(GHG) Emissions from Rainwater Harvesting Systems

A Dissertation

entitled

Life Cycle Assessment of Rainwater Harvesting Systems at Building and Neighborhood Scales and for Various Climatic Regions of the U.S. by

Jay P. Devkota

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the

Doctor of Philosophy Degree in

Engineering

Dr. Defne Apul, Committee Chair

Dr. Steven Burian, Committee Member

Dr. Ashok Kumar, Committee Member

Dr. Cyndee Gruden, Committee Member

Dr. Youngwoo Seo, Committee Member

Dr. Patricia R. Komuniecki, Dean College of Graduate Studies

The University of Toledo December 2015

This document is copyrighted material. Under copyright law, no parts of this document may be reproduced without the expressed permission of the author. An Abstract of

Life Cycle Assessment of Rainwater Harvesting Systems at Building and Neighborhood Scales and for Various Climatic Regions of the U.S.

Jay P. Devkota

Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Doctor of Philosophy Degree in Civil Engineering

The University of Toledo

December 2015

Rainwater harvesting can be a strategy to address challenges with urban water and wastewater infrastructure such as leakage, underfunding energy usage and combined sewer overflow. Rainwater harvesting system has been used for centuries to meet urban water demands such as toilet flushing, lawn irrigation, cleaning and recreational activities. Of these uses of harvested rainwater, toilet flushing is more common as it constitutes a higher percentage of indoor water use. Life cycle assessment is becoming a powerful tool to estimate environmental sustainability of rainwater harvesting systems. With growing interest in rainwater harvesting systems, it is now essential to understand and estimate the factors affecting its environmental sustainability to better design the system as well as to provide a framework for future researchers. Three research needs were identified and addressed in this study. Knowing that the prior studies lacks generalization of results to other cases, a water demand to supply ratio (D/S) ratio was proposed and demonstrated to estimate environmental impacts from rainwater harvesting systems. A decision framework was also proposed based on the result to help designers and practitioners estimate the environmental impacts without much effort. This study hypothesized that there is a lack of iii consistency in the analysis of rainwater harvesting systems. Supply and demand based approach was compared for rainwater harvesting system at ten climatic regions with one cubic meter of rainwater supplied and sanitation service in the building throughout its life time as respective functional units. Unexpectedly, the result showed that the region with lower environmental impact for one functional unit did not necessarily have lower impacts for the other functional unit making it clear that regional preference for rainwater harvesting system depends on the specific goal to be met: using harvested rainwater as a potential supplemental source versus using it to meet the water demand needs.

Implementation of rainwater harvesting system also depends on the scale of its implementation. A suburban neighborhood was selected to demonstrate the environmental sustainability of different rainwater harvesting implementation strategies (decentralized, centralized and partly centralized). The result showed that the stormwater reduction from rainwater harvesting system mainly depends on the downspout connectivity. A selection framework was also proposed to select the system with lower environmental impact based on downspout connectivity, soil type and the centralized, partly centralized, or decentralized strategy. The results of this study are expected to help designers and policy makers choose environmentally attractive rainwater harvesting scenario by looking into different factors such as building characteristics (roof area, occupancy), method of analysis and scale of implementation.

Acknowledgements

Learning theories and concepts of a computational research to address the issues on water sustainability has been one of the most challenging academic tasks I have ever faced.

This research project would not have been possible without the support of Dr. Defne Apul.

This is a great opportunity to express my respect to Dr. Defne Apul for her guidance, advice and support for the entire three and a half years. I would like to extend my gratitude to

National Science Foundation for the financial support.

I owe my deepest gratitude to the committee members Dr. Steven Burian, Dr Ashok

Kumar, Dr Cyndee Gruden and Dr. Youngwoo Seo for their assistance; without their knowledge on the specific topic; this research would not have been successful.

My special thanks go to all my friends and group members, especially Robert,

Chirjiv, Thelma, Hassan and Prayag for their valuable help and suggestions throughout my

PhD.

Last but not least, I would like to thank the Civil and Environmental Engineering

Department, at The University of Toledo for providing me financial support and quality education for those wonderful years, and Rose Marie Ackerman for always being there to provide resources whenever needed.

This dissertation is dedicated to my family and due to their inspiration and support,

I am here.

Table of Contents

Abstract ...... iii

Acknowledgements ...... v

Table of Contents ...... vi

List of Tables ...... xi

List of Figures ...... xiv

List of Abbreviations ...... xix

List of Symbols ...... xx

1 Introduction ...... 1

2 Introducing Demand to Supply Ratio as a New Metric for Understanding Life

Cycle Greenhouse Gas (GHG) Emissions from Rainwater Harvesting Systems . 13

2.1 Introduction ...... 15

2.2 Methodology ...... 17

2.2.1 Goal and Scope ...... 17

2.2.2 Life cycle inventory, impact assessment, and interpretation ...... 24

2.3 Results and Discussion ...... 26

2.3.1 Characteristics of the medium office buildings that have RWH

systems ...... 26

2.3.2 Effect of demand to supply ratio (D/S) on life cycle greenhouse

gas emissions from buildings with RWH systems ...... 29

2.3.3 Comparison of buildings with and without RWH systems (RWH

vs BAU scenarios) ...... 33

2.3.4 Per person and per area metrics ...... 36

2.3.5 On which type of building should one implement RWH? ...... 38

2.4 Conclusions ...... 42

2.5 References ...... 45

2.6 Supporting Information: Introducing Demand to Supply Ratio as a New

Metric for Understanding Life Cycle Greenhouse Gas (GHG) Emissions

from Rainwater Harvesting Systems...... 50

3 Environmental impacts from harvesting rainwater: A comparison of supply versus

demand based analysis for different climatic regions ...... 76

2.1 Introduction ...... 79

2.2 Methodology ...... 82

2.2.1 Life cycle modeling approach ...... 82

2.2.2 Building characteristics ...... 88

2.2.3 System sizing ...... 88

2.2.3.1 Demand and Supply ...... 88 vii

2.2.3.2 Tank Sizing ...... 89

2.2.3.3 Dual Piping ...... 92

2.2.3.4 Energy use by pump ...... 92

2.2.3.5 Other rainwater harvesting components ...... 93

2.2.4 Regional precipitation variation ...... 94

2.2.5 Sewer Type ...... 95

2.3 Results and Discussion ...... 97

2.3.1 Overview of key parameters affecting environmental impacts .... 97

2.3.2 Extent of Variation of Absolute Impacts (Positive bars in Figure

3.6 and 3.7) ...... 101

2.3.3 Percentage Contribution to Absolute Impacts (Positive bars in

Figure 3.6 and 3.7) ...... 104

2.3.4 Avoided Impacts from RWH system (Negative bars in Figure 3.6

and 3.7) ...... 105

2.3.5 Net Impacts from RWH system (red dots in Figure 3.6 and 3.7) 106

2.3.6 Discussion and Implementation ...... 110

2.4 Conclusions ...... 111

2.5 References ...... 113

2.6 Supporting information: Environmental impacts from harvesting

rainwater: A comparison of supply versus demand based analysis for

different climatic regions...... 118

viii

4 Environmental impacts from Implementing Rainwater Harvesting System at a

Suburban Neighborhood: A comparison of centralized versus decentralized tank

location and pervious to impervious area ratio ...... 129

4.1 Introduction ...... 130

4.2 Methodology ...... 134

4.2.1 Base watershed...... 137

4.2.2 Building classification ...... 138

4.2.3 Demand and supply ...... 138

4.2.4 Cistern sizing and cistern location ...... 139

4.2.5 Dual piping calculation ...... 142

4.2.6 Pumping design ...... 143

4.2.7 Hydrologic simulation ...... 144

4.2.8 Centralized vs decentralized ...... 146

4.2.9 Sensitivity of soil type ...... 147

4.2.10 Sensitivity of pervious to impervious area ratio ...... 147

4.2.11 Life cycle assessment ...... 148

4.3 Results and Discussion ...... 150

4.3.1 Cistern Sizing ...... 150

4.3.2 Stormwater Runoff and associated Environmental Impacts ...... 152

4.3.2.1 Comparison with BAU ...... 152

4.3.2.2 Effect of downspout connection ...... 152 ix

Downspout connected to pervious areas ...... 152

Downspout connected to impervious areas or storm drain

...... 155

4.3.3 Effect of pervious to impervious area ratio ...... 160

4.3.4 Selection framework for implementing RWH in a neighborhood

161

4.3.5 Limitation ...... 163

4.4 Conclusions ...... 163

4.5 References ...... 166

4.6 Supporting information: Environmental impacts from Implementing

Rainwater Harvesting System at a suburban neighborhood: A comparison

of centralized versus decentralized tank location and pervious to

impervious area ratio...... 171

5 Conclusions ...... 184

References ...... 187

List of Tables

Table SI - 2. 1 Inventory of BAU and RWH components for an average medium office

building ...... 56

Table SI - 2. 2 Inventory of RWH components for all the roof area variations when

occupancy was kept constant at 268 – When building is served by a

combined sewer network. Neglecting the minor components...... 58

Table SI - 2. 3 Inventory of RWH components for all the occupancy variations when

building roof area was kept constant at 53,628 sq. ft. – When building is

served by a combined sewer network. Neglecting the minor components 60

Table SI - 2. 4 Inventory of RWH components for all the roof area variations when

occupancy was kept constant at 268 – When building is served by a

separate sewer network. Neglecting the minor components ...... 62

Table SI - 2. 5 Inventory of RWH components for all the occupancy variations when

building roof area was kept constant at 53,628 sq. ft. – When building is

served by a separate sewer network. Neglecting the minor components . 64

Table SI - 3. 1 Inventory for Supply based functional unit when the building is connected

to a separate sewer network. 122

Table SI - 3. 2 Inventory for Supply based functional unit when the building is connected

to a combined sewer network...... 123

Table SI - 3. 3 Inventory for Demand based functional unit when the building is

connected to a separate sewer network...... 124

Table SI - 3. 4 Inventory for Demand based functional unit when the building is

connected to a combined sewer network...... 125

Table SI - 3. 5 Percentage change in life cycle environmental impacts due to percentage

change in precipitation across various geographical locations – supply

based functional unit. Las Vegas (with lowest mean monthly ppt) was

taken as reference; percentage change in impacts were calculated with

respect to increase from Las Vegas. A positive number indicates decrease.

...... 126

Table SI - 3. 6 Percentage change in life cycle environmental impacts due to percentage

change in precipitation across various geographical locations – demand

based functional unit. Las Vegas (with lowest mean monthly ppt) was

taken as reference; percentage change in impacts were calculated with

respect to increase from Las Vegas. A positive number indicates decrease.

...... 127

Table SI - 3. 7 Main characteristics of the precipitation and its effect on RWH system for

different climatic regions...... 128

Table 4. 1 Rainwater cistern sizes for different scenarios. There were nine, type A; five,

type B; eight, location 1; four, location 2; two; location 3 and one,

centralized cisterns. 150 xii

Table SI - 4. 1 Soil characteristics for the suburban neighborhood presented in Figure SI-

1 (collected from web soil survey, Natural Resource Conservation Service,

United States Department of Agriculture)...... 176

Table SI - 4. 2 Inventory of RWH system components for building type A...... 178

Table SI - 4. 3 Inventory of RWH system components for building type B...... 179

Table SI - 4. 4 Inventory of RWH system components for cistern location 1...... 180

Table SI - 4. 5 Inventory of RWH system components for cistern location 2...... 181

Table SI - 4. 6 Inventory of RWH system components for cistern location 3...... 182

Table SI - 4. 7 Inventory of RWH system components for centralized cistern location. 183

xiii

List of Figures

Figure 2. 1 System Boundary for the LCA ...... 20

Figure 2. 2 Schematic representation of conceptual LCA model ...... 24

Figure 2. 3 Percentage Contribution to life cycle GHG emissions for combined and

separate sewer (a) when roof area is constant and (b) when occupancy is constant.

“Sep” represents separate sewer scenario whereas “comb” represents combined

sewer scenario...... 28

Figure 2. 4 Three conditions affecting life cycle impacts from RWH systems ...... 29

Figure 2. 5 Life cycle GHG emissions for (a) constant roof area and (b) constant

occupancy ...... 30

Figure 2. 6 Decision matrix guide to implement RWH in an office building...... 40

Figure SI - 2. 1 Percentage Contribution to Life cycle GHG emission of implementing

harvested rainwater in an average medium office building. 66

Figure SI - 2. 2 RWH scenario life cycle GHG emissions in average medium office

buildingfor combined sewer– (a) constant roof area, changing occupancy and (b)

constant occupancy, changing roof area...... 67

xiv

Figure SI - 2. 3 Life cycle GHG emission in average medium office building for separate

sewer– (a) constant roof area, changing occupancy (b) constant occupancy,

changing roof area...... 68

Figure SI - 2. 4 RWH scenario life cycle GHG emissions in average medium office

building for combined sewer when the rainwater cistern is sized using Yield After

Spillage approach and daily precipitation– (a) constant roof area, changing

occupancy (b) constant occupancy, changing roof area...... 69

Figure SI - 2. 5 Life cycle GHG emissions per person in medium office for (a) constant

roof area, changing occupancy, (b) constant occupancy, changing roof area.

Business as usual (BAU) and rainwater harvesting (RWH) scenarios when

building is connected to combined sewer network...... 70

Figure SI - 2. 6 Life cycle GHG emissions per person in medium office for (a) constant

roof area, changing occupancy, (b) constant occupancy, changing roof area.

Business as usual (BAU) and rainwater harvesting (RWH) scenarios when

building is connected to separate sewer network...... 71

Figure SI - 2. 7 Life cycle GHG emissions per square meter in medium office for (a)

constant roof area, changing occupancy, (b) constant occupancy, changing roof

area. Business as usual (BAU) and rainwater harvesting (RWH) scenarios when

building is connected to combined sewer network...... 72

Figure SI - 2. 8 Life cycle GHG emissions per square meter in medium office for (a)

constant roof area, changing occupancy, (b) constant occupancy, changing roof

area. Business as usual (BAU) and rainwater harvesting (RWH) scenarios when

building is connected to separate sewer network...... 73

Figure SI - 2. 9 Life cycle GHG emissions of RWH in a small office building for

combined sewer– (a) constant roof area, changing occupancy and (b) constant

occupancy, changing roof area. Separate sewer results can be extracted from this

figure by just removing stormwater treatment...... 74

Figure SI - 2. 10 Life cycle GHG emissions of RWH in a large office building for

combined sewer– (a) constant roof area, changing occupancy and (b) constant

occupancy, changing roof area. Separate sewer results can be extracted from this

figure by just removing stormwater treatment...... 75

Figure 3. 1 System boundary of rainwater harvesting system for supply based analysis

and demand based analysis. Hatched portion represents operational phase whereas

plain region is represents construction phase. 84

Figure 3. 2 Schematic representation of conceptual LCA model for RWH systems for

different regions with two different functional units...... 87

Figure 3. 3 Nine climatic regions within the contiguous United States identified by

National Climatic Data Center (NCDC). Representative city for each climatic

regions are displayed in the Figure...... 95

Figure 3. 4 Rainwater cistern curves with respective sizes and volumetric reliabilities for

different climatic regions. City TOL represents Toledo OH, SLC represents Salt

Lake City UT, SEA represents Seattle WA, MAN-UK represents Manchester –

United Kingdom, L.VEG represents Las Vegas NV, L. CRO represents La Crosse

WI, HOU represents Houston TX, F. MYE represents Fort Myers FL, FAR

represents Fargo ND and BOS represents Boston MA...... 97

xvi

Figure 3. 5 Scatter plot matrix for cistern size (C), volumetric reliability (Vr), demand to

supply ratio (D/S0, percentage of rainwater spilled (% spillage), percentage of

days without rain, and average daily precipitation...... 100

Figure 3. 6 Life cycle environmental impacts per cubic meter of rainwater supplied for

RWH system according to supply based approach...... 108

Figure 3. 7 Plot of life cycle environmental impacts from RWH system according to

demand based approach...... 109

Figure SI - 3. 1 Rainwater tank size for different regions with respective average daily

precipitation. A second degree polynomial fit the data fairly well with an R square

value of 0.88. A R square value close to 1 represents the best fit. 121

Figure 4. 1 Schematic representation of conceptual hydrologic – LCA model used for

RWH system at neighborhood scale 136

Figure 4. 2 Schematic representation of (a) decentralized RWH tank, (b) partly

centralized RWH tank and (c) centralized RWH tank locations...... 142

Figure 4. 3 Schematic of how SWMM fits in the analysis...... 145

Figure 4. 4 Life cycle (a) runoff (gallons), (b) Energy (kWh), (c) GWP (Kg CO2 eq), (d)

Acidification (Kg H+ eq), (e) Eutrophication (Kg N eq) and (f) Ozone depletion

(Kg CFC 11 eq) of RWH system at a suburban neighborhood when downspout is

connected to storm-sewer...... 159

Figure 4. 5 Percentage reduction in stormwater runoff due to change in roof area to grass

area ratio...... 160

xvii

Figure 4. 6 Selection framework for environmental assessment of RWH system in a

neighborhood...... 162

Figure SI - 4. 1 Soil characteristics for the suburban neighborhood (collected from web

soil survey, Natural Resource Conservation Service, United States Department of

Agriculture). 172

Figure SI - 4. 2 Variation of cistern size with the increase in roof area keeping occupancy

constant...... 172

Figure SI - 4. 3 Variation of cistern size with the increase in occupancy keeping roof area

constant...... 173

Figure SI - 4. 4 Variation of cistern size with the increase in roof area and occupancy

keeping roof area to occupancy ratio constant...... 173

Figure SI - 4. 5 Life cycle (a) runoff (gallons), (b) Energy (kWh), (c) GWP (Kg CO2 eq),

(d) Acidification (Kg H+ eq), (e) Eutrophication (Kg N eq) and (f) Ozone

depletion (Kg CFC 11 eq) of RWH system at a suburban neighborhood when the

downspout is connected to grass and the soil type is sand or loam...... 174

Figure SI - 4. 6 Life cycle (a) runoff (gallons), (b) Energy (kWh), (c) GWP (Kg CO2 eq),

(d) Acidification (Kg H+ eq), (e) Eutrophication (Kg N eq) and (f) Ozone

depletion (Kg CFC 11 eq) of RWH system at a suburban neighborhood when the

downspout is connected to grass and the soil type is clay...... 175

Figure SI - 4. 7 Percentage change in stormwater runoff due to change in roof area to

grass area ratio...... 176

xviii

List of Abbreviations

BAU ………………………..Business as Usual

RWH ……………………….Rainwater Harvesting

D/S …………………………Demand to Supply ratio

CSO ………………………..Combined Sewer Overflow

LCC ………………………..Life Cycle Costing

LCA ………………………..Life Cycle Assessment

GHG ……………………….Greenhouse Gas

xix

List of Symbols

Ƞ ……………………………Rainwater catchment system efficiency.

