The cost of hybrid waste water systems: A systematic framework for specifying minimum cost-connection rates

Accepted Manuscript The cost of hybrid waste water systems: A systematic framework for specifying minimum cost-connection rates Sven Eggimann, Bernhard Truffer, Max Maurer PII:

S0043-1354(16)30576-0

DOI:

10.1016/j.watres.2016.07.062

Reference:

WR 12257

To appear in:

Water Research

Received Date: 21 April 2016 Revised Date:

23 July 2016

Accepted Date: 25 July 2016

Please cite this article as: Eggimann, S., Truffer, B., Maurer, M., The cost of hybrid waste water systems: A systematic framework for specifying minimum cost-connection rates, Water Research (2016), doi: 10.1016/j.watres.2016.07.062. 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

What is the cost minimal connection rate? Central sanitation

70%

60%

RI PT

40%

100%

Ø

Ø

Ø

Ø

$

Ø

AC C

EP

TE D

M AN U

SC

0%

Decentral sanitation

ACCEPTED MANUSCRIPT 1

The cost of hybrid waste water systems: a

2

systematic framework for specifying

3

minimum cost-connection rates

4

Eggimann Sven

1,2*

1,3

, Truffer Bernhard , Maurer Max

1, 2

1

Eawag, Swiss Federal Institute of Aquatic Science and Technology, 8600 Dübendorf, Switzerland.

7 8

2

Institute of Civil, Environmental and Geomatic Engineering, ETH Zürich, 8093 Zurich, Switzerland.

9 10

3

Faculty of Geosciences, Utrecht University, Heidelberglaan 2, NL-3584 CS Utrecht, The Netherlands.

RI PT

5 6

*Corresponding author: Eawag, Swiss Federal Institute of Aquatic Science and Technology. Überlandstrasse 133, 8600 Dübendorf, Switzerland.

13

E-mail: [email protected] (S. Eggimann)

14 15

Keywords:

16

Urban water management, engineering economics, urban structural

17

unit, geographical information system, decentralised waste water

18

management, sustainability transition

19

Abstract

20

To determine the optimal connection rate (CR) for regional waste

21

water treatment is a challenge that has recently gained the attention

22

of academia and professional circles throughout the world. We

23

contribute to this debate by proposing a framework for a total cost

24

assessment of sanitation infrastructures in a given region for the

25

whole range of possible CRs. The total costs comprise the treatment

26

and transportation costs of centralised and on-site waste water

27

management systems relative to specific CRs. We can then identify

28

optimal CRs that either deliver waste water services at the lowest

29

overall regional cost, or alternatively, CRs that result from

30

households freely choosing whether they want to connect or not. We

31

apply the framework to a Swiss region, derive a typology for regional

32

cost curves and discuss whether and by how much the empirically

33

observed CRs differ from the two optimal ones. Both optimal CRs

AC C

EP

TE D

M AN U

SC

11 12

1

ACCEPTED MANUSCRIPT 34

may be reached by introducing specific regulatory incentive

35

structures.

36

1

Sanitation services in a region may in principle be provided by

RI PT

37

Introduction

centralised or decentralised on-site waste water management

39

systems (WMS) (Libralato et al. 2012). On-site WMS enable waste

40

water to be treated geographically close to the point of generation

41

(Tchobanoglous and Leverenz 2013), making costly investments in

42

sewer networks obsolete and potentially allowing cost savings.

43

Despite the potential advantages, however, centralised WMS have

44

gained much higher market shares in most OECD countries over the

45

past century. The primary rationale for this was to assure high levels

46

of ‘urban hygiene’ (O'Flaherty 2005, Sedlak 2014). Moreover,

47

centralised WMS were promoted by public regulators because of

48

compatibility with currently existing systems, known manageability,

49

well-defined performance as well as economies of scale in both

50

waste water treatment and sewer management (Townend 1959,

51

Downing 1969, Abd El Gawad and Butter 1995, Libralato et al. 2012).

52

Over the years, institutions, organisations and the technology have

53

co-evolved, leading to shared values, a professional culture based on

54

civil engineering competences, and particular organisational forms

55

dominated by utilities under public ownership (Dominguez 2008,

56

Kiparsky et al. 2013, Fuenfschilling and Truffer 2014, Fane and Fane

57

2005, Lieberherr and Truffer 2015, Lieberherr and Fuenfschilling

58

2016). These alignments created strong path dependencies (Arthur

59

1989), so that today’s catchments are dominated by large centralised

60

waste water treatment plants (WWTP) and extensive sewer networks

61

connecting large percentages of the population. Empirically, we

62

observe a wide variety of connection rates (CR): whereas most

AC C

EP

TE D

M AN U

SC

38

2

ACCEPTED MANUSCRIPT emerging economies and developing countries are characterised by

64

very low (typically << 50%) CRs (UN 2015), some OECD countries

65

(e.g. Switzerland, Austria, the Netherlands and the United Kingdom)

66

have pushed for very high CRs (CRpresent) of >95%, whereas other

67

OECD countries (e.g. Ireland, Slovenia or Poland) have a CRpresent of

68

between 60 and 70% (OECD 2015).

69

RI PT

63

The long-term superiority of very high CR has lately been

questioned, and this has led to a call for a ‘sustainability transition’

71

towards more hybrid configurations combining centralised and on-

72

site WMS (Fane and Fane 2005, Daigger 2007, Truffer et al. 2010,

73

Larsen et al. 2013, Marlow et al. 2013). A wide range of criteria (e.g.

74

technical, environmental, public-health related, institutional, social,

75

economic) can be used to determine the optimal mixing rate. In

76

recent years, however, we can observe an increasing predominance

77

of economic efficiency criteria in the planning of network-based

78

infrastructures (Knops 2008). Economic assessments of optimal

79

infrastructure dimensioning have also gained increasing attention in

80

the field of water management, not only for waste water (Eggimann

81

et al. 2015, Lee et al. 2013), but also for drinking water (Poustie et al.

82

2014, Guo and Englehardt 2015), hydro power (Kaundinya et al.

83

2009) and seawater desalination (Shahabi et al. 2015). This

AC C

EP

TE D

M AN U

SC

70

84

heightened interest is due to strained public budgets, often leading to

85

infrastructural underinvestment (WEF 2010), the demand for more

86

infrastructure flexibility and recent advances in on-site treatment

87

technology. Furthermore, a modular approach to infrastructure

88

planning is becoming increasingly cost competitive: new sensor and

89

communication technologies allow automation and mass production

90

which drive down the cost of small standardised units (Dahlgren et al.

91

2013). Determining the optimal connection rate (OCR) therefore

92

remains a relevant question to reconsider. 3

ACCEPTED MANUSCRIPT 93

In the present paper, we focus exclusively on cost assessments, as they often play an important role in designing WMS (Maurer et al.

95

2006). The goal is to develop an encompassing framework for

96

assessing the total costs of hybrid WMS (Tchobanoglous and

97

Leverenz 2013) in a given region. Even though much effort has been

98

spent on the cost considerations of WMS (Townend 1959, Downing

99

1969, Adams et al 1972, Etnier et al. 2000, Hamilton et al. 2004,

RI PT

94

Maurer et al. 2010, Libralato et al. 2012, Eggimann et al. 2015), there

101

is a paucity of conceptual work focusing on systematic total cost

102

assessments. We build on an extensive body of work and present a

103

framework within which we deduce generic cost curves for all key

104

cost elements of a hybrid WMS. On the basis of these

105

considerations, we will provide alternative interpretations of the OCR

106

depending on specific institutional arrangements and organisational

107

set-ups of providers of WMS services. This will enable us to discuss

108

just what ‘more sustainable’ WMS configurations in specific regions

109

could be, and in particular to discuss to what extent the CRpresent

110

deviates from the various OCRs.

111

2

112

2.1

TE D

M AN U

SC

100

EP

Material and methods

AC C

Framework for total cost assessment

113

This section starts by introducing the general assumptions of our

114

framework (Section 2.1.1), and continues by identifying all key cost

115

components of centralised and on-site WMS needed for a total cost

116

assessment of hybrid WMS in a region (Section 2.1.2).

117

2.1.1

118

The framework for assessing total costs of hybrid WMS in a region

119

presented here draws on the following general assumptions:

General assumptions

4

ACCEPTED MANUSCRIPT •

120

Households and utility operators prefer each system only on the basis of average cost considerations.

121 •

122 123

1

All households have to be served either by being connected to the sewers or installing on-site WMS.

•

124

The average regional total costs at each CR are defined by the average per capita costs of both systems as well as

126

being annualised on the basis of the expected life-spans of

127

the corresponding assets.

