semadeni-Davies, A., Hernebring, C., Svensson, G., Gustafsson, L. G., 2008. The impacts of climate change

571 and urbanisation on drainage in Helsingborg, Sweden: Combined sewer system. J. Hydrol. 350, 114–125.

572 Shafieezadeh, A., Burden, L. I., 2014. Scenario-based resilience assessment framework for critical

PT
573 infrastructure systems: Case study for seismic resilience of seaports. Reliab. Eng. Syst. Safe. 132, 207-219.

574 Tavakol-Davani, H., Burian, S. J., Devkota, J., Apul, D., 2015. Performance and cost-based comparison of

RI
575 green and gray infrastructure to control combined sewer overflows. J. Sustainable Water Built Environ. 2 (2),

576 04015009.

SC
577 Vojinovic, Z., Sahlu, S., Torres, A. S., Seyoum, S. D., Anvarifar, F., Matungulu, H., Barreto, W., Savic, D.,

578 Kapelan, Z., 2014. Multi-objective rehabilitation of urban drainage systems under uncertainties. J.

U
579 Hydroinform. 16 (5), 1044-1061.
AN
580 Walker, B., Holling, C. S., Carpenter, S. R., Kinzig, A., 2004. Resilience, adaptability and transformability in

581 social-ecological systems. Ecol. Soc. 9 (2), 5.

M

582 Wang, R., Eckelman, M. J., Zimmerman, J. B., 2013. Consequential environmental and economic life cycle

583 assessment of green and gray stormwater infrastructures for combined sewer systems. Environ. Sci. Technol.
D

584 47 (19), 11189-11198.

TE

585 Willems, P., Arnbjerg-Nielsen, K., Olsson, J., Nguyen, V. T. V., 2012. Climate change impact assessment on

586 urban rainfall extremes and urban drainage: Methods and shortcomings. Atmos. Res. 103, 106-118.
EP

587 Yazdanfar, Z., Sharma, A., 2015. Urban drainage system planning and design - challenges with climate change

588 and urbanization: A review. Water Sci. Technol. 72 (2), 165-179.

C

589 Yin, J., Yan, D., Yang, Z., Yuan, Z., Yuan, Y., Zhang, C., 2016. Projection of extreme precipitation in the
AC

590 context of climate change in Huang-Huai-Hai region, China. J. Earth Syst. Sci. 125 (2), 417-429.

591 Zhou, Q., 2014. A review of sustainable urban drainage systems considering the climate change and

592 urbanization impacts. Water 6 (4), 976-992.

27
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

Highlights

A new formulation of resilience in urban drainage system was presented.

Both green and grey infrastructures could improve the system resilience.

Green roof and permeable pavement would have higher adaptability and

PT
resistibility.

There is a game for the green and grey infrastructure selection.

RI
U SC
AN
M
D
TE
C EP
AC

View publication stats