η …………………………….Combined mechanical and electrical efficiency of the pump

ϒ ……………………………Specific weight of water (N/m3)

α …………………………….Percentage of energy lost due to friction

Chapter 1

Introduction

Today’s urban water infrastructure suffers from inefficiencies, energy constraints, and lack of funding. For example, inefficient performance of urban water supply systems has been estimated to result in leaks of 10-30% and more of treated potable water in distribution systems prior to end use (U.S. EPA, 2009). Energy requirements for water services has been identified in the U.S. as an important inter-connection to water, with studies in California and Utah finding that a remarkable fraction (6-20%) of regional energy demand is required to provide water services (California Energy Commission,

2005; Larsen and Burian, 2011). Today’s aging and sprawling infrastructure is also contributing to a massive need for investment. However, estimates for the U.S. indicate that the clean water and drinking water infrastructures face a funding shortfall of more than

$500 billion if capital investment, operations and maintenance remain at current levels

(U.S. EPA, 2002). Repairs and maintenance alone on public water supply systems are estimated to require $334 billion from 2007 to 2027 (U.S. EPA, 2009). Addressing these problems requires not only substantial investment in urban water infrastructure – now and

in the future – but also changing the approach to urban water management. Rainwater harvesting can be one such strategy to address these challenges.

Rainwater harvesting has been used for centuries to meet urban water demands

(Reid, 1982) and has rapidly gained renewed interest. It has been and remains common practice in Africa, Asia, and Australia to entirely meet or supplement water supply needs

(Lassaux et al., 2007; Furmani, 2008; Han and Mun, 2008; Gould and Nissen-Petersen,

2008; Glendenning and Vervoort, 2010; Karim, 2010; Rahman et al., 2010; Tam et al.,

2010; Kahinda and Taigbenu, 2011; Alam et al., 2012; Huston et al., 2012). It is becoming common practice to not only supplement water supply but also to manage stormwater runoff, such as in Germany and the United Kingdom (Herrmann and Schmida, 1999;

Nolde, 2007; Ward et al., 2010). Promoted by the green building industry, water conservation proponents, the stormwater management community, and necessitated by recent droughts, rainwater harvesting has renewed interest in the U.S. as well (Burian and

Jones, 2010; Gleick, 2010; Jones and Hunt, 2010; Lynch and Deborah, 2010; Sample and

Liu, 2014). Kinkade-Levario (2007) reported in 2007 more than 250,000 rainwater harvesting systems were in use in the U.S.

Urban rainwater harvesting is typically accomplished by diverting rainfall-derived runoff to a location where it can be used or stored for a later beneficial use or even release

(Myers 1975). In its simplest form, rainwater harvesting is designed to capture and convey runoff from a catchment to a landscaped area for infiltration. Rainwater harvesting, in this manner, follows low-impact development based stormwater management principles seeking to recreate the natural hydrologic cycle (Dietz, 2007). In more complicated systems, harvested rainwater can be used for other indoor uses – occasionally for laundry 2

but more commonly for toilet flushing (Anand and Apul, 2010; Rahman et al., 2010). In these systems, rain is similarly harvested from the roof of the building (or from pavement

(Gomez-Ullate 2011)); then it is filtered and stored in a cistern until it is used.

Historically, many ancient civilizations used rainwater harvesting to meet water supply needs (Myers, 1975). Currently, rainwater harvesting supplied water is used primarily for non-potable water needs including toilet flushing (Bronchi et al., 1999;

Fewkes, 1998; Furumai, 2008), landscape irrigation (Li et al., 2002), crop irrigation (Yuan et al., 2003 and Liang et al., 2011), laundry washing (Bronchi et al., 1999; Angrill et al.,

2012), car and parking lot cleaning (Ghisi and Mengotti de Oliveira, 2007; Villarreal et al.

2005), water cooling (Furumai, 2008) and creation of recreational waterways (Furumai,

2008). Of the end uses of harvested rainwater, toilet flushing has been most promising because water use in buildings accounts for about 11% of fresh water withdrawals in the

US (Barber, 2009) and 27% of indoor water use goes to toilet flushing (Mayer et al 1999 and Gleick, 1996). In addition, the toilet usage does not show large daily variability

(Fewkes, 1998).

In the past couple decade’s environmental life cycle assessment has become a powerful and commonly used tool for assessing the impacts from a product or a service from cradle to grave. The basic modeling structure of a life cycle assessment is that the life cycle assessment tracks environmental impacts over the life cycle of the product or service. The life cycle environmental impacts of implementing rainwater harvesting systems vary in the literature. For example, the use of high efficiency fixtures reduces environmental impact from rainwater harvesting system (Ghisi et al., 2014; Anand and

Apul, 2010; Crettaz et al., 1999; Chiu et. al., 2009 and Racoviceanu et. al., 2010). 3

Rainwater harvesting systems have better environmental performance (Angrill et al., 2012;

Morales-Pinzón et al., 2012; Vargas-Parra et al., 2013 and Bronchi et. al., 1999) in compact and dense urban environments than in diffuse neighborhoods. The environmental impacts of rainwater harvesting system for toilet flushing was found to be lower than municipal water supply if the rainwater cistern was placed on the roof top thereby avoiding energy for pumping (Ghimire et al., 2014). When the use of rainwater for irrigation versus toilet flushing is compared, irrigation end use leads to less environmental impacts (Devkota et al., – 2015).

One issue with these prior RWH LCA studies is that they are focused on specific buildings. For example, several RWH LCA studies have focused on residential buildings

(Racoviceanu et. al., 2010; Proenca et al., 2011; ; Angrill et al., 2012; Morales-Pinz_on et al., 2012; Devkota et al., 2013; and Ghimire et al., 2014) while few have focused on educational buildings (Anand and Apul, 2010 and Bronchi et al., 1999) or dormitory

(Devkota et al., 2015). It was not only the building types that varied. Building locations, life time, end use of rainwater, sources of data, functional unit and system boundary also varied. As was also mentioned by Angrill et al. (2012), the environmental criteria to determine the sustainability of rainwater harvesting systems is still underdeveloped. The limited number of publications on this topic revealed some general trends but also indicated the high variability of the results on this topic. From these studies it is not possible to discern if the results can be generalized to other cases. This was due to variability in the systems modeled and the way the model was parameterized.

Starting out with the idea that the compact density neighborhood was more environmentally attractive than diffuse neighborhood (Angrill et al., 2012; Morales-Pinzón 4

et al., 2012; Vargas-Parra et al., 2013 and Bronchi et. al., 1999), it was clear that building roof area and occupancy played a vital role in environmental impacts from rainwater harvesting system. Therefore a common parameter that could potentially relate the building characteristics (roof area, occupancy) and the environmental impact was essential. Roof area is directly proportional to the volume of rainwater supply from the building while number of people in the building is directly proportional to the water needed to fulfill the urban water needs (toilet flushing in this case). Knowing that every building has different building roof area to occupancy combination, the ratio of occupancy to roof area, further transformed to demand to supply ratio could be the metric to estimate the environmental impacts from rainwater harvesting systems for different building types.

Another variable that could potentially affect the environmental feasibility is the functional unit. The choice of functional unit depends on how the rainwater harvesting system is designed to perform. In acting as a supplemental water source, rainwater harvesting system can serve two functions; supply of rainwater and fulfill water demand.

The specific goal and associated functional unit of RWH systems is affected by this situation. When the goal is stated from a supply perspective, and the purpose is to provide an alternative water source, an appropriate functional unit would be to supply 1 m3 of rainwater. This functional unit was used by several authors (Bronchi et al. 1999; Angrill et al. 2012; Ghimire et al. 2014 and Wang and Zimmerman 2015). On the other hand, the goal can also be stated from a demand perspective in which case the functional unit is selected based on the purpose of meeting the water demand. In this case, if the harvested rainwater is used for toilet flushing, the functional unit can be expressed as providing sanitation services during the life time of the building (Anand and Apul 2010 and Devkota 5

et al. 2015) or as toilet flushing need per person per day (Crettaz et al. 1999). The choice of functional unit depends on the research question being asked in the study and has implications on what is included in the system boundary. If a rainwater harvesting system is installed at a building where the supply of rainwater is not sufficient to fulfill the demand

(e.g. of toilet flushing), potable water will be required to supplement rainwater. In such a case a functional unit based on how water is being used (e.g. sanitation services) may be a more appropriate choice than a functional unit based on volume of rainwater provided (e.g. one cubic meter of rainwater).

The choice of functional unit has impact on what is included in the system boundary, which determines the processes included or excluded in the life cycle assessment model. Defining the system boundary is an integral and critical part of any life cycle assessment study. However most of the studies on life cycle assessment of rainwater harvesting systems either did not clearly state the system boundary or had very different system boundaries. For example, water and wastewater were included in the system boundary by some authors (that used sanitation service as the functional unit) (Anand and

Apul, 2010 and Devkota et al., 2015) but were excluded in the system boundary by others

(that used 1 m3 of rainwater as functional unit) (Bronchi et. al., 1999; Ghimire et al., 2014 and Angrill et al., 2012). The choice of system boundary depends on the functional unit and the research question being asked. Two studies identified that water and wastewater treatment has the major contribution to the environmental impact of implementing harvested rainwater (Devkota et al., 2015; Anand and Apul, 2010) whereas another study identified that the storage tank has the highest impact (Racoviceanu et al., 2010). Energy use by pump in the operational phase is identified as a major contributor to environmental 6

impact by some studies (Bronchi et al., 1999 and Ghimire et al., 2014). These inconsistent results are mainly due to the difference in the system boundary selected in the studies. Our observation from multiple studies is also supported by another study which reported that energy expenditures and emissions varied with the selection of system boundaries

(Racoviceanu et al., 2010).

One more parameter that affects the environmental feasibility of rainwater harvesting system is the scale of its implementation. For example, Farreny et al., (2011) reported that rainwater harvesting systems installed at a neighborhood scale have proven to be more economically feasible as compared to the building scale. Rainwater harvesting systems combined with a storage tunnel were reported to be economically and hydrologically efficient when implemented at the city scale in city of Toledo, Ohio to reduce combined sewer overflow (Tavakol-Davani et al., 2015 - accepted).

A neighborhood consists of various building types such as residential, commercial, institutional, super market and others. Per person water use in residential buildings is higher than other building types (Vickers, 2001). Additionally, the decision to implement RWH system in these neighborhoods would be easier as there would be less people involved in the management compared to individual residential houses and commercial buildings.

Considering the larger scale implementation of rainwater harvesting system, the designer would be interested to design the system with least cost and environmental impacts associated with it. Knowing that apartment complexes are in close proximity with each other, one rainwater cistern could be used for one building (decentralized cistern) or for multiple buildings (centralized cistern). Previous research have highlighted that centralized

rainwater harvesting system were proven to be attractive in a high density neighborhood from a water saving perspective (Vargas-Parra et al., 2013).

It is worth mentioning that rainwater harvesting system not only saves the potable water but also saves the stormwater being treated if the building is served by a combined sewer network. Rainwater harvesting systems have the capacity to reduce combined sewer overflow events. Every year, about 3.2 billion cubic meters of sewage and stormwater from combined sewer overflow events contaminate U.S. water bodies with diluted raw sewage

(U.S. EPA, 2004). Capture of harvested rainwater reduces the stormwater runoff and in turn reduces the frequency and/or volume of combined sewer overflow events. The amount of stormwater runoff reduction generally increases as cistern size increases, but not necessarily linearly due to the complexities of urban watersheds. For example, doubling storage capacity by adding a second rain barrel is known to only increase the runoff volume reduction by 1-2% for a cross section of cities in the U.S. (Steffen et al., 2013).

Substantially increasing the cistern size from 189 L to 1890 L doubles the runoff volume reduction to 12%-16%. The effectiveness of rainwater harvesting systems for stormwater control varies as a function of implementation scale for a given building rooftop – cistern size combination. Installing rainwater cisterns in a Portland neighborhood can reduce average annual runoff volume by 68% (Crowley, 2005). A single 189 liter (50 gal) rain barrel implemented at every house of a typical residential neighborhood can reduce average annual stormwater volume up to 12% in Mountain West cities such as Salt Lake City, Utah and as low as 4% in Savannah, Georgia (Steffen et al., 2013). In San Diego, California, reductions of 2.5% - 13.7% (total outflow volume) and 0.4% - 1.9% (peak flow rate) can

be achieved for the smallest to largest cistern capacities (227-liter to 7,571 liter) for a range of watershed implementation extents (25% to 100%) (Walsh et al., 2014).

As per the earlier discussions, the literature on LCA of RWH system lacks data driven conclusions on generalization of the results as well as the factors affecting environmental sustainability. In case of generalization of the results, there seemed a necessity of a common metric or a factor that affects environmental sustainability of rainwater harvesting systems and that can be used to relate both the input and output parameters. We argued that the developed metric is expected to be helpful in providing a framework for the future researchers. We proposed and demonstrated the building characteristics (occupancy and roof area) can be mathematically represented as the ratio of annual water demand to rainwater supply (D/S). Therefore, the first goal of this study was to demonstrate the effect of D/S ratio on environmental emissions of RWH systems. The prior research on RWH system also lacked consistency in analyzing the system. Some research focused on supply based approach to estimate environmental impacts while others proposed demand based approach. We hypothesize that the choice of functional unit and associated system boundary is affected by the approach chosen. With the two mostly used functional units, namely ‘one cubic meter of rainwater supplied’ and ‘provision of sanitation service in a building’ for supply and demand based analysis, the second goal of this study was to address the dual issue of supply and demand associated with RWH systems by considering two separate but related functional units. The effect of functional unit was demonstrated for ten cities in different climatic regions. With the results from the building scale, it was evident that the prior research also lacks the environmental aspects of hydrologic benefits associated with RWH system. Knowing that the multifamily 9

apartment consist of 40% of total housing supply in suburban settlements, implementation of RWH system at a suburban neighborhood with multifamily apartment complexes would reduce higher potable water supply and stormwater runoff as non-potable water consumption in residential buildings was highest among commercial and residential building. We hypothesize that a single rainwater cistern could supply rainwater to a single building or for multiple buildings. Also there is no strict guideline on the pervious area to impervious area ratio while developing a neighborhood. Therefore the third goal of this study was to evaluate environmental sustainability of RWH system at a suburban neighborhood with different RWH implementation strategies (centralized, partly centralized or decentralized). Towards addressing the above three mentioned research needs, this dissertation asks three questions:

Question 1: What are the factors affecting environmental sustainability of

rainwater harvesting system and how do they affect environmental sustainability

of rainwater harvesting systems?

Question 2: What is the effect of different analysis approach in environmental

sustainability of rainwater harvesting system? What is the environmental

sustainability of rainwater harvesting system at different climatic regions of US?

Question 3: What is the environmental sustainability of rainwater harvesting system

at a neighborhood scale? What are the various rainwater cistern implementation

strategies? What would be the effect of pervious and impervious area ratio in the

environmental impacts of implementing rainwater harvesting system at a

neighborhood scale?

To answer these questions, two types of tools were developed and used to prepare separate manuscripts. The tools developed were:

1) Previously developed building scale EIOLCA model was modified to create a

process based LCA model.

2) Neighborhood scale hydrology-LCA model using process based life cycle data.

Four manuscripts were planned based on the above research questions.

Manuscript 1: Devkota, J., Apul, D., Burian, J. and Davani H. (accepted with revisions)

Review of Rainwater Harvesting Benefits for Urban Water Infrastructure Sustainability.

Critical Reviews in Environmental Science and Technology.

Manuscript 2: Devkota, J., Apul, D., Burian, J. and Davani H. (2015) Introducing Demand to Supply Ratio as a New Metric for Understanding Life cycle Greenhouse Gas (GHG)

Emissions from Rainwater Harvesting Systems. Journal of Cleaner Production.

Manuscript 3: Devkota, J., Apul, D., Jeswani, H. and Azapagic, A. (in prep) Environmental

Impacts from Harvested Rainwater: A comparison of supply versus demand based analysis for different climatic regions. In preparation for submission to International Journal of Life cycle Assessment.

Manuscript 4: Devkota, J., Dietrich, A., Yarlagadda, R., Gruden, C., Burian, S., and Apul,

D.S. (in prep) Environmental impacts from Implementing Rainwater Harvesting System at a Suburban Neighborhood: A comparison of centralized versus decentralized tank location and pervious to impervious area ratio. In preparation for Journal of Industrial Ecology. 11

Chapter 1 focuses on background and scope of the study. This chapter is partly adapted from manuscript 1. Rest of the three manuscripts are presented as a standalone separate chapter in this dissertation, each having their own literature review, methods, results, conclusions and references. Chapter 2 focuses on answering the first question. In this chapter impacts of implementing RWH systems are compared with business as usual scenario (municipally supplied potable water system) for an average medium office building in US. First question focusing on factors affecting sustainability of RWH systems is answered in Chapter 2 and part of Chapter 3. Two major factors (Demand to supply ratio and combined versus separate sewer) are discussed in Chapter 2. Chapter 3 followed the extension of factors affecting environmental sustainability of RWH System. Supply and demand based analysis approach are introduced for the first time to evaluate environmental impacts from rainwater harvesting systems. Second question focusing on precipitation variability is also answered in Chapter 3. Nine climatic regions of United States as well as one climatic region from United Kingdom are selected to best describe the effect of regional precipitation pattern in the environmental sustainability of rainwater harvesting system. Third question focusing on environmental impacts at a neighborhood scale and various rainwater cistern implementation strategies is answered in Chapter 4. A SWMM-

LCA coupled model was developed to evaluate environmental impacts across neighborhood scale. The context and research needs related to each question are explained in detail in separate chapter.