RI PT

125

• We use average costs as a meaningful approximation for

129

individual household sanitation costs. We are aware that

130

actual tariff systems often diverge from these average costs,

131

as they may include block tariffs, subsidies, base fees or

132

connection fees (OECD 2010).

M AN U

SC

128

• To ensure human and environmental health, centralised and

134

on-site WMS need to fulfil the same functionality and provide

135

an equivalent service. This implies that on-site WMS have to

136

be equipped with treatment performance comparable to that

137

of centralised WWTPs, and that on-site effluent disposal is

138

possible (e.g. infiltration or on-site discharge into waters). We

139

consequently assume that the sewers are built exclusively for

EP

TE D

133

140

waste water transportation and no synergies with storm water evacuation have to be accounted for (cf. Section 4.3).

AC C

141

• We assume that our region consists only of households and

142

143

aggregated households in urban structural units (see Section

144

2.2.2) and no industry.

1

2

We conduct the entire cost assessment procedure by means of average cost calculations for each system. As we are interested in long-term optimal equilibrium solutions, this assumptions may be justified.

2

However, this restriction is not decisive for the general argument that we develop and merely implies that the cost curves of the 5

ACCEPTED MANUSCRIPT • We neglect transaction costs, i.e. the costs of switching from

145 146

147 148

one WMS to another.

2.1.2

Total costs of hybrid WMS

The total WMS costs C can be subdivided into waste water treatment C and waste water transport C  costs. For

150

 centralised WMS, treatment occurs in one large WWTP C

151

 whereas for decentralised WMS, treatment is on-site C .

152

Transportation is either road-based in case of decentralised WMS

153

  C  or sewer-based for centralised WMS C  . The total

154

regional cost C

 

C

SC

 

of a WMS can thus be specified as:

    = C   + C + C   + C   

156

M AN U

155

RI PT

149

(1)



Figure 1 shows the generic functional forms of the cost  

components of C

as a function of the CR along the respective

158

sensitivity bands. The average total costs of either system (e, f) can

159

be calculated on the basis of the cost function of the centralised (a, c)

160

and decentralised WMS (b, d). Finally, the average total regional

161

costs at a specific CR (assuming that this CR corresponds to a share

162

p of households connected to the centralised system, whereas 1-p

AC C

EP

TE D

157

163

have on-site treatment) can be expressed as the weighted sum

164

(dotted red line in g):

165

 

C

  (CR) = p ∗ C (CR) + (1 − p) ∗ C (&')

(2)

166

Each of the specific shapes of the cost curves is based on

167

different assumptions (outlined below): they are either derived

centralised WMS have a much bumpier shape than our idealised representation. 6

ACCEPTED MANUSCRIPT directly from explicit cost data (a, b) or are model-based (c, d). The

169

shapes are idealised, i.e. they vary depending on the specific case

170

study. However, although we have varied the underlying

171

assumptions to obtain cost ranges for each cost element, we find that

172

the behaviour of each cost curve can be described in fairly generic

173

terms:

174

RI PT

168

 Centralised treatment (C , Figure 1a): the prevailing key

economic argument given in the literature to realise high CRs

176

relates to economies of scale (inter alia Townend 1959, Downing

177

1969, Libralato et al. 2012). The likely decrease in average per

178

capita costs for a large WWTP is inversely proportional to the

179

number of households connected. In the literature, it is commonly

180

implicitly assumed that the WWTP perfectly fits the demand of the

181

connected users for each CR, and idle capacities are neglected.

182

However, as investments in WWTP are typically based on the

183

peak performance during the planning horizon (typically 20 to 30

184

years) (Hug et al. 2010), neglecting idle capacities underestimates

185

treatment costs in catchments with positive or negative growth.

186

The sensitivity indication in Figure 1a reflects two extreme cases

187

of idle capacities: the bottom border neglects idle capacities

188

altogether while the top border indicates an investment scenario

AC C

EP

TE D

M AN U

SC

175

189

that considers maximum idle capacities. For the latter scenario,

190

we calculate an initial expenditure of one WWTP serving the

191

whole catchment and distribute this investment equally amongst

192

the connected population at each CR.

193

 Centralised transportation (C  , Figure 1c): sewer

194

networks enable the transportation of waste water to the WWTP.

195

Sewer construction and operation costs are heavily influenced by

196

geography, settlement distribution or population density. We 7

ACCEPTED MANUSCRIPT consequently find decreasing marginal costs for higher CRs and

198

complex cost functions depending on the geography (cf. Adams et

199

al 1972, Hamilton et al. 2004, Maurer et al. 2010, Eggimann et al.

200

2015). In reality, most sewer systems are built up iteratively,

201

where each new settlement structure to be connected leads to

202

particular cost curves in terms of shape and cost level. Thus

203

clustered settlement structures prevent constant cost increases

204

and lead to ‘jumps’ (Zvoleff et al. 2009) in the cost curve. Figure

205

1c shows a generic sewer cost function with increasing average

206

per capita costs for higher CRs as a result of heterogeneous

207

settlement structures, and correspondingly higher costs for

208

connecting more distant settlements. The sensitivity indication

209

reflects differences in the cost curve depending on the catchment

210

context factors outlined above. Whereas cost curves can be

211

derived from the detailed cost data of existing networks, model-

212

based approaches allow us to overcome the lack of data or the

213

influence of legacy infrastructures and to systematically assess

214

sewer costs across different catchments.

TE D

M AN U

3

SC

RI PT

197

 Decentralised treatment (C , Figure 1b): the costs of on-

EP

215

site treatment are largely independent of specific CRs, and the

217

generic cost function is thus constant (Figure 1b). However, the

AC C

216

218

installation costs may differ depending on local conditions (e.g.

219

rural or urban setting) and on the system type (Singh et al. 2015).

220

We also find economies of scale for on-site treatment, i.e. per

3

For the idealistic cost curve representation in Figure 1c we assume that the dimension of the CR is ordered in a way that enables a monotonic presentation of the sewer cost curve, i.e. the x-axis proceeds from houses that are near the WWTP to those more distant from it. At the same time, we assume that the settlement density is highest around the WWTP and decreases over distance. 8

ACCEPTED MANUSCRIPT capita costs are typically lower for a 20-person system than a 4-

222

person one. The level of the cost curve can consequently differ,

223

which we represent by the respective shaded area. We assume

224

that the housing structure does not change for different CR, so

225

that the cost curve remains constant over the entire CR range.

RI PT

221

 Decentralised transportation (C  , Figure 1d): in the case

226

of on-site WMS, well-functioning operation and maintenance

228

(O&M) schemes are necessary to achieve full functionality. Road-

229

based transportation needs result from professionals having to

230

access the plants as well as from residual evacuation. For

231

operating and managing on-site WMS, we find economies of

232

density (Eggimann et al. 2016), i.e. cost savings due to the

233

numbers and spatial proximity of on-site WMS. However, this

234

effect is limited to a rather small range of treatment plant

235

densities. A generic cost function describing all transport-related

236

costs is given in Figure 1d, where the range of different cost

237

functions is due to different O&M concepts.

TE D

M AN U

SC

227

Total costs: the total cost curve of centralised WMS is shown

239

in Figure 1e with its characteristic ‘u-shaped’ form (Adams et al.

240

1972). The total cost curve of on-site WMS results in a ‘hockey-

241

stick’ shape as seen in Figure 1f (Eggimann et al. 2016). The

242

resulting total average regional costs (C

243

now be derived from the total costs of both centralised and

244

decentralised WMS (Equation 2).

AC C

EP

238

245 246

2.1.3

 

) for hybrid WMS can

Optimal connection rate

The total regional cost curve and the respective centralised and

247

on-site total cost curves exhibit some notable characteristics (Figure

248

1g, Figure 2). Firstly, we argue that the basic shapes of these curves 9

ACCEPTED MANUSCRIPT are quite generically valid: both the ‘u-form’ shape of the centralised

250

system and the ‘hockey stick’ of on-site systems have been identified

251

in earlier literature (Adams et al. 1972, Eggimann et al. 2016).

252

Secondly, by ignoring the trivial cases where one curve dominates

253

the other, we expect two intersection points between these curves to

254

exist almost independently of the specific cost characteristics.

255

Thirdly, there will be a minimum on the regional total cost curve.

RI PT

249

256

These points may be interpreted as different candidates for potential OCRs. The difference between these OCRs and the

258

CRpresent can be interpreted as the regional cost improvement

259

potential of WMS. The two relevant OCRs can be characterised as

260

follows:

•

261

M AN U

SC

257

The OCRregion is defined by the minimum on the regional total cost curve where the average aggregated costs for the entire

263

region are minimal. However, the specific costs of the

264

different systems are not equal at this point, in view of higher

265

costs for on-site treatment. •

266

The lower intersection point OCRequalcost marks the CR where the specific costs of the two WMS options are equal. This

EP

267

TE D

262

268

point would be reached spontaneously if all households could opt for the cheapest system in their specific location.

4

AC C

269

270

The sanitation costs for each household are the same,

271

irrespectively of the system choice.

4

The higher-level intersection would fulfill the criterion of equal costs equally well, but it represents a substantially higher level of total regional costs, so we do not elaborate further on its significance. 10

ACCEPTED MANUSCRIPT 272

2.2

Case study application In this section, we apply the framework outlined here to an

274

empirical case-study region in order to test whether and how the

275

different OCR can be identified. We aim to derive general cost

276

patterns in the form of a configuration typology from the various

277

case-study catchments.

278

2.2.1

Canton of Glarus

We select the Canton of Glarus, a region with a population of

SC

279

RI PT

273

280

~40,000 in the north-east of Switzerland and covering an area of 685

281

km (Figure 3). We chose Glarus because it is a diverse region in

282

terms of topography and settlement distribution which provides

283

diverse contexts with respect to cost-curve configurations. This can

284

be seen in the fact that we already find different CRs there (Figure 3).

285

The region underwent an organisational reform in 2011 - the ‘Glarner

286

Gemeindereform’ - where 25 communities were merged into three.

287

We will consequently calculate cost curves for both sets of

288

communities, before and after the merger.

M AN U

TE D

289

2

In Switzerland, waste water catchments are not organised purely along administrative borders but depend on topographic settings.

291

That is why we currently find three different waste water catchments,

292

indicated with red borders in Figure 3. In this paper, we only focus on

293

the largest WWTP catchment, which we henceforth call the ‘case-

294

study catchment’.

295

2.2.2

AC C

EP

290

296

Aggregation of urban structural units

For regional or medium-scale analysis, data aggregation is

297

generally required to reduce computational complexity (Haggag and

298

Ayad 2002). To run the heuristic sewer generation algorithm

299

efficiently (Section 2.2.3), we choose an aggregation technique

11

ACCEPTED MANUSCRIPT based on urban structural units (USU), as sanitation planning is

301

closely linked to urban patterns (Spirandelli 2015, Bach et al. 2015).

302

USU are defined as ‘areas with a physignomically homogeneous

303

character, which are marked in the built-up area by a characteristic

304

formation of buildings and open spaces’ (Wickop 1998). With the

305

emergence of geographical information systems, USU are

306

increasingly used in different contexts (inter alia Osmond 2010,

307

Wang et al. 2013, Behling et al. 2015), but have so far been rarely

308

applied to the field of sanitation (for exceptions, see Schiller 2010,

309

Eggimann 2013). Different approaches have been developed to

310

classify the physiognomies of urban building which can be used to

311

define USU (Steiniger et al. 2008, Meinel and Burgdorf 2008,

312

Lüscher et al. 2009). To derive USU, we choose an approach based

313

on the spatial intersection of linear urban features (street and railway

314

networks) within the settlement area. This intersection is followed by

315

a post-processing step in which USU containing no buildings are

316

removed and smaller USU (<0.5 km ) are merged with neighbouring

317

ones. To estimate the population per USU, we disaggregate the

318

community population data according to a volumetric estimation by

319

Lwin and Murayama (2009). The population data of the USU

320

centroids is in a last step projected to the closest point on the street

321

network (see Figure 4).

M AN U

SC

RI PT

300

AC C

EP

TE D

2

322

2.2.3

Key cost components

323

For the cost calculation of the case study, we convert all local

324

currencies to US$ using purchase power parities for the year 2014

325

(World Bank, 2015). All levelised costs are given in annuities (A)

326

calculated from the net present value (NPV):

327

A = NPV

-. (-/0) -. /0



(3)

12

ACCEPTED MANUSCRIPT where q is the discount rate + 1 and n the number of years over

329

which the infrastructure is depreciated (Crundwell 2008). We adjust

330

on-site treatment costs to the year 2014 using conversion factors for

331

the U.S. price index (U.S. Census Bureau 2015). We derive the

332

various cost elements as follows (cf. Section 2.1.2):

RI PT

328

Central treatment: we use typical Swiss replacement costs to

334

estimate the large-scale WWTP costs (Figure 5). As centralised

335

costs are very unreliable for small treatment plants, we use on-

336

site treatment costs for plants smaller than 20 population

337

equivalents (PE).

M AN U

SC

333

Central transportation: in order to estimate the costs of the

339

sewer network along the whole CR spectrum, we adapt and apply

340

a heuristic sewer network generation algorithm developed by

341

Eggimann et al. (2015) which is based on sewer-design principles

342

from the real world. The adapted algorithm allows us to iteratively

343

simulate a sewer network starting from a single connected

344

household up to full catchment connection to a single WWTP. For

345

in-depth explanations of the applied algorithm and the terminology

346

used as outlined below, we refer to Eggimann et al. (2015).

347

Compared to the original algorithm, we make three adaptations:

AC C

EP

TE D

338

348

1.

We do not consider semi-decentralised solutions but

349

iteratively simulate an interconnected network. So there is

350

no need to execute the merging module, as only a single

351

WWTP exists at each iterative step. We therefore always

352

force a sewer connection in each iteration, and thus reduce

353

the system design module to two options.

354 355

2.

Due to this conceptual change, the introduction of further distance weighting factors of the Prim-based expansion

13

ACCEPTED MANUSCRIPT module yields visually more realistic sewer networks. At

357

each iteration step, we check whether there is a local

358

elevation or depression (‘hill’ or ‘valley’) <25m between two

359

nodes under consideration. If so, we multiply this distance

360

by a weighting factor dw (dw = 30). Moreover, we always

361

weight nodes which are topographically lower by dw (dw =

362

10), as pumping is necessary and is to be avoided. The

363

only exception is where the nodes under consideration form

364

edges leading to the WWTP. This is because the network

365

position of the WWTP can be switched with the considered

366

node so that pumping can also be avoided.

RI PT

356

3.

We remove the a* algorithm in the case of missing

M AN U

367

SC

5

368

connections to street networks to reduce the computational

369

burden, and use straight-line distance approximations

370

instead.

Decentralised treatment: it is challenging to determine the

TE D

371

average on-site treatment costs because a wide variety of

373

possible system alternatives exist (Maurer et al. 2012). But even

374

more importantly, the functional equivalence of on-site WMS is

375

hard to operationalise. We therefore opt for a fail-safe option and

376

include disinfection costs derived from systems based on sodium

AC C

EP

372

377

hypochlorite and UV radiation (WERF 2010). We additionally

378

assume that further costs arise due to the need to dispose of

379

effluent on-site, ignoring possible synergies with storm-water

380

management systems. We estimate the average non-spatially

381

dependent costs of on-site WMS on the basis of a selection of

5

The choice of these distance weighting factors is arbitrary and based only on visual quality inspection. We consider this a valid approximation given the intention and scope of this paper and the low sensitivity of this parameter. 14

ACCEPTED MANUSCRIPT 382

international cost literature considering the costs of materials,

383

planning and installation, sludge treatment and electricity

384

(Fletcher et al. 2007, WERF 2010, JECES 2015). The assessed

385

treatment systems are either of class C according to DIBt (2014),

386

or where the provision of nitrification or denitrification was not

387

specified we classify the systems with a range as class C-D.

388

Figure 6 shows the total cost function, including the costs of a drip

389

disposal system and for a UV disinfection unit (WERF 2010)

390

which fall in line with other cost estimations for Switzerland (cf.

391

Abegglen 2008).

Decentralised transportation: To derive transportation-related

M AN U

392

SC

RI PT

6

393

costs for on-site WMS, we use model-based cost data from

394

Eggimann et al. 2016, who provide a cost-density relationship

395

(treatment plants per km ) at regional level for a Swiss case

396

study. The authors model a cost-density relationship in a two-

397

dimensional geometrical space by means of a heuristic routing

398

algorithm. To estimate the transportation costs in relation to the

399

treatment plant density on a local scale, we derive the relationship

400

between the CR and the on-site WMS density over all CRs for the

401

whole case-study region (Figure 7).