Chapter 2

Introducing Demand to Supply Ratio as a New Metric for Understanding Life Cycle Greenhouse Gas (GHG) Emissions from Rainwater Harvesting Systems

Jay P. Devkota1, Steven J. Burian2, Hassan Tavakol-Davani2, Defne S. Apul1*,

1 2801 W. Bancroft, Department of Civil Engineering, University of Toledo, Toledo, OH,

43606

2110 S. Central Campus Dr., Department of Civil & Environmental Engineering,

University of Utah, Salt Lake City, UT, 84112

*Corresponding author tel: (419)-530-8132; email: [email protected]

Abstract

Rainwater harvesting (RWH) is a decentralized approach to meet non-potable water supply needs and stormwater management goals. Life cycle environmental impacts of RWH systems have been reported in previous studies, but the effects of different building configurations and the type of sewer connections have not been fully studied. In this study, we aim to go beyond case studies by developing an approach that shows how

RWH LCA results change for different building roof areas, occupancies, and sewer

connections. We propose and analyze the ratio of building occupancy to roof area which can be expressed as demand to supply (D/S) ratio to estimate life cycle GHG emissions of implementing RWH system when the harvested rainwater is used to flush the toilets.

Result showed that for all the building roof area to occupancy configurations considered in this study, RWH systems had lower GHG emissions except in some separate sewer scenarios. Size of the cistern, water savings as well as life cycle GHG emissions varied as a function of D/S ratio. It was found that changing roof area and occupancy have different effects on cistern size, water savings and life cycle GHG emissions measured with respect to D/S ratio. Water savings and cistern size increased until D/S equaled 1 and remained constant for higher value when roof area was constant. The duo were constant until D/S ratio of 1 and decreased for higher D/S value. Though minimum life cycle GHG emissions were noticed for spacious building, the maximum savings in emissions were noted at D/S ratio equal to or more than 1 when the building footprint was kept constant when the building was connected to a combined sewer network. For occupancy constant case, maximum savings was reported at D/S equal to 1. Similarly when the building was connected to a separate sewer network, minimum emissions as well as maximum savings were reported at lowest possible D/S value when the building footprint was constant and at D/S less than or equal to 1 when the occupancy was kept constant. A recommendation framework was provided based on the results obtained to help designers and practitioners design the RWH system to minimize GHG emissions.

2.1 Introduction

Rainwater harvesting (RWH) is gaining more attention in the United States (US) and globally to address limited water resources and water leakages, stormwater management needs, and financial challenges faced by existing water supply systems (Jones and Hunt, 2010; CEC, 2005; Mehta, 2009 and USEPA, 2002). Common uses of harvested rainwater are toilet flushing (Bronchi et al., 1999; Fewkes, 1998 and Furumai, 2008;

Ghimire et al., 2014; Devkota et al., 2013; Angrill et al., 2012), irrigation (Li and Gong,

2002 and Stout et al., 2015), laundry washing (Bronchi et al., 1999 and Angrill et al., 2012), car and parking lot cleaning (Ghisi et al., 2009 and Villarreal et al., 2005), and water cooling (Furumai, 2008). Use of harvested rainwater in toilet flushing is especially promising because it constitutes a large fraction (27 %) of indoor non-potable water use

(Mayer and William, 1999; Vickers, 2001) that does not require significant water treatment

(Krishna, 2005; Hermann and Schimda, 2000 and Coombes et al., 2002). Rainwater harvesting has stormwater management benefits as well (Walsh et al. 2014). For example,

Steffen et al. (2013) reported the potential for a reasonable implementation of rainwater harvesting to reduce up to 20% of stormwater runoff in most regions of the US. In communities served by combined sewer systems, this reduction of stormwater runoff can reduce combined sewers overflows (CSO) and ultimately reduce the estimated 3.2 billion cubic meters (USEPA, 2004) of diluted sewage released to receiving waters of the US each year by CSOs.

There is a fast growing literature on environmental life cycle assessment (LCA) of

RWH systems, with many studies focusing on using harvested rainwater for toilet flushing.

From these studies, we have some understanding of how RWH system impacts vary for different conditions. For example, RWH scenarios with high efficiency toilets have lower

GHG emissions than those with standard toilets (Anand and Apul, 2010; Crettaz et al.,

1999 and Racoviceanu et al., 2007). The cost and environmental impacts of potable water can play an important role in whether the RWH system will be economically and environmentally attractive (Devkota et al., 2013). The environmental impacts of non- potable use of rainwater for toilet flushing can be lower than that of using municipal water if rainwater cistern is placed on the rooftop thereby avoiding pumping (Ghimire et al.,

2014). Implementing RWH system in newly constructed buildings results in lower environmental impact than in renovated buildings (Angrill et al., 2012 and Devkota et al.,

2015). Using rainwater for irrigation results in lower environmental impact than using it for toilet flushing (Devkota et al., 2015). Environmental impacts from RWH systems also vary with population densities with compact densities resulting in lower impact than diffuse densities (Angrill et al., 2012).

One issue with prior RWH LCA studies is that they often studied specific buildings.

From prior studies it is not possible to discern if the results can be generalized to other cases. In this study, we aimed to go beyond case studies by developing an approach that shows how RWH LCA results change for different system configurations. We proposed and demonstrated that different system configurations can be represented by the balance between water demand and water supply in a building which can be mathematically represented as the ratio of annual water demand to rainwater supply (D/S). The water demand parameter captures the outdoor (e.g. irrigation) and indoor (e.g. toilet flushing) water uses in a building. The rainwater supply parameter captures the effects of the Roof 16

area, annual precipitation, and tank size. The interplay between demand and supply of water is also important from a green infrastructure (GI) perspective. While there are many examples of GI (e.g. bioswales, porous pavements, rain gardens, green roofs), RWH is the only GI that affects both the stormwater and the water supply infrastructure directly.

Demand and supply estimations are essential for sizing the rainwater tank (Devkota et al.,

2013; Anand and Apul, 2010; Ghisi and Ferreira, 2007; Aladenola and Adeboye, 2010 and

Palla and Lanza, 2011) and have only been used for that purpose in the RWH literature.

Our study is the first to propose the use of D/S as a key parameter for understanding the life cycle GHG emissions from implementing RWH systems.

2.2 Methodology

2.2.1 Goal and Scope

The goal of this LCA study was to demonstrate the effects of D/S on the environmental life cycle GHG emissions of RWH systems. For our analysis we chose toilet flushing as the end use for harvested rainwater because it constitutes a large fraction (37

%) of water use in office buildings (Saving Water in Office Building, EPA, 2012). The

RWH system was assumed to be implemented at a typical medium size office building in

Toledo, OH. Office buildings constitute only 15 % of the building stock (U.S. Census,

2000) and 5.1 % of the LEED certified buildings (National Green Building Adoption Index,

2014) in the United States. However, the growth of LEED certification in office buildings is very high and it is now almost the norm rather than the exception to construct or manage sustainable office buildings. (National Green Building Adoption Index, 2014). In addition, 17

adoption of a new technology by the society can be more easily facilitated if it is first introduced at the workplace rather than at individual homes. Starting out with a typical medium size office building, the system configurations were varied by changing the occupancy and roof area and calculating the corresponding D/S.

Each building configuration was modeled twice; first assuming the building is connected to a combined sewer, and second assuming the building is connected to a separate sewer. Sewer connection type is important because RWH systems pose an interesting situation with respect to combined sewers. Due to intense precipitation events during wet weather, a combined sewer system can exceed its capacity and overflow. In general, combined sewer systems are considered less environmentally friendly than separate sewers because during a combined sewer overflow (CSO), raw sewage is discharged into water bodies causing water pollution. Collecting the stormwater on site using a RWH system reduces the volume of CSOs (Steffen et al., 2013) and also avoids the treatment of this runoff at the wastewater treatment plant thereby reducing the GHG emissions (Devkota et al., 2015). For a RWH system, the GHG reduction from not treating the stormwater would be in addition to the GHG reduction from not using potable water.

In contrast, when a building is connected to a separate sewer network, there is no need for stormwater treatment since stormwater goes directly to water bodies. Hence, in a separate sewer system, the emission reductions by implementing RWH systems are only from reduction in tap water consumption. Therefore, when RWH system is implemented in a combined sewer network the GHG emission savings are higher than when it is implemented in a separate sewer network (Devkota et al., 2015).

To avoid burden shifting, a comprehensive LCA should include all life cycle phases and all relevant impact categories. Yet installation stage has large data and modeling uncertainties and is also not considered to be critical for the LCA of RWH systems

(Peuportier, 2001 and Blengini and Di Carlo, 2010). Therefore, only raw material extraction, manufacturing, transportation, operation, and disposal phases were modeled in this study. Similarly, instead of modeling a range of environmental impacts, only GHG emissions were modeled to illustrate the efficacy of the proposed D/S ratio as a key system parameter in RWH LCA modeling. We selected global warming impact over other impact categories because GHG emission data are typically easier to access and relate closely to energy usage (and often to other environmental impacts) thereby providing a key metric for business decisions (GHG Protocol).

Previous RWH LCA studies used different functional units to evaluate environmental impacts of using harvested rainwater. For example, Ghimire et al. (2014) used one cubic meter of water (for non-potable domestic and agricultural irrigation).

Angrill et al. (2012) used one cubic meter of rainwater per person per year (for laundry).

Morales Pinzón et al. (2012) used one cubic meter of rainwater for laundry for 50 years.

Bronchi et al. (1999) used water volume normalized to kg of clothes washed (50 l of water to wash 2.75 kg of clothes). Crettaz et al. (1999) used water needed for toilet flushing per person per day.

Starting out with the idea that RWH systems cannot always meet the full water needs of a building, it would be expected that additional municipal water will be required in many cases. In such situations we argue that the supply of unit volume of harvested rainwater is

not an appropriate functional unit because it would not properly capture situations where water demand is higher than available rainwater. Therefore we selected a demand based functional unit such as provision of sanitation services in the building throughout its lifetime (75 years). This functional unit expands the system boundary to include municipally supplied water for cases where water demand cannot be met with harvested rainwater. In the business as usual (BAU) scenario the sanitation services are met by flushing the toilets by municipally supplied water only. In the RWH scenario, the sanitation services are provided by flushing the toilets with rainwater (and municipally supplied water when needed). The reference flow for our functional unit was the total volume of water required to flush the toilets.

Figure 2. 1 System Boundary for the LCA

The system boundary for the RWH scenarios is shown in Figure 2.1. It includes all building infrastructure that relates specifically to toilets (items 1-11 in figure 2.1). The 20

operation phase includes the operation of water and wastewater treatment plants and the energy use by the cistern pump (items 12-14 in figure 2.1). The operation phase also includes replacement of some of the components. For example, filters were assumed to be replaced every 5 years and pumps every 20 years (Kirk and Dell’Isola, 1995). Replacement of toilets depends on various factors such as safety, aesthetics, water conservation policy and others. Water conservation guidelines recommend 20 and 25 years for toilet replacement in commercial and residential buildings, respectively (USEPA, Water

Conservation Plan Guidelines). In the model, we replaced the toilets every 35 years. This lifetime is used by building designers and practitioners (CostLab, 2015). A replacement period of 35 years has also been suggested by Kirk and Dell’Isola (1995).The BAU scenario was the same as the RWH scenario except that it did not include RWH system related components (items 1, 2, 3, 4, 5, 6, 7 and 12 in Figure 2.1). The BAU scenario includes only pipes, toilets, flush accessories, potable water, wastewater, stormwater and replacement of toilet components (items 8, 9, 10, 11, 13 and 14 in Figure 2.1). In the model, the municipally supplied water was connected to the toilets (item 13) not just in the BAU case but also for the RWH case to meet the demand in case the supplied rainwater was not enough to flush toilets in some buildings.

In determining the system boundary we initially considered only modeling the system components that are different between the RWH and BAU scenarios. This approach was taken by Devkota et al. (2013) and Devkota et al. (2015) who calculated energy and

GHG emission payback periods for RWH systems. In this study, we modeled not the relative but the absolute impacts to better understand the contributions of different system components to total impact as well as their effect in various building configurations. 21

Several other RWH LCA studies also modeled absolute impacts from rainwater harvesting systems (Angrill et al., 2012; Ghimire et al., 2014; Morales-Pinzón et al., 2012 and Vargas-

Parra et al., 2013).

The life cycle inventory including system components and their respective material types and quantities included is presented in Table SI-2.1. Separate and combined sewer network assumptions are differentiated in Table SI-2.1. The only inventory difference in the separate sewer scenario is the avoidance of stormwater treatment.

The characteristics of a typical medium office (square footage, number of stories, and occupancies) were taken from Deru et al. (2011) who compiled data on the US building stock and used this information as well as energy efficiency criteria to develop typical U.S. building dimensions and corresponding occupancies (based on 200 ft2 per person). The building square footage (53,628 ft2) was divided by the number of stories (3 stories) to calculate the roof area (17,876 ft2). Water demand for toilet flushing was estimated using building occupancy (268 occupants) (Deru et al., 2011), four flushes per person per day

(Vickers, 2001 and Alliance for Water Efficiency) and flush volume for standard toilets

(1.6 gallons per flush). The annual supply of rainwater was calculated by multiplying the monthly precipitation in Toledo by the building roof area and assuming 25% of rainwater might be lost before entering the cistern (Krishna, 2005). The equations for rainwater supply are provided in section 1 of the supporting information (equations 1 and 2 of section

2.6). To calculate D/S, the annual toilet flushing water demand was divided by the annually available rainwater supply.

To model other system configurations, the demand and supply were varied one at a time. In one set of analysis, occupancy was kept constant and roof area was varied to 22

simulate the variation in supply. In another set of analysis, roof area was kept constant and occupancy was varied to simulate the variation in demand. In varying the roof area and the occupancy, the lowest roof area (8,965 ft2) and highest occupancy (535) values

(corresponding to the higher end of the range of D/S values modeled) were determined from the International Building Code (IBC, 2006) which limits the maximum number of people that can occupy an office space (minimum 100 ft2 per person) based on egress safety concerns (IBC, 2006). The lower end of the D/S ratio (D/S=0.5) which corresponds to the largest roof area (80,620 ft2) and smallest occupancy (59 occupants) values modeled was set to one eighth of the International Building Code requirements assuming it would not be desirable to operate the building and attempts would be made to remedy the situation and increase the occupancy if the D/S ratio fell more than 8 times the maximum occupancy allowed (IBC, 2006). When occupancy was varied the functional unit (amount of sanitation services provided) also changed whereas when roof area was varied, the functional unit remained the same. Due to the two different ways of varying D/S resulting in having different effects on our chosen functional unit, we also analyzed and presented results by normalizing the GHG emissions to per person and per unit area.

2.2.2 Life cycle inventory, impact assessment, and interpretation

Figure 2. 2 Schematic representation of conceptual LCA model

The approach for life cycle inventory, impact assessment and interpretation modeling (shown in Figure 2.2) was adapted from the EEAST model (Devkota et al.,

2013). The primary differences in this study are the use of process based data instead of economic input output data, the modeling of all relevant system components (as opposed to only those that are different between BAU and RWH) and the analysis of results as a function of D/S. In this approach, building characteristics and precipitation are first used to calculate the water demand and supply volumes and then to size the cistern, concrete pad, pump, dual piping, filter, bends, valves, tees, number of toilets. The rainwater storage cistern is designed based on the comparison of volume of rainwater captured from the roof and the water demand for flushing toilets on a monthly basis. If the volume of rainwater

available from capture was more than the toilet flushing demand, then the cistern size was set equal to toilet flushing demand; otherwise the supply governs the cistern size. This method was taken from Texas Water Development Board (Krishna, 2005) and Georgia

Rainwater Harvesting Guideline (2009) and has been previously used by Farreny et al.

(2011); Krishna et al. (2005); Ghisi and Ferreira (2007); Aladenola and Adeboye (2010) and Devkota et al. (2013) It is considered an easy method for rainwater tank sizing in commercial and industrial sectors. This method may size a larger tank compared to sizing methods that use daily rainfall data but is sufficient for illustrating the role of D/S on life cycle GHG emissions. Georgia Rainwater Harvesting Manual also states that the concept of demand and supply is easier to understand if broken down on a monthly basis. (Georgia

Rainwater Harvesting Manual, 2009). However, to be comprehensive, we performed the calculations for daily tank sizing as well. Our daily tank sizing results (included in supporting information- section 2.6) show that the overall conclusions we derive from D/S analysis do not change if daily precipitation is used to tank the size instead of monthly precipitation.

Once the tank was sized, dimensions of a square concrete pad were estimated (100 mm thick and 300 mm greater than the diameter of cistern on each side) with the pad being used as the supporting foundation for the rainwater cistern. The length of the dual piping

(item 4 in figure 2.1) was calculated by estimating the distance from the rainwater tank to the toilets in each floor (Devkota et al., 2013). An equation to estimate approximate lengths of dual piping system is presented in supporting information (Equation 3 in Supporting

Information). In architectural practice, 2 % of piping is assumed to be bends, valves, and tees. This approach was adopted to estimate the number and mass of bends, valves and tees 25

used in the building. Once the systems were sized and the relevant water volumes and D/S were calculated, the energy required by pump was estimated using equation 1 and 2 provided in supporting information. The life cycle inventory (LCI) for the medium sized office (Table SI – 2.1) and how this inventory changes with varying D/S (Table SI 2.2-

Table SI 2.5) are shown in the supporting information. The emission data for these inventory were collected from GaBi (PE International, 2015) and Eco-invent databases

(Ecoinvent, 2.2). TRACI was used as an impact assessment methodology with 100 years time horizon for GHG emission estimation. Special consideration was taken while selecting the source of data. US data were used whenever available. For RWH components such as potable water treatment, wastewater treatment and concrete pad, US data were not available. Therefore, for these inventory items the electricity mix was converted from

European to US electricity mix with an assumption that all other aspects of the processes would be similar in the US. After estimating quantities and per unit emission intensities for all the RWH components, the model was run for two different scenarios; one, when the building is connected to a combined sewer network and another when the building is connected to separate sewer network. Finally, to interpret the results, life cycle emission data were analyzed as a function of D/S.