TE D

EP

2.2.4

Sensitivity analysis

AC C

402

2

403

To evaluate the model sensitivity, we calculate three different cost

404

functions for each key cost element by systematically varying the

405

underlying assumptions (Table 1). A systematic combination of all

406

resulting cost functions yields 81 different scenarios. With this

6

Nutrient recovery is especially promising for on-site WMS and affects the overall economic performance. However, we do not include this analysis in view of the scope of this paper. 15

ACCEPTED MANUSCRIPT approach, we aim to produce diversity in order to indicate sensitivity

408

rather than statistical representativeness.

409

Cost element

Description

Unit

Scenario assumptions

410

 C

Assumed idle capacity

%

0, 50, 100

411 412 413 414 415

 C 

Different minimum sewer slope (fminslope) for running the sewer network generation algorithm

%

0.5, 1, 1.5

416 417

 C

Assumed on-site WMS dimension

PE

5, 10, 15

418

 C 

Systematic cost variation %

RI PT

407

SC

-20, 0, +20

Table 1: Overview of cost scenario assumptions for all key cost

421

elements. Standard scenario values are given in bold.

422

3

423

3.1

Results

M AN U

419 420

Cost curve configurations

Appendix A gives all standard parameter calculations of the

425

former communities of Glarus. Figure 8 shows the results over all

426

cost scenarios (see Section 2.2.4) with respect to the different OCR.

427

We find very diverse OCR at the former communal level with broad

428

sensitivity ranges resulting from the cost scenarios. For today’s more

429

dispersed southern community ‘Süd’, we find lower OCRs than for

430

the more urbanised communities of ‘Nord’ or ‘Glarus’.

AC C

EP

TE D

424

431

The detailed cost curve configuration of the case study catchment

432

in Glarus is given in Figure 9. We notice that the OCRregion and the

433

OCRequalcost are at very low CR of around 0.2 and 0.4 respectively.

434

However, the total regional cost curve is more or less horizontal until

435

a CR of around 0.6 - 0.7. The standard parameter calculation is thus

436

very sensitive to small changes of any single cost component.

16

ACCEPTED MANUSCRIPT 437

3.2

438

Typology We can identify basic cost-shape behaviours on the basis of the

configurations of all the communities. This enables us to build a

440

typology that distinguishes between three major configurations (see

441

Figure 10 for typical examples):

•

442

RI PT

439

Type A: This cost curve configuration type has no

OCRequalcost, and centralised WMS costs are typically lower

444

for all CR. The OCRregion is typically very high. •

445

SC

443

Type B: For this type, we do not find a distinct OCRequalcost because the intersection point is highly sensitive to cost-

447

curve changes due to a more or less horizontal total regional

448

cost curve (we may find multiple cost curve intersections).

449

The OCRregion is typically in the middle CR range. On-site

450

WMS costs only become noticeably expensive at very high

451

CR.

Type C: For this type, we find distinctive OCRequalcost and

TE D

•

452

M AN U

446

OCRregion with clear cost differences. Typically, we observe a

454

distinctive exponential increase of the centralised costs at

455

relatively low CR, leading to low OCRs.

4

Discussion

AC C

456

EP

453

457

We now reflect critically on the case study application and our

458

framework in general. We then elaborate the institutional conditions

459

under which the different OCRs could be realised. Finally, we identify

460

potential research needs.

461

4.1

462 463

Case study application and OCR typology The case study application confirms that we can indeed identify

the conceptually outlined OCRs on a real example. We find very 17

ACCEPTED MANUSCRIPT different OCRregion and OCRequalcost for the former and merged

465

communities depending on the local geography, ranging from very

466

low to very high CR. For example, we note that the WWTP

467

catchment along the new community ‘Süd’ is unsuitable for a large

468

centralised WMS. We see that the OCR depends on the chosen

469

scale and catchment boundary, which is to be expected, as the

470

topographic characteristics also depend on the chosen system

471

boundaries. Specific characteristics of the various catchments

472

enabled us to derive a typology of cost curves. However, the

473

boundaries of this typology are fuzzy, so it only represents a broad

474

classification.

SC

M AN U

475

RI PT

464

An important finding for our case study catchment is that both the calculated OCRs are lower than the CRpresent. We argue that this is

477

because these sewers were not constructed primarily from a cost

478

optimisation point of view and the regulators have often introduced a

479

mandatory connection rule in order to force a higher number of

480

households to connect to the sewers than a direct cost comparison

481

would suggest. The reason for these regulations often lie in the

482

argument that centralised systems are easier to control than a myriad

483

of on-site WMS, or that the latter cannot cope in terms of treatment

484

performance criteria (inter alia Moelants et al. 2008, Buchanan et al.

AC C

EP

TE D

476

485

2014). However, for this paper we assume that neither of these

486

arguments will be valid if on-site WMS are properly designed and if

487

appropriate business models are installed to run them. We thus

488

presume that the role of institutions responsible for applying the

489

mandatory connection rule or investment subsidies for centralised

490

WMS explain why CRpresent is much higher than OCRequalcost or

491

OCRregion in countries like Switzerland (cf. Eggimann et al. 2015).

18

ACCEPTED MANUSCRIPT 492

High uncertainties result from the various cost scenarios. This is largely due to the cost curve configurations: for many communities

494

(including the case study catchment estimated here), the total

495

regional cost curve is rather flat and the cost values of OCRregion and

496

OCRequalcost are thus very close. Consequently, only minor cost

497

differences would lead to very different intersection points on the cost

498

curve. This suggests that a focus on OCRequalcost might be a viable

499

option if OCRregion is hard to implement.

RI PT

493

Finally, territorial reforms are a challenge in water governance

501

(OECD 2015b). The organisational centralisation in Glarus is in line

502

with the general tendency to centralisation throughout Switzerland.

503

We argue that this creates an opportunity to reach lower-cost CR

504

because larger organisations are likely to develop higher professional

505

competencies to run both centralised and on-site WMS (Maurer et al.

506

2012b). This is especially interesting in the case of on-site treatment,

507

where larger contracts result in more standardised and professional

508

operation and management.

509

4.2

M AN U

TE D

In order to decide which of the two candidate OCRs is more likely to be implemented, we have to take a closer look at the incentive

AC C

511

The institutional and organisational setting of OCR

EP

510

SC

500

512

513

structures and regulatory arrangements in the specific regions:

i.)

To reach the OCRregion, a central decision maker would

514

have to determine the total regional cost curve and

515

identify the lowest cost point. In most empirical cases,

516

this will coincide with the lowest cost point of the total

517

centralised cost curve. It would therefore be sufficient to

518

require central operators to connect new households as

19

ACCEPTED MANUSCRIPT 7

long as their total average cost curve decreases.

520

Beyond that point, households would have to seek

521

services from companies that offer on-site alternatives.

522

As a consequence, users connected to the centralised

523

WMS would have lower costs than those serviced by on-

524

site WMS. The rationale for this arrangement is that the

525

total amount of money spent in the region would then be

526

lowest. However, this solution would imply that users

527

connected to the centralised WMS would pay

528

substantially lower tariffs for their WMS services than

529

those that have to rely on on-site solutions. Such price

530

differences could lead to political protests. This problem

531

could be circumvented if the provision of WMS services

532

for the whole region was delegated to a monopoly

533

provider who would be obliged to charge households

534

equal tariffs while minimizing the overall costs in the

535

region. One way to implement such a solution would be

536

for a public utility to build up equal professional

537

competence in both centralised and decentralised WMS

538

and be subjected to tight price regulation. Alternatively,

SC

RI PT

519

EP

TE D

M AN U

8

the OCRregion could be reached by a private monopoly

539

operator who bids for a long-term service contract

AC C

540 541

through a public call for tenders (Demsetz 1968).

542

7

Following this logic, the central operator may not maximise his profit and consequently needs to be regulated, as he would otherwise connect too many households (the profit maximum lies somewhere between both OCR).

8

In order not to complicate matters, we assume that utilities would be able to charge tariffs on a cost-plus basis. 20

ACCEPTED MANUSCRIPT 543

ii.)

However, it may not be feasible to reach the OCRregion under specific conditions: there may be strong political

545

preferences in the region for individual households to

546

choose their service provider freely, and monopoly

547

providers (public or private) may meet with resistance.

548

Moreover, it may prove difficult to build up professional

549

competencies in both centralised and decentralised

550

WMS within a single organisation. In these situations, the

551

OCRequalcost might be a second-best option, as the costs

552

would be the same for all households while various

553

organisations could compete to supply them. The

554

OCRequalcost could be reached if a public or private

555

organisation running the centralised system were

556

required to offer its services at average cost and would

557

be prohibited from turning down customers. Households

558

would be free to choose either to connect to a sewer or

559

to accept services from one of the potentially many

560

suppliers of on-site WMS. In this case, the centralised

561

system would expand to the point where the average

562

cost curves of the two systems intersect, i.e. the

SC

M AN U

TE D

EP

OCRequalcost.

AC C

563

RI PT

544

564

We can deduce from the general cost-curve characteristics of our

565

case study that the following relationship holds for countries with very

566

high CRpresent: OCRregion < OCRequalcost < CRpresent. The first inequality

567

is given by the shape of the cost curves and is generic. The second

568

is very likely to hold in countries which have installed regulations

569

such as mandatory connection rules. Otherwise, competition would

570

likely lead to market shares that are close to or at around the

571

OCRequalcost.

21

ACCEPTED MANUSCRIPT We may summarise our framework for calculating the total

573

regional cost for hybrid WMS systems as follows: the shape of the

574

type C cost curve indicates two potential OCR that would be superior

575

to the present CR. However, which of these OCR is actually reached

576

depends on the role specification of households, the central system

577

operator, the on-site suppliers and the regulator. Getting away from

578

current mixing ratios will therefore depend on comprehensive reforms

579

(including organisations and regulations) and cannot be considered

580

purely as a matter of cost.

581

4.3

SC

Critical reflection and research needs

The full cost assessment for regional WMS represents at least a

M AN U

582

RI PT

572

583

first step towards determining more sustainable WMS services.

584

However, it is not enough merely to assess the costs.

585

Most cost assumptions relating to the costs of on-site treatment were chosen on the conservative side in this paper (including for

587

disinfection and on-site infiltration). However, these may be subject

588

to considerable changes in the future, for instance if economies of

589

scale could be reaped in manufacturing (Adler 2007, Dahlgren et al.

590

2013). On the other hand, assumptions about effluent disposal would

591

require a more sophisticated analysis including storm water

AC C

EP

TE D

586

592

evacuation. As far as transportation costs are concerned, lacking

593

economies of scale and the challenges involved in establishing fully

594

functional O&M schemes are usually considered as the key

595

disadvantages of on-site WMS (cf. Eggimann et al. 2016). However,

596

off-grid infrastructure systems also possess specific advantages,

597

although these are hard to express in monetary terms: the

598

independence from a sewer network increases the flexibility to

599

respond to socio-economic or technological boundary conditions

600

(Panebianco and Pahl-Wostl 2006, Hug et al. 2010). It also reduces 22

ACCEPTED MANUSCRIPT interdependence-related disruptions (Rinaldi et al. 2001) and lessens

602

the potential environmental impact in case of failure of a single plant,

603

whether due to malfunctions, earthquakes (Hamada 2014) or

604

terrorism (Panebianco and Pahl-Wostl 2006). Centralised and on-site

605

WMS thus offer unique strengths and weaknesses which are often

606

intangible and difficult express in to monetary terms (cf. Gikas and

607

Tchnobanoglous 2009, Libralato et al. 2012, Larsen et al. 2013,

608

Vousvouras 2013). However, the quantification of non-monetary

609

advantages or disadvantages goes far beyond this study as it would

610

require a research approach of much greater scope (cf. Morera et al.

611

2015, Arora et al. 2015, Naik and Stenstrom 2016).

SC

M AN U

612

RI PT

601

With the aid of the framework presented here, we can address the question of the degree to which on-site WMS can be considered as

614

substitutes from an economical point of view. However, we refer to

615

the literature (inter alia Larsen et al. 2013, Libralato et al. 2012)

616

concerning the key assumption as to whether on-site WMS can be

617

considered as functionally equivalent from a technological point of

618

view.

In this study, we assume stable context conditions even though

EP

619

TE D

613

many exogenous factors affect infrastructure planning, such as

621

changing public goals or environmental concerns (Hansman et al.

622

2006). Furthermore, WMS are exposed to different long-term

623

dynamics (e.g. population, role of industry, water consumption

624

trends), making it very challenging to plan optimal systems

625

(Dominguez and Gujer 2006). However, this study provides valuable

626

insights into changing population and settlement dynamics related to

627

sanitation costs: we showed that different catchments result in

628

diverse characteristic cost configurations, which gives an indication

629

of what cost configurations may look like and evolve for future

AC C

620

23

ACCEPTED MANUSCRIPT projected catchments. For instance, let us assume an anticipated

631

increase in settlement area together with sprawling tendencies of the

632

settlement distribution for an urban catchment classified as type A.

633

For such a case, we might expect a cost configuration shift from type

634

A towards type C. On the other hand, for catchments classified as

635

type A, urban infill or settlement shrinking in rural areas (Siedentop

636

and Fina 2010) shifts the cost curve from type A towards type C. A

637

final assumption of the framework outlined here is that the basic

638

choice for households in a particular region is either between a fully

639

centralised or a small-scale on-site WMS. We believe that in reality

640

the choice is indeed often limited to these basic two options, namely

641

either to connect to a large centralised WWTP or to select small-

642

scale package treatment plants.

643

5

SC

M AN U

Conclusion

In this paper, we have prepared the way towards achieving a

TE D

644

RI PT

630

cost-based identification of lowest cost connection rates (CR) in a

646

given region. We thus contribute to the broader debate about

647

sustainable CR with a cost analysis over the whole CR continuum. In

648

particular, we identify a potentially optimal CR (OCRregion) from a total

649

regional cost point of view, and a second-best CR (OCRequalcost)

650

which may be easier to implement under specific institutional and

651

organisational conditions.

AC C

EP

645

652

The framework presented here suggests that the OCRregion may

653

be achieved if the operator of the centralised WMS is required to

654

expand his service as long as his average costs decrease.

655

Alternatively, the OCRregion is reached if a single operator runs both

656

WMS alternatives and tariffs are regulated either explicitly or through

657

a call for tenders relating to service contracts. The second-best

24

ACCEPTED MANUSCRIPT OCRequalcost could be reached if individual households can choose

659

freely between central and on-site WMS. A potentially intermediary

660

form would be for households beyond the OCRregion to be charged

661

tariffs proportional to the costs for a sewer connection on the basis of

662

actual household connection costs, thus increasing their incentive to

663

choose on-site WMS. In our case study, we find relatively small cost

664

differences between the two OCR, which suggests that opting

665

directly for the OCRequalcost is an advisable option. We argue that

666

neither OCR can be reached without regulating the centralised WMS

667

and introducing adequate policy measures.

SC

RI PT

658

We optimise CR by building on long-term average costs, thus

669

assuming that the context conditions remain static in the long run. In

670

further elaborations of the framework, it would make sense to include

671

dynamic considerations (e.g. changing settlement patterns or

672

population dynamics).

TE D

M AN U

668

Finally, we believe that a holistic consideration is needed in view

674

of the complexity of the question of cost-efficient CR for sustainable

675

urban water management. We conclude that this discussion cannot

676

be separated from analyses of the respective organisational,

677

institutional and regulatory arrangements in a region and argue for a

678

co-evolution of technological advances in on-site WMS with the

679

prevailing institutional and organisational arrangements.

680

Acknowledgements

681

Our special thanks go to Ivana Logar, Roy Brouwer and Matteo

682

Mattmann for their suggestions to improve the manuscript. We thank

683

Rachel Barrett for her help with the data preparation and Richard

684

Michell for the proofreading. We also thank the Canton of Glarus for

685

their data provision.

AC C

EP

673

25

ACCEPTED MANUSCRIPT Appendix A

687

References

688 689 690

ABD EL GAWAD, H., BUTTER, J., 1995. Clustering of towns and villages for centralized wastewater treatment. Water Science and Technology, 32(11), 85–95.

691 692 693

ABEGGLEN, C.K., 2008. Membrane bioreactor technology for decentralized wastewater treatment and reuse. Dissertation, ETH Zürich, Switzerland.

694 695 696

ADAMS, B.J., DAJANI, J.S., GEMMELL, R.S., 1972. On the centralization of wastewater treatment facilities. Water Resources Bulletin 8(4), 669–678.

697 698 699

ADLER, C., 2007. Market potential of a membrane based wastewater treatment plant for decentralised application in China. Master Thesis. Eawag, Dübendorf, Switzerland.

700 701 702

ARORA, M., MALANO, H., DAVIDSON, B., NELSON, R., GEORGE, B., 2015. Interactions between centralized and decentralized water systems in urban context: A review. WIREs Water 2015(2), 623–634.

703 704 705

ARTHUR, W.B., 1989. Competing technologies, increasing returns, and lock-in by historical events. The Economic Journal, 99(394) 116– 131.