2.3 Results and Discussion

2.3.1 Characteristics of the medium office buildings that have RWH systems

The D/S of the modeled average medium size office building is 2.25. A D/S value greater than one indicates that there is not sufficient rainwater to meet the toilet flushing 26

demand. Due to the high D/S ratio, the available rainwater is able to meet only 45 % of the water demand in the medium office building. For all of the building configurations modeled, the major life cycle impacts of using RWH system come from wastewater and potable water treatment (Figure 2.3). Similar to findings of Anand and Apul (2010) and

Devkota et al. (2015) the impacts from operational phase are much higher than those of construction phase. The operation phase dominates the life cycle impacts due to the long life time (75 years) of the system. Many of the infrastructure components within the building (pipes, filters, toilets, vents, overflow drain, flush tank, pump, sewer drain pipe, bends valves tees and filter replacement) contribute to less than 1 % of the total life cycle

GHG emissions and can be omitted from the analysis without introducing significant error

(represented by others in Fig SI- 2.1).

(a) 100% 90% 80% 70%

60% to life cycle GHGcycle life to 50% 40%

emissions 30% 20% 10%

Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep

Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Percentage contributionPercentage 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

(b) 100% 90% 80% 70%

to life cyclelife to 60% 50% 40% 30% 20%

GHG emissionsGHG 10%

Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep

Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb Comb

Percentage contributionPercentage 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 Demand to Supply (D/S) ratio 100%

Cistern0% Concrete Pad Energy use by pump

S… S… S… S… S… S… S… S…

C… C… C… C… C… C… C… C… Potable waterC… treatment Waste water treatment Stormwater treatment 4.50 4.25 4.00 3.75 3.50 3.25 3.00 2.75 2.50 2.25 2.00 1.75 1.50 1.25 1.00 0.750.50 Figure 2. 3 Percentage Contribution to life cycle GHG emissions for combined and separate sewer (a) when roof area is constant and (b) when occupancy is constant. “Sep” represents separate sewer scenario whereas “comb” represents combined sewer scenario.

2.3.2 Effect of demand to supply ratio (D/S) on life cycle greenhouse gas emissions

from buildings with RWH systems

Figure 2. 4 Three conditions affecting life cycle impacts from RWH systems

400 (a) More crowded building 535 occupants 350 Average medium office building 268 occupants 300

250 More spacious building 59 occupants 200

150 CO2 CO2 eq) 100

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS Life cycle cycle Life emissions GHG (in thousands Kg 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 400

350 Bigger building

CO2 CO2 2 (b) 80,620 ft roof area Average medium office building

Kg 300 17,915 ft2 roof area Smaller building 8,965 ft2 250 roof area

200

150

100

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

LIfe cycle GHG LIfe GHG cycle emissions (in thousands Demand to supply ratio (D/S)

700Stormwater Waste water Potable water

B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… -300 R… Energy use by pump Concrete Pad Cistern 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

Figure 2. 5 Life cycle GHG emissions for (a) constant roof area and (b) constant occupancy

The effect of D/S ratio on life cycle impacts from RWH systems depends on the three conditions summarized in Figure 2.4 and quantitatively displayed in Figure 2.5 and

Figure SI-2.2. In communities with separate sewers, when the available rainwater (supply) exceeds toilet flushing demand (D/S<1) the excess runoff from the building roof goes to the storm sewer (Condition 1). In communities served with combined sewer system, this excess runoff is conveyed to a wastewater treatment plant (Condition 2). The additional life cycle greenhouse gas emissions resulting from treatment of this excess runoff can be seen on the left side of Figures 2.5a and 2.5b where D/S<1 (see RWH bars). The impacts of treating excess runoff are highest when D/S is at its minimum which can be achieved with either a building with large roof or a building with very few occupants. The impacts from the excess runoff decrease as the D/S ratio increases towards one. When the D/S is equal to one or greater, there is no excess runoff and therefore there are no additional impacts from treating this runoff.

In this study, we assumed all excess roof runoff to be treated at a wastewater treatment plant in combined sewer communities. However, in reality, only some of this storm water and some of the toilet flushed wastewater are treated at the wastewater treatment plant during combined sewer overflows. The percentages that would be treated depend on many factors such as sewer network and capacity, rain intensity, and wastewater treatment plant capacity. Future research with more refined modeling is needed to account for these factors.

In Condition 3, available rainwater is less than the toilet water demand (D/S > 1) and is supplemented with municipally supplied water (Figure 2.4). In this condition, as the

D/S increases, so do the amounts of water supplied from the municipality and the wastewater resulting from toilet flushing. When the occupancy increases (Figure 2.5a,

RWH bars), there is a uniform increase in life cycle emissions resulting from linear increase in potable water and wastewater volumes. However, the GHG emissions do not increase uniformly when roof area changes (Figure 2.5b, RWH bars). In this case, as the building roof area decreases and D/S increases, the increasing impact from potable water supply counteracts the decreasing impact from smaller cistern and concrete pad. The relative changes in impacts from the cistern and from potable water supply result in a nonlinear and less pronounced effect on the life cycle impact compared to the effect of occupancy change.

It is interesting to note how the emissions from cistern change as a function of D/S.

For the medium size office building the emissions from cistern is 7 % of the total emissions

(Figure 2.3 and Figure SI-2.1). The percentage contribution from cistern decreases as demand to supply ratio increases. It can be seen that spacious buildings (D/S = 0.5) have higher contribution from cistern (up to 23%) whereas the contribution decreases (to 5%) for crowded building (D/S = 4.5). As the percentage contribution from cistern decreases

(with increase in D/S ratio), the contribution from potable water treatment increases. This is due to the increase in demand and insufficient rainwater. The percentage contribution by cistern to the total life cycle GHG emission is similar for combined and separate sewer for

D/S value more than 1. For the D/S value less than 1, separate sewer has higher percentage contribution. This is because separate sewer avoids the treatment of stormwater and has lower total life cycle GHG emissions than combined sewer (Figure 2.5 and Figure SI-2.3 for D/S value less than 1).

The result of this research and the trends observed are only applicable to greenhouse gas emissions. It is further recommended to investigate whether similar trends would be achieved for other impact categories such as eutrophication potential, acidification potential, ecotoxicity and others. The implications of combined versus separate sewers as shown on the left side of Figure 2.5 (D/S<1) may especially be different. With greenhouse gas emissions, impacts occur when the runoff is not released directly to the environment but treated at a wastewater treatment plant. In contrast, eutrophication and ecotoxicity impacts would still occur when this runoff is treated at a wastewater treatment plant but may be lower than impacts coming from direct release of the runoff to surface water bodies.

Future research incorporating water quality parameters in the LCA model is needed to better understand the eutrophication and ecotoxicity tradeoff between treatments at a wastewater treatment plant versus release of runoff in a separate sewer.

The results presented above were obtained when the rainwater tank was sized using mean monthly precipitation. In cases where more rigor is needed the tank can be sized using daily precipitation data typically resulting in smaller tank sizes. We repeated the calculations using daily tank sizing method and observed no change (from monthly tank sizing based results) in how the GHG emissions vary with respect to D/S (Figure SI-2.4).

2.3.3 Comparison of buildings with and without RWH systems (RWH vs BAU

scenarios)

For combined sewer scenarios, the BAU bars are always above the RWH bars for all varied demand to supply ratios suggesting that the RWH implementation always results 33

in lower life cycle GHG emission in all the building configuration variations considered

(Figure 2.5). For the case when occupancy is changing, the distance between the BAU and

RWH bars increases until D/S =1 after which it remains constant (Figure 2.5a). This is because rainwater is used to flush the toilets in RWH which would go to the wastewater treatment plant and require treatment in BAU until D/S = 1. As D/S increases more than one, there is not enough rainwater to flush the toilets and hence the distance between BAU and RWH becomes constant. This finding suggests that in comparing buildings with different occupancy but same roof area, the life cycle emissions would continue to increase with the highest GHG emission savings from implementing RWH systems being achieved when D/S is equal to one or greater. This findings supports the results from Morales-Pinzón et al. (2012) who stated that optimum scale of RWH system is high density developments

(which in our case would be higher D/S values).

Interestingly, a contrary trend is observed when roof area changes while occupancy is kept constant. The highest savings from implementing RWH systems (i.e. the difference in total emissions between RWH and BAU buildings) occur when D/S≤1 (Figure 2.5 b).

At this low D/S value, the roof area and corresponding captured rainwater volume are large.

When some of this rainwater is used in toilet flushing instead of being sent to a combined sewer, there are large emission savings. For D/S > 1, the demand would not be sufficient to fulfill the demand and hence with no stormwater treatment and decrease in cistern size, the GHG emission savings decrease with the increase in D/S.

In communities connected with separate sewer systems, the difference in life cycle emissions between BAU and RWH is minimal (Figure SI-3 in Supporting Information). In addition, in the lowest D/S value modeled, the GHG emissions were even higher for RWH. 34

This is because the separate sewer does not typically include stormwater treatment (and it was assumed not for this study). The GHG emission reduction that is achieved from not sending the storm water runoff to wastewater treatment plant is not realized in separate sewers.

In our particular analysis, buildings with RWH systems have lower total life cycle

GHG emissions compared to BAU scenario indicating that RWH systems are always attractive as compared to municipal water supply system. The only exception is the separate sewer scenario with lower D/S value. Previous research has shown mixed results on this. Bronchi et al. (1999); Crettaz et al. (1999); Ghimire et al. (2014) (pumping option) and Racoviceanu and Karney (2010) have shown that life cycle environmental impacts of

RWH system are higher than BAU scenarios whereas Devkota et al. (2013) and Anand and

Apul (2010) have shown the opposite. Our D/S analysis for BAU and RWH comparisons sheds some light into why mix results have been reported in previous studies. Use of different building life times and different life cycle inventory data sources also likely caused the mixed results observed in the literature. Use of lower life time of the system can be one reason for RWH system to be environmentally taxing. We used a 75 system life time whereas Ghimire et al. (2014) used 50 years and Racoviceanu and Karney (2010) used

20 years. However, as shown in our study, the D/S also affects the results. It is quite possible that previous case studies were done at different D/S values. We recommend that future RWH LCA studies report D/S values so their work can be more easily interpreted in the future.

2.3.4 Per person and per area metrics

Per person and per area metrics have been used to report energy use in buildings

(Harvey, 2013), and potable water savings (Hermann and Schmida, 1999). Devkota et al.

(2015) also has reported life cycle greenhouse gas emissions per person from RWH systems (Devkota et al., 2015). Our study reported that the life cycle GHG emissions

(absolute emission) of implementing harvested rainwater in an average medium office building are 800 kg CO2 equivalent per person and 12 Kg CO2 equivalent per square feet when connected to a combined sewer network (Figure SI – 2.5). The per person greenhouse gas emissions from an average medium office building (600 kg CO2 eq) is found to be comparable with the per person greenhouse gas emission from an educational dormitory

(800 Kg CO2 eq – combined sewer network) (Devkota et al., 2015). Though educational dormitory has higher total life cycle GHG emissions than office building (520,000 kg CO2 eq versus 150,000 kg CO2eq), the per person emissions are comparable since the D/S values are close (D/S=2.25 in medium office and D/S = 3.3 in dormitory). Our study reported that with the increase in D/S value, the per person emission from RWH system decrease (Figure SI – 2.5). At D/S value of 3.3, the life cycle emissions reported in our study was 615 Kg CO2 eq (combined sewer) which is quite comparable to the value reported for the dormitory (Devkota et al., 2015).

The per person and per unit area emissions for the different D/S values are shown in Figures SI – 2.5,SI – 2.6, SI-2.7 and SI-2.8. Unlike total emissions that depend on whether the roof area or the occupancy is changed, the emissions normalized to per person and per unit area result in the same value for a given D/S regardless of whether the roof

area or the occupancy is changed. These findings suggest that emissions per person or per unit area analyzed as a function of D/S may be a more universal and robust analysis and characterization of RWH systems than the analysis of total emissions as a function of D/S.

The per person GHG emission of implementing RWH system for the various D/S ratios varied between 500 to 600 Kg CO2 eq when the building was connected to a separate sewer system and 600 to 1,000 Kg CO2 eq when the building is connected to a combined sewer system. The per unit area emissions vary from 25 to 275 kg CO2 eq per square meter for separate sewer and from emissions vary from 50 to 275 kg CO2 eq per square feet for combined sewer. Even though the D/S range is the same (0.5 to 4.5) the variation is much higher for per unit area (about 5 fold) emissions than for per person (less than 2 fold) emissions. This finding implies that the emissions per unit area are more sensitive to the D/S parameter than the emissions per person. In other words, changing the roof area or the occupancy will cause a greater response in the emission per area metric than the emission per person metric.

Interestingly the per person and per unit area emission trends are very similar to the absolute emission trends discussed in section 3.2. The per person emissions trends (Figure

SI – 2.5 a and b & Figure SI – 2.6 a and b) were found to be the exact same as the total greenhouse gas emission trend for RWH systems with varying roof area (Figure 2.5 b).

Similarly, the per unit area emissions trends (Figure SI – 2.7 a and b & Figure SI – 2.8 a and b) were found to be the exact same as the total greenhouse gas emission trend for RWH system with varying occupancy (Figure 2.5 a). Therefore, the interpretations of the emission trends (both totals and the difference between BAU and RWH) as a function of

D/S discussed in section 3.2 apply to emission per person and emission per unit area results as well.

2.3.5 On which type of building should one implement RWH?

The findings from the LCA results can be used in two different ways as shown in

Figure 2.6. Sometimes the goal is to find the building design that would have the least GHG emissions from RWH systems (Type A situation). In other cases, the goal may be to determine how much GHG emission savings can be achieved by implementing a RWH system (Type B situation).

In comparing buildings with varying occupancy (Type A situation) the building with the lowest GHG emissions will be the one with lowest occupancy irrespective of the sewer network. In Type B situation, there is little difference between RWH and BAU scenarios in separate sewer cases. However, in combined sewer systems, building with higher occupancy resulting in D/S ≥1 will maximize the GHG savings compared to BAU scenario.

In comparing buildings with varying roof area, choosing a large roof area resulting in D/S≤1 reduces both the GHG emissions and the GHG savings (in RWH compared to

BAU scenario). However, if it is a combined sewer, then the building with roof area corresponding to D/S =1 should be selected to have the least GHG emissions.

In this study we proposed D/S as an important metric that should be reported in

LCA RWH systems. As part of this work, roof area and occupancy were changed one at a time. This one at a time variation shows that changes in occupancy have a greater impact on life cycle GHG emissions than changes in roof area (compare figures 2.5a and 2.5b).

Therefore, while more rainwater can be collected from buildings with larger roof area 38

which would reduce the environmental impact of RWH system, the building with smaller number of occupants would further reduce the impact. This result validates the findings from Morales-Pinzón et al. (2012) where they stated higher rainwater supplies (in our study we represent by D/S < 1) produce higher energy use thus emitting low GHG emission per functional unit. In future work, it would be interesting to advance the D/S approach by studying how the emissions change when the roof area and occupancy are varied simultaneously. We also recommend further advancing the D/S approach using per unit area and per person metrics for other types of buildings.

Figure 2. 6 Decision matrix guide to implement RWH in an office building.

It should be noted that while these design guidelines have been derived from simulating medium size office buildings, the life cycle GHG emission trends would be the same for small and large office buildings as well (Figure SI – 2.9 and Figure SI – 2.10).

The primary difference among different size buildings is the relevant D/S ratios. For example, for the medium size office building, the lowest D/S modeled was 0.5 which is

not very likely to occur in real life for this size of a building. The D/S of the average medium office (D/S = 2.25) is derived from the occupancy load (200 ft2 per occupant) used by Deru et al. (2011). The international building code specifies half of this area (100 ft2 per occupant) as the minimum required for egress safety (IBC, 2006)) resulting in the highest

D/S ratio (D/S= 4.5) modeled for the medium sized office. As such, the lowest D/S value modeled for the medium sized office (D/S = 0.5) corresponds to 8 times fewer occupants or 8 times greater roof area than what the international building code specifies. These types of buildings are likely not to be a large fraction of the existing and future building stock.

However, D/S = 0.5 would be a relevant occupancy or roof area for small office buildings that would have D/S = 0.75 based on Deru et al’s occupancy load (Deru et al., 2011) and

D/S = 1.6 based on the international building code (IBC, 2006). In the case of large office buildings, the D/S value based on Deru et al’s occupancy load is 8.4. In large office buildings, low D/S values would therefore not be expected and even D/S = 1 may not be a commonly found design.

Finally, it should be noted that these results apply specifically to the system boundary modeled in this study. The system boundary in this study was developed specifically for RWH system analysis and does not include the building shell. The LCA of a whole building would include the shell and would explicitly model the GHG emissions resulting from higher amounts of construction materials used in buildings that have larger roof areas. In addition, a whole building LCA would likely show that the larger roof would require larger heating, cooling, and lighting operational energy and associated GHG emissions. In comparing different building designs, would variations in occupancy still cause greater impacts than variations in roof area when the LCA is conducted for the whole 41

building? Research is needed to answer this question and understand to what extent the

RWH based results apply to whole building LCAs.

2.4 Conclusions

Several conclusions can be drawn from this study of demand to supply ration on

RWH system GHG emissions:

1. Effect of combined vs separate sewer:

The life cycle GHG emissions of using RWH system in an office building depends

whether the building is connected to a combined sewer or separate sewer. When D/S<1,

the emissions from RWH systems are higher in combined sewer systems than in

separate sewer systems. When D/S≥1, combined and separate sewers behave the same

for RWH systems.

2. Effect of D/S on GHG emissions

The GHG emission increased linearly with increasing occupancy for D/S > 1 cases

when the building footprint was constant. The increase in emission was due to the

increase in potable water and wastewater treatment. However if the occupancy was

further reduced (D/S < 1), the cistern size and its associated GHG emissions from

manufacturing was reduced. On the other hand, keeping occupancy constant required

higher treatment of stormwater (for D/S < 1). The GHG emission increased linearly

with increasing occupancy for D/S > 1 cases when the building footprint was constant. 42

The increase in emission was due to the increase in potable water and wastewater treatment. However if the occupancy was further reduced (D/S < 1), the cistern size and its associated GHG emissions from manufacturing was reduced. On the other hand, keeping occupancy constant required higher treatment of stormwater (for D/S < 1). The

GHG emission increased linearly with increasing occupancy for D/S > 1 cases when the building footprint was constant. The increase in D/S value decreased the stormwater treatment thereby decreasing the GHG emissions for D/S < 1. The emission reductions from the cistern (though small) and the emission increase from higher use of municipal water supply in the opposite directions not resulting in a linear change in impacts as a function of D/S. Though the relative impacts from the cistern was smaller (potable water treatment also not being significantly high as compared to wastewater treatment), the nature of the curve was somewhat guided by the change in impacts from the cistern.

3. Design selection based on absolute life cycle emissions

If decision to implement RWH is based on absolute life cycle emissions (standalone

RWH system comparisons) and different building configurations are being considered for constant roof variable occupancy, the building with least GHG emissions will be the one with least possible occupancy. For building configurations having varying roof area (keeping constant occupancy) and separate sewer scenario, emissions do not vary much with varying roof but a large roof area resulting in D/S≤1 has lowest emissions whereas emissions are very high when roof area is large resulting inD/S<1 for combined sewer scenario. In such case, choose a roof area resulting in D/S=1 for lowest emissions. 43

4. Design selection based on relative emissions from RWH (comparison with BAU)

If decision to implement RWH is based on relative emissions (comparison of RWH

system with BAU) and different building configurations are being considered for

constant roof variable occupancy, a large occupancy has highest emission savings. For

building configurations having varying roof area (keeping constant occupancy) large

roof area has the highest savings.

5. Water savings

Water savings from implementing RWH system changes as a function of D/S ratio.

Maximum savings are reported at D/S equal to one and the savings increases for the

building configuration when the supply is more than demand.

6. Per person and per square meter metrics

Per person and per area based metrics have been introduced in RWH system to report

life cycle GHG emissions. Both the trend and magnitude of per person as well as per

area impacts are independent of varying occupancy or roof area and can be used as a

universal metric to estimate life cycle GHG emissions.

Acknowledgements

This study was funded by National Science Foundation’s Environmental Sustainability

(grant # 1236660).

2.5 References

Aladenola, O. O., & Adeboye, O. B., 2010. Assessing the potential for rainwater harvesting. Water Resource Management, 24(10), 2129-2137.

Alliance for Water Efficiency. Available online from: http://www.allianceforwaterefficiency.org/office_buildings.aspx

Anand, C., Apul, D.S., 2010., Economic and environmental analysis of standard, high efficiency, rainwater flushed, and composting toilets. Journal of Environment Management 92, 419-428.

Angrill, S., Farreny, R., Gasol, C. M., Gabarrell, X., Viñolas, B., Josa, A., and Rieradevall, J., 2012. Environmental analysis of rainwater harvesting infrastructures in diffuse and compact urban models of Mediterranean climate. International Journal of Life Cycle Assessment, 17(1), 25-42.

Blengini, G. A., & Di Carlo, T., 2010. The changing role of life cycle phases, subsystems and materials in the LCA of low energy buildings. Energy and Buildings, 42(6), 869-880.

Bronchi, V., Jolliet, O. and Crettaz, P., 1999. Life Cycle Assessment of Rainwater use for Domestic Needs. Institute of soil and water management - Ecosystem management, CH-1015 Lausanne EPFL, Switzerland.

California Energy Commission, 2005. California’s Water- Energy Relationship. Available online from: http://www.energy.ca.gov/2005publications/CEC-700- 2005-011/CEC-700-2005-011-SF.PDF (accessed 1.14.2014)

Coombes, P.J., Kuczera, G. and Kalma, J.D., 2002. Economic, water quantity and quality results from a house with a rainwater tank in the inner city. Proceedings of the 27th Hydrology and Water Resources Conference. Melbourne, Australia.

CostLab, 2015. Available online at: http://whitestoneresearch.com/Costlab.

Crettaz, P., Jolliet, O., Cuanillon, J. M., Orlando, S., 1999. Life cycle assessment of drinking water and rain water for toilets flushing. J. Water Serv. Res. Tec. 48, 73- 83.

Deru, M., Field, K., Studer, D., Benne, K., Griffith, B., Torcellini, P., and Crawley, D. 2011. US Department of Energy commercial reference building models of the national building stock.

Devkota, J., Hannah, S., and Apul, D., 2015 Life Cycle Based Evaluation of Harvested Rainwater Use in Toilets and for Irrigation. Submitted to Journal of Clean Production, 95, 311-321.

Devkota, J., Schlachter, H., Anand, C., Phillips, R., and Apul, D., 2013. Development and application of EEAST: A life cycle based model for use of harvested rainwater and composting toilets in buildings. Journal of Environment Management 130, 397-404. Ecoinvent, 2010. Ecoinvent v2.2 database. Swiss Centre for Life Cycle Inventories, Dübendorf, Switzerland.

Farreny, R., Gabarrell, X., & Rieradevall, J., 2011. Cost-efficiency of rainwater harvesting strategies in dense Mediterranean neighbourhoods. Resources, conservation and recycling, 55(7), 686-694.

Fewkes, A., 1998. The use of rainwater for WC flushing: the field testing of a collection system. Building and Environment 34(6), 765-772.

Furumai, H., 2008. Rainwater and reclaimed wastewater for sustainable urban water use. Phys. Chem. Earth - Parts A/B/C; 33(5), 340-346.

Georgia Rainwater Harvesting Guidelines, 2009. Available online from: http://www.conservewatergeorgia.net/resources/Georgia_Rainwater_Harvesting_ Guidelines.pdf Accessed Dec 2014.

GHG protocol. Available online from: http://www.ghgprotocol.org/files/ghgp/public/FAQ.pdf Accessed May 14 2014.

Ghimire, S. R., Johnston, J. M., Ingwersen, W. W., & Hawkins, T. R. 2014. Life Cycle Assessment of Domestic and Agricultural Rainwater Harvesting Systems. Environment Science and Technology, 48(7), 4069-4077.

Ghisi, E., & Ferreira, D. F., 2007. Potential for potable water savings by using rainwater and greywater in a multi-storey residential building in southern Brazil. Building and Environment, 42(7), 2512-2522.

Ghisi, E., Tavares, D. D. F., and Rocha, V. L., 2009. Rainwater harvesting in petrol stations in Brasilia: Potential for potable water savings and investment feasibility analysis. Resource Conservation and Recycling,. 54(2), 79-85.

Harvey, L. D. (2013). Recent advances in sustainable buildings: review of the energy and cost performance of the state-of-the-art best practices from around the world. Annual Review of Environment and Resources, 38, 281-309.

Herrmann, T., Schimda, U., 2000. Rainwater utilization in Germany: efficiency, dimensioning, hydraulic and environmental aspects, Urban Water Journal 1, 307- 316.

IBC, I. (2006). International building code. International Code Council, Inc.(formerly BOCA, ICBO and SBCCI), 4051, 60478-5795.

Jones, M. P., and Hunt, W. F., 2010. Performance of rainwater harvesting systems in the southeastern United States. Resources, Conservation and Recycling, 54(10), 623- 629.

Kirk, S. J., & Dell'Isola, A. J., 1995. Life cycle costing for design professionals (pp. 1-3). New York: McGraw-Hill.

Krishna HJ. The Texas manual on rainwater harvesting, edition 3, Texas water development board 2005 http://www.ecy.wa.gov/programs/wr/hq/pdf/texas_rw_harvestmanual_3rdedition. pdf (accessed 1.14.2014)

Li, X. Y., and Gong, J. D., 2002. Effects of different ridge: furrow ratios and supplemental irrigation on crop production in ridge and furrow rainfall harvesting system with mulches. Agricultural Water Management, 54(3), 243-254.

Mayer, P.W., William, B.D., 1999. Residential End Uses of Water. American Water Works Association Research Foundation (AWWA), Denver.

Mehta, M., 2009. Water Efficiency Saves Energy: Reducing Global Warming Pollution through Water Use Strategies. Natural Resource Defense Council. Available online from http://www.nrdc.org/water/files/energywater.pdf (accessed 1.14.2014)

Morales-Pinzón, T., Lurueña, R., Rieradevall, J., Gasol, C. M., & Gabarrell, X., 2012. Financial feasibility and environmental analysis of potential rainwater harvesting systems: a case study in Spain. Resources, Conservation and Recycling, 69, 130- 140.

National Green Building Adoption Index, 2014. Available online from: http://www.cbre.com/o/international/AssetLibrary/Green-Building-Adoption- Index-2014_Final.pdf

Palla, A., Gnecco, I., & Lanza, L. G., 2011. Non-dimensional design parameters and performance assessment of rainwater harvesting systems. Journal of Hydrology, 401(1), 65-76.

PE International, 2014. Gabi database. http://www.pe-international.com. 47

Peuportier, B. L. P., 2001. Life cycle assessment applied to the comparative evaluation of single family houses in the French context. Energy and buildings, 33(5), 443-450.

Racoviceanu, A. I., & Karney, B. W., 2010. Life-cycle perspective on residential water conservation strategies. Journal of Infrastructure Systems,16 (1), 40-49.

Report to congress: Impacts and control of CSOs and SSOs, National Pollution discharge elimination system, USEPA, 2004. Available online from http://www.epa.gov/npdes/pubs/csossoRTC2004_executive_summary.pdf

Saving Water in Office Building, EPA, 2012. Available online from: http://www.epa.gov/watersense/commercial/docs/factsheets/offices_fact_sheet_50 8.pdf. Last accessed July 2015. Steffen, J., Jensen, M., Pomeroy, C. A., & Burian, S. J., 2013. Water Supply and Stormwater Management Benefits of Residential Rainwater Harvesting in US Cities. JAWRA Journal of the American Water Resources Association, 49(4), 810-824.

Stout, D. T., Walsh, T. C., & Burian, S. J. (2015). Ecosystem services from rainwater harvesting in India. Urban Water Journal, (ahead-of-print), 1-13.

United States Environmental Protection Agency (USEPA). 2002. The clean water and drinking water infrastructure gap analysis. EPA 816-R-02-020. Available from http://water.epa.gov/aboutow/ogwdw/upload/2005_02_03_gapreport.pdf (accessed 1.14.2014)

US Census, 2000. Commercial Buildings -- Number and Size, by Principal Activity. Available online from: http://www.allcountries.org/uscensus/1227_commercial_buildings_number_and_s ize_by.html

USEPA, Water Conservation Plan Guidelines. Available online from: http://www.epa.gov/WaterSense/docs/app_b508.pdf. Accessed May 2015.

Vargas-Parra M V, Villalba G, and Gabarrell X, (2013). Applying exergy analysis to rainwater harvesting systems to assess resource efficiency.Resources, Conservation and Recycling, 72, 50-59.

Vickers A. 2001. A Handbook of water use and conservation. Water Flow Press. Amherst: MA.

Villarreal, E. L., and Dixon, A., 2005. Analysis of a rainwater collection system for domestic water supply in Ringdansen, Norrköping, Sweden. Building and Environment, 40(9), 1174-1184. 48

Walsh, T. C., Pomeroy, C. A., & Burian, S. J. (2014). Hydrologic modeling analysis of a passive, residential rainwater harvesting program in an urbanized, semi-arid watershed. Journal of Hydrology, 508, 240-253.

2.6 Supporting Information: Introducing Demand to Supply Ratio as a New

Metric for Understanding Life Cycle Greenhouse Gas (GHG) Emissions from

Rainwater Harvesting Systems.

1. Water supply calculations:

A. Using mean monthly precipitation:

Vr = Ravg x ƞ x A (2.1)

Where,

Vr = Volume of rainfall available for capture in one month (gal)

Ravg = Average monthly precipitation (in)

Ƞ = Rainwater catchment system efficiency (default value = 85 %).

A = Building roof area (ft2).

B. Using long term daily precipitation and Yield After Spillage approach:

Yt = MIN ( Dt , Vt-1 + It ) (2.2)

Yt = Volume of rainwater supplied to urban water demands (gallons)

Dt = Daily water demand (gallons).

Vt-1 = volume of rainwater in the tank at the end of previous time step t (gallons)

It = inflow or roof runoff (gallons)

2. Pump Energy calculation:

It is assumed that the tank is placed on the ground. Therefore a pump is required to transport the rainwater from the cistern to the toilets in respective floors. Energy delivered to pump was estimated using following equation:

P = ϒ x Q x (he + hp) x [1+α]/η (2.3)

Then the pump’s life cycle energy requirements are calculated as:

E = P x 365 [days/year] x 24 [hrs/day] x 0.001 [kW/W] x 75 years (2.4)

Where,

P = Power delivered to Pump (W)

E = Annual energy required by Pump (kWh)

Q = Flow rate (m3/sec). h = he + hp = sum of elevation head and pressure head provided by the pump (m).

η = combined mechanical and electrical efficiency of the pump (assumed 65 %; Cengel and Cimbala, 2005).

ϒ = Specific weight of water (N/m3)

α = Percentage of energy lost due to friction (assumed 0.3%; Cheng, 2002).

Pressure head provided by the pump (hp) is set equal to the pressure provided by the city

(assumed 35.2m of water or 50 psi). Elevation head provided by pump (he) is set equal to the building height.

3. Dual piping length calculation

Length of dual piping required for conveying harvested rainwater to toilets is calculated by assuming that one primary pipe runs all the way to the top floor and that each toilet is approximately 3ft away from this pipe. In addition, the horizontal length of pipe from the cistern to the main pipe is assumed to be half of the square root of the roof area.

LP = (3ft x No. of Toilets) + (H – H / stories) + [(Roof area)^0.5] / 2 (2.5)

Where,

LP = Length of piping required (ft).

H = Height of the proposed building (ft).

푇 = Number of toilets per floor. 푓

4. Tank sizing a. Using Monthly Precipitation

Rainwater cistern is designed based on a comparison of rainwater water supply and building water demand on a monthly basis. If the volume of water required to flush toilets and irrigate is more than the available rainwater supply, then the cistern size is based on the supply and vice versa:

Vc = V (If V

Vc = Vr (If V>Vr) (2.7)

Where,

Vc = Volume of cistern (gallons)

V = Total volume of water required to flush toilets in one month (gallons)

Vr = Volume of rain available for capture in one month (gallons)

b. Using Daily Precipitation i. Methodology

Alternatively to mean monthly approach to size the rainwater tank, YAS approach and long term daily (13 years) precipitation of Toledo, Ohio were used to size the rainwater tank to compare the variation of life cycle GHG emissions to the D/S ratio values for an average medium office building. This method was adopted from Mitchell (2007).

The efficiency of the rainwater tank is measured in terms of volumetric reliability which can be expressed as:

∑ 푌푡 Vr = volumetric reliability = ……………………………………………...….. (2.8) ∑ 퐷푡

Where,

Yt = Volume of rainwater supplied to urban water demands (gallons)

Dt = Daily water demand (gallons).

In the YAS approach, Yt and Dt are calculated using:

Yt = MIN ( Dt , Vt-1 + It ) …………………………………………………………. (2.9)

Vt = MIN (Vt-1 + It – Yt , C – Yt ) ………………………………………………… (2.10)

Where,

Vt = volume of rainwater in the tank at the end of time step t (gallons) 53

Vt-1 = volume of rainwater in the tank at the end of previous time step t (gallons)

It = inflow or roof runoff (gallons)

C = Capacity of the rainwater tank (gallons)

Daily spillage (St) is calculated from the daily mass balance equation (equation 7)

Vt = Vt-1 + It + Pt – Et – St – Lt - Yt ……………………………………..……….. .(2.11)

Where:

Pt = incident precipitation on received by the tank (gallons)

Et = evaporation (gallons)

St = amount of spillage due overflow from tank (gallons)

Lt = seepage or leakage (gallons)

Evaporation (Et), Incident precipitation (Pt) and Leakage (Lt) are very small and are is not

considered in this study.

A series of rainwater storage tank sizes were assessed and their corresponding

volumetric reliabilities were calculated using equation 4. The size of the tank was

considered optimum when any increase in the tank size does not have any significant

change in its volumetric reliability.

ii. Results

The size of rainwater tank with the daily precipitation of Toledo and YAS approach

was found to be 12,500 gallons which is significantly smaller than the size obtained from

mean monthly sizing approach (23,120 gallons). The variation of life cycle GHG emission 54

with respect to demand to supply ratio is presented in Figure SI-3. Even though the rainwater tank size was smaller, the results from daily approach did not showed significant difference and the trend was similar to that obtained from mean monthly approach. This was due to small contribution from manufacturing the rainwater cistern to the life cycle

GHG emissions. Wastewater treatment and potable water treatment being highest contribution to the life cycle GHG emissions, any change in cistern size was counterbalanced. Therefore results from mean monthly approach is only presented in the main body of this paper.

Reference:

Cengel, Y., Cimbala, J., 2005. Fluid Mechanics: Fundamentals and Applications.

McGraw Hill Higher Education, Boston.

Cheng CL., Study of the inter-relationship between water use and energy conservation for a building. Energy Build. 2002; 34(3), 261-66.

Mitchell, V. G. (2007). How important is the selection of computational analysis method to the accuracy of rainwater tank behaviour modelling? Hydrological processes, 21(21),

2850-2861.