706 707 708 709

BACH, P.M., STAALESEN, S., MCCARTHY, D.T., DELETIC, A., 2015. Revisiting land use classification and spatial aggregation for modelling integrated urban water systems. Landscape and Urban Planning, 143, 43–55.

710 711 712 713

BEHLING, R., BOCHOW, M., FOERSTER, S., ROESSNER, S., KAUFMANN, H., 2015. Automated GIS-based derivation of urban ecological indicators using hyperspectral remote sensing and height information. Ecological Indicators, 48, 218–234.

714 715 716

BUCHANAN, J.R., 2014. Decentralized wastewater treatment. Book chapter, volume three, 'Remediation of Polluted Water', in the Series 'Comprehensive Water Quality and Purification. Elsevier, Oxford, UK.

717 718

CRUNDWELL, F.K., 2008. Finance for engineers. Springer, London, UK.

719 720

DAHLGREN, E., GÖÇMEN, C., LACKNER, K., VAN RYZIN, G., 2013. Small modular infrastructure. The Engineering Economist, 58(4), 231–264.

721 722

DAIGGER, G.T., 2007. Wastewater Management in the 21st Century, Journal of environmental engineering, 133(7), 671–680.

723 724

DEMSETZ, H., 1968. Why Regulate Utilities? Journal of Law and Economics, 11(1), 55–65.

725 726

DIBt, 2014. Approval guidelines for small wastewater treatment plants (Zulassungsgrundsätze Kleinkläranlagen). Berlin, Germany.

AC C

EP

TE D

M AN U

SC

RI PT

686

26

ACCEPTED MANUSCRIPT DOMINGUEZ, D., GUJER, W., 2006. Evolution of a wastewater treatment plant challenges traditional design concepts. Water Research, 40(7), 1389–1396.

730 731

DOMINGUEZ, D., 2008. Handling future uncertainty: Strategic planning for the infrastructure sector. Dissertation, ETH Zürich, Switzerland.

732 733

DOWNING, P.B., 1969. The economics of urban sewage disposal. Praeger, New York, NY, USA.

734 735 736 737 738

EGGIMANN, S., 2013. Potential for a sustainability transition of decentralised wastewater systems for Switzerland on the level of urban structural units (Potenzial für eine Nachhaltigkeits-Transition dezentraler Abwassersysteme auf Strukturtypebene in der Schweiz). Master Thesis, University of Zürich, Switzerland (in German).

739 740 741

EGGIMANN, S., TRUFFER, B., MAURER, M., 2015. To connect or not to connect? Modelling the optimal degree of centralisation for wastewater infrastructures. Water Research, 84, 218–231.

742 743

EGGIMANN, S., TRUFFER, B., MAURER, M., 2016. Economies of density for on-site treatment. Water Research, 101, 476–489.

744 745 746 747 748

ETNIER, C., NELSON, V., PINKHAM, R., 2000. Economics of decentralised wastewater treatment systems: Direct and indirect costs and benefits. In: National research needs conference proceedings: Risk-based decision making for onsite wastewater treatment, St. Louis, MO, USA.

749 750 751

FANE, A.G., FANE, S.A., 2005. The role of membrane technology in sustainable decentralized wastewater systems. Water Science and Technology, 51(10), 317–325.

752 753

FLETCHER, H., MACKLEY, T., JUDD, S., 2007. The cost of a package plant membrane bioreactor. Water Research, 41(12), 2627–2635.

754 755 756

FUENFSCHILLING, L., TRUFFER, B., 2014. The structuration of sociotechnical regimes—Conceptual foundations from institutional theory. Research Policy, 43(4), 772–791.

757 758 759

GIKAS, P., TCHOBANOGLOUS, G., 2009. The role of satellite and decentralized strategies in water resources management. Journal of Environmental Management, 90(1), 144–152.

AC C

EP

TE D

M AN U

SC

RI PT

727 728 729

760 761 762

GUO, T., ENGLEHARDT, J.D., 2015. Principles for scaling of distributed direct potable water reuse systems: A modeling study. Water Research, 75, 146–163.

763 764 765

HAGGAG, M.A, AYAD, H.M., 2002. The urban structural units method: a basis for evaluating environmental prospects for sustainable development. Urban Design International, 7(2), 97–108.

766 767

HAMADA, M. (EDS), 2014. Critical urban infrastructure handbook. CRC Press, Boca Raton, FL, USA.

768 769 770 771 772

HAMILTON, B.A., PINKHAM, R.D., HURLEY, E., W ATKINS, K., LOVINS, A.B., MAGLIARO, J., ETNIER, C., NELSON, V., 2004. Valuing decentralised wastewater technologies: A catalog of benefits costs and economic analysis techniques. Rocky Mountain Institute, Snowmass, CO, USA. 27

ACCEPTED MANUSCRIPT HANSMAN, R.J., MAGEE, C., NEUFVILLE, R. DE., ROBINS, R., ROOS, D., 2006. Research agenda for an integrated approach to infrastructure planning, design and management. International Journal of Critical Infrastructures, 2(2/3), 146–159.

777 778 779 780

HUG, T., DOMINGUEZ, D., MAURER, M., 2010. The cost of uncertainty and the value of flexibility in water and wastewater infrastructure planning. Proceedings of the Water Environment Federation, Cities of the Future/Urban River Restoration, 487–500.

781 782 783

JECES, 2015. Personal communication with Satoru Takahashi, Japan Education Center of Environmental Sanitation, 22 October 2015.

784 785 786 787

KAUNDINYA, D.P., BALACHANDRA, P., RAVINDRANATH, N.H., 2009. Gridconnected versus stand-alone energy systems for decentralized power—A review of literature. Renewable and Sustainable Energy Reviews, 13(8), 2041–2050.

788 789 790 791

KIPARSKY, M., SEDLAK, D.L., THOMPSON, B.H., TRUFFER, B., 2013. The innovation deficit in urban water: The need for an integrated perspective on institutions, organisations, and technology. Environmental Engineering Science, 30(8), 395–408.

792 793 794 795 796

KNOPS, H., 2008. The impact of technical characteristics of network industries upon the governance of infrastructure adequacy. In: Arts, G., Dicke, W., Hancher, L. (eds), New perspectives on investment in infrastructures. Amsterdam University Press, Amsterdam, The Netherlands.

797 798 799

LARSEN, T.A., UDERT, K.M., LIENERT, J. (EDS), 2013. Source separation and decentralization for wastewater management. IWA Publishing, London, UK.

800 801 802 803

LEE, E.J., CRIDDLE, C.S., BOBEL, P., FREYBERG, D.L., 2013. Assessing the scale of resource recovery for centralized and satellite wastewater treatment. Environmental Science and Technology, 47(19), 10762–10770.

804 805 806 807

LIBRALATO, G., GHIRARDINI, A.V., AVEZZÙ, F., 2012. To centralise or to decentralise: An overview of the most recent trends in wastewater treatment management. Journal of Environmental Management, 94(1), 61–68.

808 809 810 811

LIEBERHERR, E., TRUFFER, B., 2015. The impact of privatization on sustainability transitions: A comparative analysis of dynamic capabilities in three water utilities. Environmental Innovation and Societal Transitions, 15, 101–122.

812 813 814 815

LIEBERHERR, E., FUENFSCHILLING, L., 2016. Neoliberalism and sustainable urban water sectors: A critical reflection of sector characteristics and empirical evidence. Environment and Planning C: Government and Policy, 1-16.

816 817

LÜSCHER, P., W EIBEL, R., BURGHARDT, D., 2009. Integrating ontological modelling and Bayesian inference for pattern

AC C

EP

TE D

M AN U

SC

RI PT

773 774 775 776

28

ACCEPTED MANUSCRIPT classification in topographic vector data. Computers, Environment and Urban Systems, 33, 363–374.

820 821 822

LWIN, K., MURAYAMA, Y., 2009. A GIS approach to estimation of building population for micro-spatial analysis. Transactions in GIS, 13(4), 401–414.

823 824 825

MARLOW, D.R., MOGLIA, M., COOK, S., BEALE, D.J., 2013. Towards sustainable urban water management: A critical reassessment. Water Research, 47(20), 7150–61.

826 827 828 829

MAURER, M., ROTHENBERGER, D., LARSEN, T.A., 2006. Decentralised wastewater treatment technologies from a national perspective: at what cost are they competitive? Water Science and Technology: Water Supply, 15(6), 145–154.