Table SI - 2. 1 Inventory of BAU and RWH components for an average medium office building

Life Process

span description Quantity (years)

Phase

Scenario Combi Separa Item ned te Units 75 PVC granulate suspension - Pipe SPVC 21.32 21.32 kg 35 vitrified clay Toilets 223.17 223.17 kg

Toilet flush tank 35 vitrified clay with accessories 100.70 100.70 kg 75 PVC granulate

suspension - Sewer drain pipe SPVC 29.39 29.39 kg Construction Transportation from Truck – plant to installation gasoline power 3177.2 3177.2 point – 100 miles 8 8 kg-km 75 PVC granulate

Business as Usual Business as Bends valves and suspension - tees SPVC 12.97 12.97 kg Toilet replacement 35 vitrified clay

(every 35 years) 223.17 223.17 kg Potable water/ Municipal 197,59 197,59 month water at tap 0 0 ltr Municipal Operation Wastewater sewage from treatment/ month residence 285 198 m3 75 Galvanized steel sheet at 3631.2 3631.2

Storage tank plant 2 2 kg 20 Electric water Pumps pump – 0.5 hp 10.70 10.70 kg 5 Floating tank filter 3.63 3.63 kg 75 PVC granulate

Construction suspension - Pipes (dual piping) SPVC 42.64 42.64 kg Rainwater Harvesting 75 PVC granulate suspension - Overflow drain SPVC 0.96 0.96 kg

75 PVC granulate Vent for RWH tank suspension - (PVC) SPVC 0.96 0.96 kg 35 vitrified clay Toilets 223.17 223.17 kg Toilet flush tank 35 vitrified clay with accessories 100.70 100.70 kg 75 PVC granulate suspension - Sewer drain pipe SPVC 29.39 29.39 kg 75 PVC granulate Bends valves and suspension - tees SPVC 0.65 0.65 kg Transportation from Truck – plant to installation gasoline power 37797 37797 point – 100 miles 4.15 4.15 kg-km 75 Concrete mix at plant (Density: 2,380 kg/m3. Ingredients: Cement 300 kg, Water 190 kg, Gravel Concrete Pad 1,890 kg) 1.57 1.57 m3 Potable water/ Municipal 108,67 108,67 month water 5 5 ltr Municipal Wastewater sewage from treatment/ month residence 205 198 m3 Electricity at Energy use by grid - medium pump/ month voltage 81 81 MJ

Operation Pump replacement 20 Electric water (every 20 years) pump – 0.5 hp 10.70 10.70 kg Toilet replacement 35 vitrified clay (every 35 years) 223.17 223.17 kg Filter replacement 5 (every 5 years) 3.63 3.63 kg

Table SI - 2. 2 Inventory of RWH components for all the roof area variations when occupancy was kept constant at 268 – When building is served by a combined sewer network. Neglecting the minor components.

RWH design parameters Quantities

/mon) /mon)

er replacement

3 3

S.N. Roof Area (sq ft) Roof area to occupancy (sq ft) Demand (gal/mon) Supply (gal/mon) Deamand/ Supply Cistern Volume (gal) Cistern (kg) Concrete Pad (m3) Filt (lb) Energy use by pump (kWh) Potable water required (ltr/mon) Waste water (m Stormwater (m 1 8070 30.1 52198 10437 5.00 10437 3137.7 0.9 3.63 9.99 158081 198 0 2 8491 31.7 52198 10982 4.75 10982 3218.5 1.0 3.63 10.51 156019 198 0 3 8965 33.4 52198 11595 4.50 11595 3307.1 1.0 3.63 11.10 153699 198 0 4 9495 35.4 52198 12280 4.25 12280 3403.5 1.1 3.63 11.76 151104 198 0 5 10080 37.6 52198 13037 4.00 13037 3506.8 1.2 3.63 12.48 148240 198 0 6 10749 40.1 52198 13902 3.75 13902 3621.3 1.2 3.63 13.31 144965 198 0 7 11517 43.0 52198 14895 3.50 14895 3748.4 1.3 3.63 14.26 141205 198 0 8 12403 46.3 52198 16041 3.25 16041 3889.9 1.4 3.63 15.36 136867 198 0 9 13436 50.1 52198 17377 3.00 17377 4048.7 1.5 3.63 16.64 131810 198 0 10 14658 54.7 52198 18958 2.75 18958 4228.8 1.6 3.63 18.15 125827 198 0 11 16124 60.1 52198 20854 2.50 20854 4435.2 1.8 3.63 19.96 118650 198 0 12 17915 66.8 52198 23170 2.25 23170 4675.0 2.0 3.63 22.18 109881 198 0 13 20155 75.2 52198 26067 2.00 26067 4958.7 2.2 3.63 24.95 98914 198 0 14 23034 85.9 52198 29791 1.75 29791 5301.0 2.5 3.63 28.52 84819 198 0 15 26873 100.2 52198 34756 1.50 34756 5725.8 2.9 3.63 33.27 66024 198 0

S.N. RWH design parameters Quantities

(sq ft)

/mon) /mon)

3 3

Roof Area (sq ft) Roof area to occupancy Demand (gal/mon) Supply (gal/mon) Deamand/ Supply Cistern Volume (gal) Cistern (kg) Concrete Pad (m3) Filter replacement (lb) Energy use by pump (kWh) Potable water required (ltr/mon) Waste water (m Stormwater (m 16 32248 120.3 52198 41708 1.25 41708 6272.3 3.4 3.63 39.93 39709 198 0 17 40310 150.3 52198 52135 1.00 52135 7012.7 4.2 3.63 49.91 239 198 0 18 53747 200.4 52198 69514 0.75 52198 7016.9 4.2 3.63 49.97 0 198 66 19 80620 300.7 52198 104270 0.50 52198 7016.9 4.2 3.63 49.97 0 198 197 20 161241 601.3 52198 208541 0.25 52198 7016.9 4.2 3.63 49.97 0 198 592

Table SI - 2. 3 Inventory of RWH components for all the occupancy variations when building roof area was kept constant at 53,628 sq. ft. – When building is served by a combined sewer network. Neglecting the minor components

RWH design parameters Quantities

/mon) /mon)

3 3

S.N. Occupancy Roof area to occupancy (sq ft) Demand (gal/mon) Supply (gal/mon) Deamand/ Supply Cistern Volume (gal) Cistern (kg) Concrete Pad (m3) Filter replacement (lb) Energy use by pump (kWh) Potable water required (ltr/mon) Waste water (m Stormwater (m 1 594 30.1 115632 23120 5.00 23120 4669.9 2.0 3.63 22.13 350195 438 0 2 564 31.7 109850 23120 4.75 23120 4669.9 2.0 3.63 22.13 328310 416 0 3 535 33.4 104069 23120 4.50 23120 4669.9 2.0 3.63 22.13 306424 394 0 4 505 35.4 98287 23120 4.25 23120 4669.9 2.0 3.63 22.13 284538 372 0 5 475 37.6 92506 23120 4.00 23120 4669.9 2.0 3.63 22.13 262653 350 0 6 446 40.1 86724 23120 3.75 23120 4580.7 2.0 3.63 22.13 240767 328 0 7 416 43.0 80942 23120 3.50 23120 4669.9 2.0 3.63 22.13 218881 306 0 8 386 46.3 75161 23120 3.25 23120 4669.9 2.0 3.63 22.13 196996 284 0 9 356 50.2 69379 23120 3.00 23120 4669.9 2.0 3.63 22.13 175110 263 0 10 327 54.7 63598 23120 2.75 23120 4669.9 2.0 3.63 22.13 153224 241 0 11 297 60.2 57816 23120 2.50 23120 4669.9 2.0 3.63 22.13 131339 219 0 12 267 66.9 52034 23120 2.25 23120 4669.9 2.0 3.63 22.13 109453 197 0 13 238 75.2 46253 23120 2.00 23120 4669.9 2.0 3.63 22.13 87567 175 0 14 208 86.0 40471 23120 1.75 23120 4669.9 2.0 3.63 22.13 65682 153 0 15 178 100.3 34690 23120 1.50 23120 4669.9 2.0 3.63 22.13 43796 131 0 60

S.N. RWH design parameters Quantities

)

/mon) /mon)

3 3

Occupancy Roof area to occupancy (sq ft Demand (gal/mon) Supply (gal/mon) Deamand/ Supply Cistern Volume (gal) Cistern (kg) Concrete Pad (m3) Filter replacement (lb) Energy use by pump (kWh) Potable water required (ltr/mon) Waste water (m Stormwater (m 16 149 120.4 28908 23120 1.25 23120 4669.9 2.0 3.63 22.13 21910 109 0 17 119 150.5 23126 23120 1.00 23120 4669.9 2.0 3.63 22.13 25 88 0 18 89 200.6 17345 23120 0.75 17345 4044.9 1.5 3.63 16.60 0 66 22 19 59 300.9 11563 23120 0.50 11563 3302.6 1.0 3.63 11.07 0 44 44 20 30 601.9 5782 23120 0.25 5782 2335.3 0.6 3.63 5.53 0 22 66

Table SI - 2. 4 Inventory of RWH components for all the roof area variations when occupancy was kept constant at 268 – When building is served by a separate sewer network. Neglecting the minor components

RWH design parameters Quantities

/mon)

ergy use by

S.N. Roof Area (sq ft) Roof area to occupancy (sq ft) Demand (gal/mon) Supply (gal/mon) Deamand/ Supply Cistern Volume (gal) Cistern (kg) Concrete Pad (m3) Filter replacement (lb) En pump (kWh) Potable water required (ltr/mon) Waste water (m 1 8070 30.1 52198 10437 5.00 10437 3137.7 0.9 3.63 9.99 158081 198 2 8491 31.7 52198 10982 4.75 10982 3218.5 1.0 3.63 10.51 156019 198 3 8965 33.4 52198 11595 4.50 11595 3307.1 1.0 3.63 11.10 153699 198 4 9495 35.4 52198 12280 4.25 12280 3403.5 1.1 3.63 11.76 151104 198 5 10080 37.6 52198 13037 4.00 13037 3506.8 1.2 3.63 12.48 148240 198 6 10749 40.1 52198 13902 3.75 13902 3621.3 1.2 3.63 13.31 144965 198 7 11517 43.0 52198 14895 3.50 14895 3748.4 1.3 3.63 14.26 141205 198 8 12403 46.3 52198 16041 3.25 16041 3889.9 1.4 3.63 15.36 136867 198 9 13436 50.1 52198 17377 3.00 17377 4048.7 1.5 3.63 16.64 131810 198 10 14658 54.7 52198 18958 2.75 18958 4228.8 1.6 3.63 18.15 125827 198 11 16124 60.1 52198 20854 2.50 20854 4435.2 1.8 3.63 19.96 118650 198 12 17915 66.8 52198 23170 2.25 23170 4675.0 2.0 3.63 22.18 109881 198 13 20155 75.2 52198 26067 2.00 26067 4958.7 2.2 3.63 24.95 98914 198 14 23034 85.9 52198 29791 1.75 29791 5301.0 2.5 3.63 28.52 84819 198 15 26873 100.2 52198 34756 1.50 34756 5725.8 2.9 3.63 33.27 66024 198

S.N. RWH design parameters Quantities

/mon)

Roof Area (sq ft) Roof area to occupancy (sq ft) Demand (gal/mon) Supply (gal/mon) Deamand/ Supply Cistern Volum (gal) Cistern (kg) Concrete Pad (m3) Filter replacement (lb) Energy use by pump (kWh) Potable water required (ltr/mon) Waste water (m 16 32248 120.3 52198 41708 1.25 41708 6272.3 3.4 3.63 39.93 39709 198 17 40310 150.3 52198 52135 1.00 52135 7012.7 4.2 3.63 49.91 239 198 18 53747 200.4 52198 69514 0.75 52198 7016.9 4.2 3.63 49.97 0 198 19 80620 300.7 52198 104270 0.50 52198 7016.9 4.2 3.63 49.97 0 198 20 161241 601.3 52198 208541 0.25 52198 7016.9 4.2 3.63 49.97 0 198

Table SI - 2. 5 Inventory of RWH components for all the occupancy variations when building roof area was kept constant at 53,628 sq. ft. – When building is served by a separate sewer network. Neglecting the minor components

RWH design parameters Quantities

/mon)

aste wateraste

S.N. Occypancy Roof area to occupancy (sq ft) Demand (gal/mon) Supply (gal/mon) Deamand/ Supply Cistern Volume (gal) Cistern (kg) Concrete Pad (m3) Filter replacement (lb) Energy use by pump (kWh/mon) Potable water required (ltr/mon) W (m 1 594 30.1 115632 23120 5.00 23120 4669.9 2.0 3.63 22.13 350195 438 2 564 31.7 109850 23120 4.75 23120 4669.9 2.0 3.63 22.13 328310 416 3 535 33.4 104069 23120 4.50 23120 4669.9 2.0 3.63 22.13 306424 394 4 505 35.4 98287 23120 4.25 23120 4669.9 2.0 3.63 22.13 284538 372 5 475 37.6 92506 23120 4.00 23120 4669.9 2.0 3.63 22.13 262653 350 6 446 40.1 86724 23120 3.75 23120 4669.9 2.0 3.63 22.13 240767 328 7 416 43.0 80942 23120 3.50 23120 4669.9 2.0 3.63 22.13 218881 306 8 386 46.3 75161 23120 3.25 23120 4669.9 2.0 3.63 22.13 196996 284 9 356 50.2 69379 23120 3.00 23120 4669.9 2.0 3.63 22.13 175110 263 10 327 54.7 63598 23120 2.75 23120 4669.9 2.0 3.63 22.13 153224 241 11 297 60.2 57816 23120 2.50 23120 4669.9 2.0 3.63 22.13 131339 219 12 267 66.9 52034 23120 2.25 23120 4669.9 2.0 3.63 22.13 109453 197 13 238 75.2 46253 23120 2.00 23120 4669.9 2.0 3.63 22.13 87567 175 14 208 86.0 40471 23120 1.75 23120 4669.9 2.0 3.63 22.13 65682 153 15 178 100.3 34690 23120 1.50 23120 4669.9 2.0 3.63 22.13 43796 131 64

S.N. RWH design parameters Quantities

on)

/mon)

Occypancy Roof area to occupancy (sq ft) Demand (gal/mon) Supply (gal/mon) Deamand/ Supply Cistern Volume (gal) Cistern (kg) Concrete Pad (m3) Filter replacement (lb) Energy use by pump (kWh/m Potable water required (ltr/mon) Waste water (m 16 149 120.4 28908 23120 1.25 23120 4669.9 2.0 3.63 22.13 21910 109 17 119 150.5 23126 23120 1.00 23120 4669.9 2.0 3.63 22.13 25 88 18 89 200.6 17345 23120 0.75 17345 4044.9 1.5 3.63 16.60 0 66 19 59 300.9 11563 23120 0.50 11563 3302.6 1.0 3.63 11.07 0 44 20 30 601.9 5782 23120 0.25 5782 2335.3 0.6 3.63 5.53 0 22

0.3% 7.3% 0.3% 4.4%

19.7%

68.0%

Cistern Concrete Pad Energy use by pump Water treatment Wastewater treatment Others

Figure SI - 2. 1 Percentage Contribution to Life cycle GHG emission of implementing harvested rainwater in an average medium office building.

500 (a) 450

400 More crowded building 540 Occupants 350

(MT CO2 e) CO2 (MT 300 Average medium office 250 270 Occupants 200 More spacious building 150 30 Occupants 100 50

Life time GHG emission emission GHG time Life 0

0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

250 Bigger building (b) 160,000 ft2 roof area Smaller building Average medium office 8,000 ft2 roof area 200 18,000 ft2 roof area

150 (MT CO2 e) CO2 (MT

100

Life time GHG emission emission GHG time Life

0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 Demand to supply ratio

500Stormwater Waste water Potable water required 0 Energy0.25 use0.50 by pump0.75 1.00 1.25 1.50Concrete1.75 2.00 Pad2.25 2.50 2.75 3.00 3.25Cistern3.50 3.75 4.00 4.25 4.50

Figure SI - 2. 2 RWH scenario life cycle GHG emissions in average medium office buildingfor combined sewer– (a) constant roof area, changing occupancy and (b) constant occupancy, changing roof area.

400 More crowded building (a) 540 Occupants 350

300 Average medium office 268 Occupants

250 (MT CO2 e) (MTCO2 200

150 More spacious building 30 Occupants 100

Life time GHG emissionGHGtime Life

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 400 (b) 350 Smaller building Average medium office 300 Bigger building 8,000 ft2 roof 160,000 ft2 roof 18,000 ft2 roof area area 250 area

(MT CO2 e) (MTCO2 200

150

100

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS

Life time GHG emissionGHGtime Life 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.75 3 3.25 3.5 3.75 4 4.25 4.5 Demand to supply ratio 700

Waste water Potable water Energy use by pump Concrete Pad Cistern

B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… -300 R… 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

Figure SI - 2. 3 Life cycle GHG emission in average medium office building for separate sewer– (a) constant roof area, changing occupancy (b) constant occupancy, changing roof area.

900 800 Cistern Concrete Pad Energy use by pump Potable water Waste water Stormwater 700 600 (a)

(MT CO2 e) CO2 (MT 500 400 300 200 100

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS Life time GHG emission emission GHG time Life 0.23 0.46 0.68 0.92 1.14 1.37 1.59 1.82 2.04 2.27 2.50 2.70 2.94 3.23 3.33 3.70 3.85 4.17 4.35 4.55

600 Cistern Concrete Pad Energy use by pump Potable water Waste water Stormwater 500 (b)

400 (MT CO2 e) CO2 (MT 300

200

100

Life time GHG emission emission GHG time Life

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.23 0.46 0.68 0.92 1.14 1.37 1.59 1.82 2.04 2.27 2.50 2.70 2.94 3.23 3.33 3.70 3.85 4.17 4.35 4.55 Demand to supply ratio (D/S)

Figure SI - 2. 4 RWH scenario life cycle GHG emissions in average medium office building for combined sewer when the rainwater cistern is sized using Yield After Spillage approach and daily precipitation– (a) constant roof area, changing occupancy (b) constant occupancy, changing roof area.

1.60

1.40 (a) 1.20

1.00

0.80

0.60

0.40

0.20

0.00

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

Life cycle Life GHG emission per person (MT

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 1.60

1.40 (b) 1.20

1.00

0.80

0.60 eq) 0.40

0.20

GHg emission emission GHg per person (MT CO2 0.00

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

LIfe cycle LIfe Demand to supply ratio (D/S)

700Stormwater Waste water Potable water

-300

B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… Energy use by pump Concrete Pad Cistern R… 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

Figure SI - 2. 5 Life cycle GHG emissions per person in medium office for (a) constant roof area, changing occupancy, (b) constant occupancy, changing roof area. Business as usual (BAU) and rainwater harvesting (RWH) scenarios when building is connected to combined sewer network.

1.00 (a)

0.80

0.60

0.40

0.20

0.00

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

Life cycle Life GHG emission per person (MT

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 1.00 (b) 0.80

0.60

0.40 eq) 0.20

0.00

GHg emission emission GHg per person (MT CO2

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

LIfe cycle LIfe Demand to supply ratio (D/S)

700Stormwater Waste water Potable water

-300

Figure SI - 2. 6 Life cycle GHG emissions per person in medium office for (a) constant roof area, changing occupancy, (b) constant occupancy, changing roof area. Business as usual (BAU) and rainwater harvesting (RWH) scenarios when building is connected to separate sewer network.

0.35 (a) 0.3

0.25

0.2

0.15

0.1

0.05

Life cycle GHG emission (MT CO2 e) CO2 (MT emission GHGcycle Life

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.500.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 4.75 5.00 0.35 (b) 0.3

0.25

0.2

0.15

0.1

0.05

Life cycle GHg emission (MT CO2 e) CO2 (MT emission GHgcycle Life

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.500.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 4.75 5.00 0.5

0 Potable water required Waste water treatment Stormwater treatment

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS Cistern Concrete Pad Energy use by pump RWHS 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 4.75 5.00

Figure SI - 2. 7 Life cycle GHG emissions per square meter in medium office for (a) constant roof area, changing occupancy, (b) constant occupancy, changing roof area. Business as usual (BAU) and rainwater harvesting (RWH) scenarios when building is connected to combined sewer network.

0.3 (a)

0.25

0.2

0.15

0.1

0.05

Life cycle GHG emission (MT CO2 e) CO2 (MT emission GHGcycle Life

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.500.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 4.75 5.00

0.3 (b)

0.25

0.2

0.15

0.1

0.05

Life cycle GHg emission (MT CO2 e) CO2 (MT emission GHgcycle Life

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS RWHS 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50 4.75 5.00 0.5

0 Potable water required Waste water treatment Stormwater treatment

BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU BAU

Figure SI - 2. 8 Life cycle GHG emissions per square meter in medium office for (a) constant roof area, changing occupancy, (b) constant occupancy, changing roof area. Business as usual (BAU) and rainwater harvesting (RWH) scenarios when building is connected to separate sewer network.

45 (a) 40

(MT CO2 eq) CO2 (MT 25

Life time GHG emission emission GHG time Life

0.27 0.36 0.44 0.52 0.60 0.68 0.77 0.85 0.93 1.01 1.09 1.18 1.26 1.34 1.42 1.50

70 (b)

50 (MT CO2 eq) CO2 (MT 40

0.23 0.24 0.25 0.27 0.28 0.30 0.31 0.34 0.36 0.38 0.42 0.45 0.50 0.55 0.61 0.69 0.80 0.95 1.16 1.50 Life time GHG emission emission GHG time Life Demand to supply ratio

700

Waste-300 water Potable water Energy use by pump Concrete Pad Cistern

B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… B… R… 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 3.25 3.50 3.75 4.00 4.25 4.50

Figure SI - 2. 9 Life cycle GHG emissions of RWH in a small office building for combined sewer– (a) constant roof area, changing occupancy and (b) constant occupancy, changing roof area. Separate sewer results can be extracted from this figure by just removing stormwater treatment.

4000 (a) 3500

3000

2500 (MT CO2 eq) CO2 (MT

2000

1500

1000

500

Life time GHG emission emission GHG time Life

0.13 0.98 1.83 2.68 3.53 4.38 5.23 6.09 6.94 7.79 8.64 9.49

10.34 11.19 12.05 12.90 13.75 14.60 15.45 16.30

4000 (b) 3500

3000

2500 (MT CO2 eq) CO2 (MT 2000

1500

1000

500

Life time GHG emission emission GHG time Life

0.39 0.41 0.43 0.46 0.49 0.52 0.56 0.60 0.65 0.72 0.79 0.89 1.00 1.16 1.37 1.68 2.16 3.04 5.11 Demand to supply ratio 16.04

700

Waste-300 water Potable water Energy use by pump Concrete Pad Cistern

Figure SI - 2. 10 Life cycle GHG emissions of RWH in a large office building for combined sewer– (a) constant roof area, changing occupancy and (b) constant occupancy, changing roof area. Separate sewer results can be extracted from this figure by just removing stormwater treatment.

Chapter 3

Environmental impacts from harvesting rainwater: A comparison of supply versus demand based analysis for different climatic regions

Jay P. Devkota1, Harish Jeswani2, Adisa Azapagic2, Defne S. Apul1*,

1 2801 W. Bancroft, Department of Civil Engineering, University of Toledo, Toledo, OH,

43606.

2 School of Chemical Engineering and Analytical Science, University of Manchester,

United Kingdom.

*Corresponding author tel: (419)-530-8132; email: [email protected]

Abstract

Purpose

Rainwater harvesting (RWH) and RWH life cycle assessments (LCAs) are becoming popular among practitioners and researchers. RWH serves two purposes; supply of rainwater and fulfilling the urban water demand. Water supply from RWH systems and the extent of the supply to meet the demand depend on geographical location but the environmental implications of these factors have not been previously studied. This study

76 is the first study to introduce multiple functional units in the LCA of RWH systems to compare the dual functional nature of RWH system for various geographical locations.

Methods

Two different purposes of the RWH system were analyzed using two different functional units for nine climatic regions in the US and for one city in the UK. One cubic meter of rainwater supplied was used as the supply based functional unit while provision of sanitation service in the building throughout its life time was used as the demand based functional unit. Rainwater cistern was sized using Yield After Spillage approach and long term daily precipitation. Other RWH components were also designed accordingly. With the respective system boundaries for two functional units, eight different impact categories were estimated and compared among different climatic regions using environmental emission data from GaBi and Ecoinvent databases.

Results and discussion

The size of the rainwater cistern varied from 19 m3 (5% reliability) to 265 m3 (70% reliability) within the climatic regions. Locations that had frequent and evenly distributed precipitation events had smaller cistern sizes with higher volumetric reliability compared to cities that had intense and less frequent precipitation patterns. Manchester had the least net impact from implementing RWH system when analyzed using supply based functional unit while Houston (in case of separate sewer), Boston (in case of combined sewer) and

Las Vegas (in case of greenhouse gas emission) had the least net impact when analyzed using the demand based functional unit. The absolute environmental impacts varied

77 significantly among regions for supply based functional unit while the variation was negligible when demand based functional unit was used. Rainwater cistern and energy use by pump were major contributors to life cycle impact for supply based while wastewater and potable water treatment were major contributors for demand based approach.

Conclusions

This study concludes that the region that would have the most environmental benefits from RWH system depends on the choice of the functional unit. Since the results from one functional unit to the other cannot be directly converted, we recommended to compare supply and demand before selecting the functional unit for LCA of RWH systems.

If the demand is greater than what can be supplied from RWH, the supply based functional unit (1m3 of harvested rainwater) falls short of creating results that may be helpful for decision making.

Key Words

Rainwater harvesting system, life cycle assessment, regional precipitation variation, rainwater tank sizing, functional unit, demand and supply.

2.1 Introduction

RWH is the practice of capturing, storing and utilizing rainwater. By managing the water on site, RWH can decentralize and diversify both the water supply and stormwater management infrastructures. Harvested rainwater can be used for irrigation (Bronchi et al.

1999; Fewkes 1998; Furumai 2008); landscape irrigation (Li et al. 2002), crop irrigation

(Yuan et al. 2003 and Liang et al. 2011), laundry washing (Bronchi et al. 1999; Angrill et al. 2012), car and parking lot cleaning (Ghisi and Mengotti de Oliveira 2007; Villarreal et al. 2005), water cooling (Furumai 2008) and creation of recreational waterways (Furumai

2008). There is a growing interest in utilizing harvested rainwater for toilet flushing in residential and commercial buildings to reduce the potable water demand (Anand and Apul

2010; Devkota et al. 2013; Devkota et al. 2015; Ghimire et al. 2014; Vargas-Parra et al.

2013; Proenca et al. 2011; Racoviceanu et. al. 2010). In residential buildings about 27% of the indoor water (or 45-100 L per capita per day) is used for toilet flushing (Mayer et al. 1999 and Gleick 1996), while in hotels, dorms, office and educational buildings this percentage is likely to be higher since toilets and sinks are the primary uses of water in these buildings (Crites R. 1998). RWH use in toilet flushing can therefore largely reduce the potable water demand in buildings.

One important aspect of life cycle assessment (LCA) is that the same product can be analyzed using different functional units. The definition of the functional unit and the selection of corresponding system boundary have significant effects on the comparability and reliability of LCA results (Martinez-Blanco et al. 2014). The concept of multiple functional units has mostly been used in agricultural food production. Unit mass of food

(e.g. 1kg of commercial dry matter, one ton of or one single commercial fruit), nutritional 79 value of food (e.g. 1kg of caffeic acid or sinapic acid equivalent), and area of land are some common functional units in food LCAs (Haas et al. 2001; Hayashi 2006; Basset-Mens and

Van der Werf 2005 and Martínez-Blanco et al. 2011). The use of different functional units have been helpful in better interpreting and understanding the environmental burden, productivity and farm income (Roy et al. 2009).

RWH systems can also be conceptualized as having different functions. In acting as a water source, a RWH system may or may not be able to meet the water demand. We argue that the specific goal and associated functional unit of RWH systems is affected by this situation. When the goal is stated from a supply perspective, and the purpose is to provide an alternative water source, an appropriate functional unit would be to supply 1 m3 of rainwater. This functional unit was used by several authors (Bronchi et al. 1999; Angrill et al. 2012; Ghimire et al. 2014 and Wang and Zimmerman 2015). On the other hand, the goal can also be stated from a demand perspective in which case the functional unit is selected based on the purpose of meeting the water demand. In this case, if the harvested rainwater is used for toilet flushing, the functional unit can be expressed as providing sanitation services during the life time of the building (Anand and Apul 2010 and Devkota et al. 2015) or as toilet flushing need per person per day (Crettaz et al. 1999). The supply based functional unit does not consider the extent of demand that can be met using harvested rainwater. In many cases, supply of rainwater may not be sufficient to fulfill the demand in which case potable water from municipal supply would be required to supplement the harvested rainwater. In these situations the functional unit on how the water is being used may be more appropriate compared to the functional unit based on the volume of rainwater supplied from the RWH system.

The choice of functional unit has effect on what is to be included in the system boundary which itself then effects the environmental impacts from RWH systems. For example, RWH LCAs that have used one cubic meter of rainwater supplied (supply based) as functional unit reported that energy use by pump and cistern were major contributors

(Bronchi et al., 1999 and Ghimire et al. 2014) while wastewater treatment and potable water treatment were the major contributor for the studies that have used sanitation service

(demand based) as the functional unit (Anand and Apul 2010 and Devkota et al. 2015). The results from one functional unit cannot be converted from one to another unless similar set of system boundary parameters has been selected for the study.

The goal of this research was to address the dual issue of supply and demand associated with RWH systems by considering two separate but related functional units. The effect of different functional units was studied on various climatic regions. For this purpose, an LCA model of a RWH system was developed for two different functional units for ten cities in different climatic regions. Nine of these were based in the US (Toledo, Salt

Lake City, Seattle, Las Vegas, La Crosse, Houston, Fort Myers, Fargo and Boston) and one in the UK (Manchester). The functional units were defined as ‘one cubic meter of rainwater supplied’ and ‘provision of sanitation service in a building’ for supply and demand based analysis, respectively. For the supply-based study, the goal was to consider the life cycle environmental impacts of supplying water from an alternative source

(rainwater) irrespective of the demand. In the demand-based study, the goal was to estimate the impacts of meeting the demand from rainwater harvesting system and potable water supply in case when harvested rainwater was not enough.

2.2 Methodology

2.2.1 Life cycle modeling approach

We used LCA to evaluate environmental impacts of using RWH system in an office building for both demand and supply based functional units across various climatic regions.

The modeling approach was adapted from the EEAST model (Devkota et al. 2013). This model first designs the RWH system based on building characteristics and precipitation followed by the estimation of environmental impacts using LCA data from the GaBi (PE

International 2014) and Ecoinvent (2010) databases.

Two functional units, ‘one cubic meter of rainwater supply’ and ‘provision of sanitation service in a building’, were chosen to estimate and compare the impacts of implementing a RWH system in an office building. The system boundary for both functional units was from ‘cradle to grave’, comprising construction, operation and decommissioning of the RWH system. However, the choice of the functional unit has implications on what exactly was included within the system boundaries. A schematic representation of the system boundary for both functional units is shown in Figure 3.1. For the supply-based functional unit, the system boundary includes construction and installation of RWH components for collecting and supplying rainwater, such as rainwater cistern, concrete pad, pump, pipes, bends and valves, filter and overflow pipe (Figure 3.1; items 1 through 7). The operational stage of supply based system includes the replacement of the filter and pump (Figure 3.1; replacement of items 3 and 6) as well as the energy used by the pump and the treatment of the stormwater (Figure 3.1; items 8 and 9). Since the use of rainwater avoids the need for potable water as well as stormwater treatment (in case of combined sewer network), the credits for avoided potable water treatment and avoided 82 stormwater treatment (where relevant) were also included in the analysis. It is interesting to note that the previous studies that used the supply based functional unit did not consider stormwater treatment contributing to the impact or to the avoided burden (Angrill et al.

2012; Ghimire et al. 2014; Morales-Pinzón et al., 2012 and Vargas-Parra et al., 2013). Our approach here is different because we argue that the use of rainwater harvesting system not only reduces the potable water but also avoids the treatment of stormwater being diverted from entering the combined sewer.

For the demand based functional unit, the system boundary in the construction phase includes everything included in the supply based analysis (Figure 3.1: items 1 through 7) plus toilet specific items such as the toilets themselves, flush accessories, the underdrain and vent pipes (Figure 3.1; items 10, through 13). In the operational stage, in addition to the components included in supply-based system, the demand-based system consist of additional potable water treatment (if required), wastewater treatment and toilets with accessories replacement (Figure 3.1; items 14 and 15). In cases where there was not enough rainwater to meet the demand, the potable water was required and therefore its treatment was also included in the system boundary. Credits for the avoided treatment of water and stormwater were also applied. The detailed equations used to estimate the volume of potable water, wastewater and stormwater for both the functional units are provided in section SI-3.1 of supporting information.

Figure 3. 1 System boundary of rainwater harvesting system for supply based analysis and demand based analysis. Hatched portion represents operational phase whereas plain region is represents construction phase.

A schematic representation of the methodology is shown in Figure 3.2. RWH components (rainwater cistern, concrete pad, pump, dual pipes, filter, bends valves and tees, toilets, flush tanks, overflow pipe, concrete pad, under-drain and vent) were designed based on the precipitation of nine climatic regions of US as well as Manchester-UK. Roof area and occupancy for an average medium sized office building were assumed according to the U.S. Department of Energy commercial reference building model developed for the national building stock (Deru et al. 2011). As the geographical location changed, the quantities of some RWH components, for example, cistern, concrete pad, energy use by

84 pump, potable water required and stormwater also changed, whereas other components

(wastewater treatment, dual piping, filter, pump, toilets and accessories, overflow pipe, under-drain and vent) remained the same for all cities considered. All the components included in the construction phase (marked with * in Figure 3.2) are different for different climatic regions but similar for the two functional units. The replacement components such as filter and pumps, were assumed to be replaced every 5 and 20 years, respectively (Kirk and Dell’Isola 1995). Toilets and flush accessories were assumed to be replaced every 35 years (Kirk and Dell’Isola 1995 and Costlab 2015). A detailed inventory for the supply and demand based functional units for both combined and separate sewer scenario is presented in supporting information (Table SI-3.1 to Table SI-3.4).

Life cycle inventory data were sourced from the Ecoinvent v 2.2 (Ecoinvent 2010) and Gabi databases (PE International 2014). USLCI data incorporated in GaBi software were used for the RWH components in US regions whenever available. For potable water treatment, wastewater treatment and concrete pad, US specific data were not available in the databases. For these, the original electricity mix was replaced with the US grid with an assumption that the processes were similar. GaBi database does not have the electricity mix for different climatic regions of US as defined later in this study (See section 3.2.4).

Therefore, an average electricity mix for the U.S. was used as a representative of the electricity mix for all the regions. Wang and Zimmerman (2015) used eGRID database for carbon grid intensity of electricity production but this database (eGRID) does not have data for other environmental impact categories reported in this manuscript. In addition, eGRID database only includes the operational phase emissions and not the supply chain effects.

Assuming the life time of the building as 75 years, the life cycle environmental impacts (in

85 terms of energy (net calorific energy value), greenhouse gas emission, acidification potential, eutrophication potential, ozone depletion potential, ecotoxicity air, ecotoxicity soil and ecotoxicity water) were estimated for the two different functional units using

TRACI as impact assessment method (Bare, 2011). A detailed equation of the impact estimation for the two different functional units is explained in section SI-3.2 of the supporting information (section 3.6).

Precipitation

Roof Area – Avg. Toledo, Boston, Seattle, Houston, Occupancy – Avg. med. office building Fort Myers, Las Vegas, Manchester- med. office building in US (53,628 Sq ft) UK, Fargo, La Crosse & Salt lake (268) City

RWH system design

Functional Unit

Demand based functional unit Supply based functional unit (One (Sanitation Service in the building) cubic meter of rainwater supplied) throughout its life time.

Quantity Estimation and LCA

Demand based FU Supply based FU

Construction phase Operational phase Construction phase Operational 1. Cistern* 8. Stormwater 1. Cistern* phase 2. Concrete pad* treatment 2. Concrete pad* 8. Stormwater 3. Pump* 9. Energy use by 3. Pump* treatment 4. Dual pipes* pump 4. Dual pipes* 9. Energy use by 5. Overflow pipe* 13. Potable water 5. Overflow pipe* pump 6. Filter* treatment 6. Filter* 10. Replacements: 7. Bends, valves, tees 14. Wastewater 7. Bends, valves, filter, pump. 10. Vent pipe* treatment tees* 11. Toilet* 16. Replacements: 12. Flush accessories toilet, filter, pump. 15. Under drain*

Life cycle environmental impacts for different regional precipitation variation

Demand based FU Supply based FU

Figure 3. 2 Schematic representation of conceptual LCA model for RWH systems for different regions with two different functional units.