830 831 832

MAURER, M., W OLFRAM, M., HERLYN, A., 2010. Factors affecting economies of scale in combined sewer systems. Water Science & Technology, 62(1), 36–41.

833 834 835 836

MAURER, M., BUFARDI, A., TILLEY, E., ZURBRÜGG, C., TRUFFER, B., 2012. A compatibility-based procedure designed to generate potential sanitation system alternatives. Journal of Environmental Management, 104, 51–61.

837 838 839

MAURER, M., CHAWLA, F., HORN, J., STAUFER, P., 2012B. Water Sanitation 2025 (Abwasserentsorgung 2025 in der Schweiz). Schriftreihe der eawag, Dübendorf, Switzerland (in German).

840 841 842 843 844

MEINEL, G., BURGDORF, M., 2008. Automatic derivation of urban land use pattern and integration in a Geographical Information System. Forschungen, Bundesministerium für Verkehr, Bau und Stadtentwicklung, Bundesamt für Bauwesen und Raumordnung, Bonn, Germany (in German).

845 846 847

MOELANTS, N., JANSSEN, G., SMETS, I., VAN IMPE, J., 2008. Field performance assessment of onsite individual wastewater treatment systems. Water Science and Technology, 58(1), 1–6.

848 849 850 851

MORERA, S., COMAS, J., POCH, M., COROMINAS, L., 2015. Connection of neighboring wastewater treatment plants: economic and environmental assessment. Journal of Cleaner Production, 90, 34– 42.

AC C

EP

TE D

M AN U

SC

RI PT

818 819

852 853 854

NAIK, K.S., STENSTROM, M.K., 2016. A feasibility analysis methodology for decentralized wastewater systems – Energyefficiency and cost. Water environment research, 88(3), 201-209.

855 856 857

PANEBIANCO, S., PAHL-W OSTL, C., 2006. Modelling socio-technical transformations in wastewater treatment—A methodological proposal. Technovation, 26(9), 1090–1100.

858 859 860 861

POUSTIE, M.S., DELETIC, A., BROWN, R.R., W ONG, T., DE HAAN, J., SKINNER, R., 2014. Sustainable urban water futures in developing countries: the centralised, decentralised or hybrid dilemma. Urban Water Journal, 1–16.

862 863 864

RINALDI, S.M., PEERENBOOM, J.P., KELLY, T.K., 2001. Identifying, understanding, and analyzing critical infrastructure interdependencies. IEEE Control Systems Magazine, 21(6), 11–25. 29

ACCEPTED MANUSCRIPT SCHILLER, G., 2010. Cost evaluation of the adaptation of wastewater treatment systems under shrinkage (Kostenbewertung der Anpassung zentraler Abwasserentsorgungssysteme bei Bevölkerungsrückgang). IÖR Schriften, 51, Rhombos (in German).

869 870 871

SEDLAK, D., 2014. Water 4.0: The past, present, and future of the world's most vital resource. Yale University Press, New Haven & London.

872 873 874 875

SHAHABI, M.P., MCHUGH, A., ANDA, M., HO, G., 2015. Comparative economic and environmental assessments of centralised and decentralised seawater desalination options. Desalination, 376, 25– 34.

876 877 878

SIEDENTOP, S., FINA, S., 2010. Urban sprawl beyond growth: the effect of demographic change on infrastructure costs. Flux (n° 79-80), 90– 100.

879 880 881

SINGH, N.K., KAZMI, A.A., STARKL, M., 2015. A review on full-scale decentralized wastewater treatment systems: techno-economical approach. Water Science & Technology, 71(4), 468–478.

882 883 884

SPIRANDELLI, D., 2015. Patterns of wastewater infrastructure along a gradient of coastal urbanization: A study of the Puget Sound Region. Land, 4(4), 1090–1109.

885 886 887

STEINIGER, S., BURGHARDT, D., W EIBEL, R., 2008. An approach for the classification of urban building structures based on discriminant analysis techniques, Transactions in GIS, 12(1), 31–59.

888 889 890 891

TCHOBANOGLOUS, G., LEVERENZ, H., 2013. The rationale for decentralization of wastewater infrastructure. In: Larsen, T. A., Udert, K. M., Lienert, J. (eds), Source separation and decentralization for wastewater management, IWA Publishing, London, UK.

892 893 894

TOWNEND, C.B., 1959. The economics of the disposal of sewage and trade effluents. Bulletin of the World Health Organization, 20, 535– 562.

895 896 897 898

TRUFFER, B., STÖRMER, E., MAURER, M., RUEF, A., 2010. Local strategic planning processes and sustainability transitions in infrastructure sectors. Environmental Policy and Governance, 20, 258–269.

AC C

EP

TE D

M AN U

SC

RI PT

865 866 867 868

899 900 901

UN, 2015, Environmental Indicators. Inland Water Resources, Wastewater. Archived by WebCite® http://www.webcitation.org/6gQ6z23bk

902 903

U.S. CENSUS BUREAU, 2015. Construction Price Indexes. Archived by WebCite® http://www.webcitation.org/6d6sN7StR

904 905

OECD, 2010. Pricing water resources and water and sanitation services. OECD Publishing, London, UK.

906 907

OECD, 2015. Environment at a Glance 2015: OECD Indicators. OECD Publishing, Paris, France.

908 909

OECD, 2015B. Water and cities. Ensuring sustainable futures. OECD Studies on Water. OECD Publishing, Paris, France.

30

ACCEPTED MANUSCRIPT O'FLAHERTY, B., 2005. City economics. Harvard University Press, USA.

912 913 914

OSMOND, P., 2010. The urban structural unit: towards a descriptive framework to support urban analysis and planning. Urban Morphology, 14(1), 5-20.

915 916 917 918

VOUSVOURAS, C.A., 2013. Large-scale, small-scale, and hybrid water utilities in cities of the developing world: The impact of scale beyond “economies of scale”. Dissertation, University of St. Gallen, Switzerland.

919 920 921 922

VSA, 2011. Costs and performances of sewage disposal (Kosten und Leistungen der Abwasserentsorgung). Verband Schweizer Abwasser- und Gewässerschutzfachleute, Glattbrugg, Switzerland (in German).

923 924 925 926

W ANG, H.F., QIU, J.X., BREUSTE, J., FRIEDMAN, R.C., ZHOU, W.Q., W ANG, X.K., 2013. Variations of urban greenness across urban structural units in Beijing, China. Urban Forestry and Urban Greening, 12(4), 554–561.

927 928

WEF, 2010. Global Risks 2010: A Global Risk Network Report. A Global Risk Network Report. Geneva, Switzerland.

929 930

WERF, 2010. Performance and cost of decentralised unit processes. Archived by WebCite® http://www.webcitation.org/6d6qEmdU6.

931 932 933 934

W ICKOP, E., 1998. Environmental quality targets for urban structural units in Leipzig with a view to sustainable urban development. In: Breuste, J., Feldmann, H., Uhlmann, O. (eds), Urban Ecology. Springer, Berlin, Germany.

935 936

W ORLD BANK, 2015. PPP conversion factor. International comparison program database. Available at: http://data.worldbank.org/

937 938 939

ZVOLEFF, A., KOCAMAN, A.S., HUH, W.T., MODI, V., 2009. The impact of geography on energy infrastructure costs. Energy Policy, 37(10), 4066–4078.

AC C

EP

TE D

M AN U

SC

RI PT

910 911

31

b.) Average per capita decentralised treatment costs

ACCEPTED MANUSCRIPT

80 40 60 20 Connection rate [%]

0

100

SC 100

[in $ per PE and year]

TE D 80 40 60 20 Connection rate [%]

80 40 60 20 Connection rate [%]

f.) Total average per capita decentral WMS costs

EP

0

AC C

[in $ per PE and year]

cen

cen

[in $ per PE and year]

dec

Ctransport

0

100

dec dec dec Ctot ( Ctransport + Ctreatment )

80 40 60 20 Connection rate [%]

e.) Total average per capita central WMS costs

cen

Sensitivity indication Illustrative cost function

M AN U

0

Ctot ( Ctransport + Ctreatment )

100

d.) Average per capita decentral transportation costs

Sensitivity indication Illustrative cost function

[in $ per PE and year]

cen

Ctransport

c.) Average per capitacentral transportation costs

80 40 60 20 Connection rate [%]

RI PT

dec

Ctreatment

0

Sensitivity indication Illustrative cost function

[in $ per PE and year]

Sensitivity indication Illstrative cost function

[in $ per PE and year]

cen

Ctreatment

a.) Average per capita centralised treatment costs

0

100

80 40 60 20 Connection rate [%]

100

g.) Total average per capita regional costs

[in $ per PE and year]

Ctot

Total average regional costs

0

80 40 60 20 Connection rate [%]

100

Figure 1: Idealised average per capita cost functions over all CR. The final cost curve (configuration g) corresponds to the cost type C in Figure 10. See also Section 2.2.3 for the underlying material and methods.

60 40 80 Connection rate [%]

100

M AN U

5 95 Ø CRpresent

Decentralised Centralised Regional Population serviced [in %]

50 50 Ø

AC C

EP

Total cost

OCRequal

TE D

OCRregion

14 86 Ø

SC

20

46 54 Ø

Average annuities [in $ per PE]

0

RI PT

Average annuities [in $ per PE]

ACCEPTED MANUSCRIPT

50

Figure 2: Calculated OCR, costs and distribution of central and on-site WMS from an exemplary cost curve configuration. A CRpresent of 95% is shown. With help of the bar charts, the costs of the three total cost curves are visualised for all three CR. The numbers on the bar charts show the percentages of the population serviced with centralised or decentralised systems respectively for each CR.

ACCEPTED MANUSCRIPT

DE

AT

Bilten

RI PT

Nie derurnen O b erur n e n

Näfels

SC al

Nest

M AN U

Riedern Glarus

Glarus [today, after merge]

Luch sin ge n

Braunwald

Case-study catchement Data Sources: Federal office of Topography and Canton of Glarus

Current connection rate [in %]

Elm

EP

AC C

WWTP catchement border

Matt

Rüti

Betschwanden

Former community border

e

n

Mitlödi

Mühlehorn

Sool Engi

n de

Süd

Schwändi

sl Ha

Glarus

Sc hw an

TE D

Nord

Mollis

a nd ne En

IT

Filzbach

CH

Obstalden

FR

Linthal

5 km

¢

< 91.00 91.01 - 94.00 94.01 - 97.00 97.01 - 100.00

Figure 3: Case study area showing the present and former community boundaries as well as different WWTP catchments. White areas belong to external catchment areas or are not part of the Canton of ‘Glarus’.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Settlement area

Post-processing area

Figure 4: Schematic representation of USU generation.

Building

Street network

USU centroid

Projection

y = 410.05x -0.25 y = 410.05x 2

ACCEPTED MANUSCRIPT

R² = 0.51

SC

RI PT

300 300

M AN U

200 200

EP

TE D

100 100

00

-0.25

R = 0.519

0

AC C

Annuities

[$ per capita per year]

400 400

20,000

40,000

60,000

80,000

100,000

Population equivalent

Figure 5: Swiss capital and O&M expenditures for centralised treatment (VSA 2011), assuming an average lifespan of 30 years and a discount rate of 2%.

700 Total costs incl. UV Total costs excl. UV Primary & secondary treatment Drip disposal UV disinfection

RI PT

600

SC

500

M AN U

400

TE D

300

EP

200 100 0

AC C

Annuities for decentralised WMS excl. transportation related costs [$ per PE and year]

ACCEPTED MANUSCRIPT

0

10

20

30

40

Population equivalent

Figure 6: Average treatment cost data for on-site WMS.

50

RI PT

4

M AN U

SC

3 2

0

20

EP

0

TE D

1

AC C

[treatment plants per km2]

On-site WMS density

ACCEPTED MANUSCRIPT

5

40

60

Connection rate [%]

80

100

Figure 7: Calculated relationship between the CR and the treatment plant density for the whole case study region as a basis for estimating local transportation-related costs.

ACCEPTED MANUSCRIPT

100 Oberurnen Lindthal

RI PT

80

SC

Mühlehorn

OCRregion

M AN U

60

Sool

TE D

40

Betschwanden Mollis Glarus

Haslen

Schwändi

Rüti Luchsingen Filzbach

Matt

Engi

Näfels Bilten

Nord

Elm Obstalden

case study catchment

EP

Süd

AC C

20

Braunwald

0

0

20

40

60

80

100

OCRequalcost Figure 8: Visualisation of the cost scenario calculations for all former (black) and current merged communities of the case study region ‘Glarus’ (coloured), including the case study catchment. The scenario sensitivity is indicated by error bars representing one standard deviation.

800

M AN U

SC

600

TE D

400

EP

200 0

RI PT

ACCEPTED MANUSCRIPT

Total cost Decentralised Centralised Regional OCRregion

0

AC C

Annuities

[$ per capita per year]

1,000

20

40

60

Connection rate [%]

80

100

Figure 9: Thick lines show standard parameter calculations, shaded areas indicate scenario uncertainties (maximum extent over all 81 scenarios).

600 400 200

RI PT

Annuities [in $ per PE]

Example type C

800

0 0

SC

typ eC

ACCEPTED MANUSCRIPT 1,000

Annuities [in $ per PE]

M AN U

[in $ per PE]

60 40 80 Connection rate [%]

Total cost decentralised centralised regional

OCRregion OCRequalcost

pe

400 200 0

1,000 A

Example type B

600

100

ty

100

800

0

Annuities [in $ per PE]

20

EP

0

TE D

B type

AC C

Annuities

1,000

20 40 60 80 Connection rate [%]

20 40 60 80 Connection rate [%]

100

Example type A

800 600 400 200 0

0

20 40 60 80 Connection rate [%]

100

Fig. 10: Cost curve configuration typology with examples. For each example type, standard parameter runs (thick line) and scenario uncertainties are provided (cf. Figure 9).

ACCEPTED MANUSCRIPT Highlights •

We perform a total-cost analysis of hybrid wastewater management systems

•

We find a socially optimal connection rate and a second-best

•

RI PT

connection rate Optimal connection rates rely on organisational and

AC C

EP

TE D

M AN U

SC

institutional arrangements

1,000

1,000

800

800

800

600

600

600

400

400

400

200

200

200

1,000

Näfels

600

400

400

200

200 0

20

40

60

80

0

100

1,000

Obstalden

800

800

600

600

400

400

200

200 0

20

1,000

40

60

80

100

800

40

0

60

80

Mollis

100

200 0

1,000

0.2

0.4

0.6

0.8

1.0

Haslen

0 0 1,000

0

20

40

60

80

100

Mühlehorn

0

20

40

60

80

100

0 0 1,000

80

100

0 0

600

600

400

400

200

200 0

1,000

20

40

60

80

100

80

100

0 0 1,000

Schwanden

400

400

400

200

200

200 0

20 40 60 80 Connection rate [%]

100

20 40 60 80 Connection rate [%]

100

60

Nestal

20

1,000

Ennenda

40

200

800

0

100

20

400

600

1.0

80

Filzbach

600

600

0.2 0.4 0.6 0.8 Connection rate [%]

100

800

600

0

80

200

800

0

60

400

800

800

40

600

800

0

20

800

EP

400

0

20

1,000

Glarus

600

0

Oberurnen

RI PT

0

100

SC

80

600

0

Costs [in $ per PE]

60

800

1,000

Costs [in $ per PE]

40

800

0

Costs [in $ per PE]

20

TE D

Costs [in $ per PE]

1,000

0

ACCEPTEDNiederurnen MANUSCRIPT

M AN U

0

Bilten

AC C

Costs [in $ per PE]

1,000

0 0

40

60

Luchsingen

20

40

60

Schwändi

Fig. A.1: Communal cost curve configurations. The results show standard parameter calculations, including scenario uncertainties indicated by shaded areas.

800

600

600

600

400

400

400

200

200

200

60

80

100

Matt

0 1,000

800

800

600

600

400

400

200

200

0 1,000

0

20

40

60

80

100

Braunwald

0 1,000

0

20

40

60

80

Elm

100

0 0 1,000

0

20

40

60

80

100

Linthal

0 0 1,000

400

400

400

200

200

200

1,000

TE D

600

60

Riedern

800 600

80

100

0

20

40

Decentralised

200

Centralised Regional 0

02

40

60

Connection rate [%] Fig A.1: (continued).

80

100

60

80

Connection rate [%]

Total cost

400

0

0

80

100

60

80

100

80

100

Rüti

OCRregion

20

40

Betschwanden

800

600

40

60

200

600

02

40

400

800

0

20

600

800

0

Engi

800

SC

40

M AN U

20

EP

Costs [in $ per PE]

0

RI PT

800

1,000

Costs [in $ per PE]

1,000

ACCEPTED MANUSCRIPT Sool

800

0

Costs [in $ per PE]

1,000

Mitlödi

AC C

Costs [in $ per PE]

1,000

100

0 0

20

40

60

Connection rate [%]