2.2.2 Building characteristics

An average medium sized office building in the US according to National Building database, Department of Energy (Deru et al. 2011) was used as a base building. Office building are about 5.1% of the LEED certified buildings (National Green Building

Adoption Index 2014) in the United States and this number is expected to increase because it is now a norm for almost every office buildings to adopt sustainable construction practices (National Green Building Adoption Index 2014). For our analysis we chose toilet flushing as the end use for harvested rainwater because it constitutes a large fraction (37

%) of water use in office buildings (Saving Water in Office Building, EPA 2012).

According to Commercial Building Energy Consumption Survey, there are 404,000 medium office buildings in the United States (CBECS 2012) which constitute 15 % of the building stock (U.S. Census 2000). The medium sized office building considered in this study has 3 stories (excluding the basement) with a roof area of 17,876 sq ft and accommodates 268 occupants (Deru et al. 2011).

2.2.3 System sizing

2.2.3.1 Demand and Supply

Water demand for toilet flushing was estimated using, building occupancy (268 occupants, four flushes per person per day (Vickers 2001 and Alliance for Water Efficiency

2015) and standard flush volume for standard toilets (1.6 gallons per flush). The daily supply of rainwater was estimated by multiplying the daily precipitation of different climatic regions by the roof area of the building assuming that 25% (roof runoff coefficient 88

= 0.75) of the rainwater might get lost before entering the cistern (Krishna 2005). A roof runoff coefficient of 0.75 to 0.95 has been used in previous studies to estimate roof runoff

(McCuen 2004; TxDOT 2009 and Viessman and Lewis 2003). The ratio of toilet flushing demand to rainwater supplied from the roof (D/S) was also calculated as explained in

Devkota et al. (2015 – accepted). The relation between D/S, spillage and environmental impacts were analyzed.

2.2.3.2 Tank Sizing

Both daily (Khastagir and Jayasuriya 2011; Imteaz et al. 2011; Ward et al. 2010;

Angrill et al. 2012; Fewkes and Wam 2000; Mitchell 2007 and Devkota et al. 2015) and monthly (Krishna 2005; Georgia Rainwater Harvesting Guideline 2009; Farreny et al.

2011; Ghisi and Ferreira 2007; Aladenola and Adeboye 2010 and Devkota et al. 2013) tank sizing methods have been used in RWH LCA studies. Rainwater cistern is oversized when monthly precipitation data is used to size it (Khastagir and Jayasuriya 2011). Yield before spillage (YBS) and yield after spillage (YAS) approaches are the most widely used approaches in the daily mass balance model. In this study we used YAS approach with long term daily precipitation to size the rainwater tank as it gives a conservative estimate of tank yield (Mitchell 2007 and Fewkes and Butler 2000). Use of 10-year precipitation pattern does not significantly reduce the accuracy of the estimated yield of rainwater tank compared to the 50-year precipitation pattern (Mitchell 2007). In this study 15-year daily precipitation (from the year 2000 to 2014) data were used to size the rainwater tank.

Volumetric reliability (Vr) of a rainwater cistern, which is also a measure of the water saving efficiency of the tank, is used to size the tank. Vr can be calculated using:

푛 ∑푡=1 푌푡 Vr = volumetric reliability = 푛 ……………………………………………...….. 3.1 ∑푡=1 퐷푡

Where,

3 Yt = Volume of rainwater supplied from the cistern– daily (m )

3 Dt = Daily water demand (m ).

n = number of days (15 * 365 = 5475)

It should be noted that “t” is for every day.

In the YAS approach, Yt and Dt are calculated using:

Yt = MIN ( Dt , Vt-1 + It ) ……………………………………………………. 3.2

Vt = MIN (Vt-1 + It – Yt , C – Yt ) …………………………………………… 3.3

Where,

3 Vt = volume of rainwater in the cistern at the end of time step t (m )

3 Vt-1 = volume of rainwater in the cistern at the end of previous time step t (m )

3 It = inflow or roof runoff – daily (m )

C = Capacity of the rainwater cistern (m3)

Daily spillage (St) is calculated from the daily mass balance equation (equation 4)

Vt = Vt-1 + It + Pt – Et – St – Lt - Yt …………………………………….….………...3.4

Where:

3 Pt = incident precipitation received by the tank – daily (m )

3 Et = daily evaporation (m )

3 St = amount of spillage due to overflow from tank – daily (m )

3 Lt = seepage or leakage - daily (m ) 90

Evaporation (Et), incident precipitation (Pt) and leakage (Lt) are very small and are not considered in this study.

The first step in sizing a rainwater cistern using YAS approach was to assume various cistern sizes at an equal interval. Based on the initial storage volume (Vt−1), rainfall inflow on first day (It) and daily water demand (Dt), the volume of rainwater supply from the cistern (Yt)) was then calculated for the first time step t using equation 3.2. It is important to note that rainwater supply (from roof) may not be always equal to the rainwater supply from cistern unless there is no spillage from the cistern. After usage on first day, the volume of water remaining in the cistern at the end of time step t also called

Vt was calculated using equation 3.3. Daily spillage was calculated by comparing the daily demand and supply using Equation 3.4. The initial storage volume (Vt−1) for the second time step would be the volume of water at the end of first time step (Vt). Similar to the first time step, Yt and Vt for the second time step were then estimated using Equations 3.2 and 3.3 respectively and so on. Volumetric reliability of the rainwater tank was estimated by dividing long term daily supply with long term daily demand. The volumetric reliabilities of series of rainwater tanks assumed in the first step were then assessed using Equation 3.1.

The size of the rainwater tank was considered optimum when any increment in the tank size only changes the volumetric reliability by 1% or less.

We also note that Vr may appear as an inverse of D/S ratio. D/S ratio was estimated using equation 3.5. While Vr gives an estimate of efficiency rainwater supply from the cistern, D/S gives an estimate of the ratio of average daily toilet flushing demand to average daily rainwater supply from the roof. Vr is equal to D/S ratio when daily spillage (St) is zero. Use of these parameters are helpful in estimating the effect of rainwater supply with

91 and without spillage. Spillage is important because in case of combined sewer scenario, it required treatment and hence increases the energy and emissions.

퐷푡 D/S = ……….………………………………………………………… 3.5 푌푡+푆푡

Where,

3 Yt = Volume of rainwater supplied from the cistern– daily (m )

3 Dt = Daily water demand (m ).

3 St = amount of spillage due to overflow from tank – daily (m )

2.2.3.3 Dual Piping

Dual piping is required to supply the rainwater from the rainwater cistern to the toilets. Length of dual piping required for conveying harvested rainwater to toilets was calculated by assuming that one primary pipe runs to the top floor and that each toilet was approximately 3 ft away from this pipe. In addition, the horizontal length of pipe from the cistern to the main pipe was assumed to equal to the width of the building. The pipe was assumed to be made of PVC.

2.2.3.4 Energy use by pump

It is assumed that the tank was placed on the ground. Therefore a pump was required to transport the rainwater from the cistern to the toilets in respective floors. Energy delivered to pump was estimated using equation 3.6:

P = ϒ x Q x (he + hp) x [1+α]/η ………………………………………………………3.6

Then the pump’s life cycle energy requirements (E) are calculated as:

E = P x 365 [days/year] x 24 [hrs/day] x 0.001 [kW/W] x 75 years ………………..3.7

Where,

P = Power delivered to Pump (W)

E = Annual energy required by Pump (kWh)

Q = Flow rate (m3/sec).

h = he + hp = sum of elevation head and pressure head provided by the pump (m).

η = combined mechanical and electrical efficiency of the pump (assumed 65 %;

Cengel and Cimbala 2005).

ϒ = Specific weight of water (N/m3)

α = Percentage of energy lost due to friction (assumed 0.3%; Cheng 2002).

Equations 3.6 and 3.7 are valid for both supply and demand based approach.

Pressure head provided by the pump (hp) was set equal to the pressure provided by the city

(assumed 35.2m of water or 50 psi). Elevation head provided by pump (he) was set equal to the building height.

2.2.3.5 Other rainwater harvesting components

A square concrete pad of thickness 0.1 m (four inches) was assumed as the supporting foundation for the rainwater cistern and the dimensions were designed in such a way that the side of the square concrete pad would be 0.3 m (one foot) greater than the diameter of the cistern on each side. A floating filter made of plastic with the filter media was provided inside the rainwater cistern to separate out the floating matters in the tank

93 such as dead leaves, tree branches, dead birds and insects. Toilets and flush tank made of vitrified clay were assumed for the study. Other pipes and fittings (overflow pipe, bends, valves, tees, under drain) made of PVC were assumed.

2.2.4 Regional precipitation variation

Nine different climatic regions within US as defined by National Climatic Data

Center (NCDC 2015) were selected to see the variation in life cycle impacts of using RWH system in an average medium office building in different regions of US (Figure 3.3). The selected cities with their respective climatic regions are presented in Figure 3.3. In addition, one region in the United Kingdom was also selected to see the variation in precipitation across countries. Manchester being one of the cities in the UK with high and frequent precipitation events was selected as one of the case study in our research.

SEATTLE

LA-CROSSE FARGO

BOSTON LAS VEGAS

TOLEDO

SALT LAKE CITY

HOUSTON FORT MYERS

Figure 3. 3 Nine climatic regions within the contiguous United States identified by National Climatic Data Center (NCDC). Representative city for each climatic regions are displayed in the Figure.

2.2.5 Sewer Type

Previous research reported that one of the major factors affecting the sustainability of RWH system is whether the building is connected to a combined or separate sewer network (Devkota et al. 2015). In a combined sewer system, if there are no combined sewer overflows, the excess runoff from the building (the spilled water from the rainwater tank) is treated at a wastewater treatment plant increasing the environmental impacts due to the operation of the treatment plant. Combined sewer overflows reduce the volume of stormwater runoff that reaches the wastewater treatment plant but this volume can only be 95 determined using a watershed scale analysis (Tavakol-Davani et al., 2015 - accepted).

Therefore, for simplicity, in this study, as in Devkota et al (accepted) we assumed all stormwater is being treated at the wastewater treatment plant. Out of the climatic regions considered, Northeast, Central, Upper Midwest, Southeast, Northwest, West and

Manchester- United Kingdom have both combined and separate sewers whereas South,

Southeast and North Rockies & Plains have only separate sewers. Therefore, in this study we assumed that the building was connected to a separate sewer network for all the regions under consideration, while regions such as Northeast, Central, Upper Midwest, Southeast,

Northwest, West and Manchester- United Kingdom were also analyzed for the building connected to a combined sewer network.

2.3 Results and Discussion

2.3.1 Overview of key parameters affecting environmental impacts

80% HOU, 265 m3 Vr =

3 60% BOS, 66 m , Vr = 55% MAN-UK, 28m3, Vr = F.MYE, 3 3 76 m , TOL, 47m , Vr = Vr = 3 40% SEA, 47m , Vr = 42% L.CRO, 57m3, Vr = 42% FAR, 38m3, Vr = 28%

20% SLC, 28m , Vr = 20% Volumetric Volumetric reliability(Vr) L.VEG, 19m3, Vr = 0% 0 50 100 150 200 250 Rainwater cistern size (m3)

Figure 3. 4 Rainwater cistern curves with respective sizes and volumetric reliabilities for different climatic regions. City TOL represents Toledo OH, SLC represents Salt Lake City UT, SEA represents Seattle WA, MAN-UK represents Manchester – United Kingdom, L.VEG represents Las Vegas NV, L. CRO represents La Crosse WI, HOU represents Houston TX, F. MYE represents Fort Myers FL, FAR represents Fargo ND and BOS represents Boston MA.

The cistern size for the climatic regions varied from 19 m3 (Las Vegas) to 265 m3

(Houston) (Figure 3.4). In this study, we provide an overview of several parameters that played an important role in environmental impacts from rainwater harvesting systems.

These parameters varied considerably across the different cities (Table SI 3.7) and were often correlated, albeit not perfectly, to each other (Figure 3.5). A higher cistern volume

97 implied a higher environmental impact due to higher mass of steel used in the cistern which affect the impacts from manufacturing and transporting the cistern.

The volumetric reliabilities (Vr) of the cisterns provide a measure of how reliable the system is in supplying rainwater. Higher Vr implies larger volumes of rainwater supplied and less of a need for municipally supplied water. In comparing the different regions where we assumed the demand was constant, a higher Vr means a larger volume of rainwater supplied. Vr generally increased with tank size and then plateaued after the optimum tank size (Figure 3.4). Cities with little available rain typically start out with a low Vr and remained at a low Vr even at larger tank sizes (e.g. Las Vegas and Salt Lake

City) whereas in other cities, Vr increases, albeit at different rates, ultimately reaching a plateau at a considerably higher Vr value (e.g. Houston, Boston, Fort Myers) (Figure 3.4).

The water to demand to supply (D/S) is a measure of whether there is sufficient rainwater to meet the demand from toilet flushing. A value of D/S greater than one implies insufficient rainwater to meet the demand. For the different climatic regions, D/S varied from 1.2 (Houston) to 16.7 (Las Vegas) suggesting that RWH system in any of the cities considered were not able to supply enough water to meet the toilet flushing demand in the average medium office building (Table SI 3.7). In other words, additional water supply from municipality was needed in all cases. Vr and D/S ratio were inversely related with some irregularities (Figure 3.5 b and 3.5 g). These two parameters would be exactly inversely related when the daily spillage was zero. For the region with least spillage, (Salt

Lake City, spillage = 1%), the D/S ratio (4.8) and Vr (20%) were close to inversely related.

Similarly for the region with higher overflow spillage (Houston, spillage = 15%), the D/S

98 ratio (1.2) was not even close to inverse of Vr (70%). Due to the spillage, on the denominator of D/S ratio (refer to equation 5), higher spillage caused some irregularities.

Spillage affects environmental impacts because the spilled rainwater is assumed to be treated at the wastewater treatment plant if the building is connected to a combined sewer system. Interestingly, even though the supply was less than demand, spillage from the cistern was observed for all the climatic regions. Here we report percentage of spilled rainwater but this percentage is directly correlated with the volume of spilled water. Percent spillage from various climatic regions varied from 1% (Las Vegas and Salt Lake City) to

15% (Houston). Dry cities (Las Vegas and Salt Lake City) had the least spillage (Table SI-

3.7) whereas wet cities (Houston, Fort Myers) had the highest spillage of 15% and 14%, respectively. The amount of spillage was primarily an effect of the way the tank was designed. During high intensity rain events, larger volumes of rainwater were available for capture, but there was also more spillage because the tank size was optimized in such a way that it filled in a single day rain event and emptied in fulfilling one day demand

(Mitchell 2007).

Ultimately, the cistern size was determined from Vr and spillage and these three parameters were loosely correlated to each other as well as ancillary parameters such as

D/S, average daily precipitation, and rainfall frequency measured in this study as percentage of days without rain. However, the correlations were not perfect because the tank sizing method itself was not linear (Figure 3.5). If the tank sizing approach with monthly precipitation had been used, much smoother correlations would have been observed and the percentage of days without rain would not even have any effect on the environmental results.

Figure 3. 5 Scatter plot matrix for cistern size (C), volumetric reliability (Vr), demand to supply ratio (D/S0, percentage of rainwater spilled (% spillage), percentage of days without rain, and average daily precipitation. 100

2.3.2 Extent of Variation of Absolute Impacts (Positive bars in Figure 3.6 and 3.7)

The variation of absolute impacts among climatic regions was much higher for the supply based analysis than the demand based analysis (Figure 3.6 and 3.7). For example, keeping Las Vegas (with lowest precipitation) as the base point, the maximum percent difference in absolute impact was 97% for the supply based analysis (for water ecotoxicity,

Fort Myers, combined sewer case, Table SI 3.5) but only 23 % for the demand based analysis (for energy, Houston separate sewer case; Table SI 3.6) when connected to a combined sewer for supply based analysis.

Supply based analysis

The absolute impacts in the supply based analysis showed a high variation across regions because the cistern size and volume of spillage varied considerably across regions.

The effect of this variation was further enhanced when the impacts were divided by the supplied rainwater volume which itself varied. For example, Las Vegas had the smallest tank size and the least volume of rainwater supplied. Yet, the volume of rain was many more times smaller than the volume of tank size. Taking Manchester as a comparison, Las

Vegas’s volume of rainwater supplied was 10 times smaller but its tank size was only two times smaller (Table SI 3.7). Therefore, the very small value of the volume of rainwater in the denominator greatly increased the impact of the cistern making Las Vegas the city with the highest energy demand in Figure 3.6.

Many times in an LCA analysis, the goal is to reduce the impacts. The supply based analysis showed that for absolute impacts to be small, the volume of rainwater supplied

(Vr) had to be high and the cistern size (C) and stormwater treatment (volume of spillage measured as percentage spillage) had to be low. These conditions were best met when 101 there was a continuous supply of plenty rainwater throughout the year as illustrated in the case of energy demand for Manchester, which was the lowest among all cities. Manchester received considerably high amount of rain (3.4mm/day) but not as high as Houston (4.5 mm/day) and Boston (3.6 mm/day) (Table SI 3.7). Yet, Houston’s rain was unreliable and the low frequency of rain resulted in a very large tank size (265 m3), much greater than those of Boston (66 m3) and Manchester (28 m3). Because of the infrequent but large spurts of rain, there was also high spillage in Houston (1m3/day) compared to Manchester (0.3 m3/day) and Boston (0.3 m3/day) (Table SI 3.7). Therefore, despite having a large annual rain, Houston was not a good place to implement rainwater harvesting because of the large cistern size and high spillage. In comparing Boston to Manchester, both had relatively close average daily precipitation. But Manchester’s rain was more uniform throughout the day

(days without rain = 10 %) resulting in a much smaller tank than in Boston (days without rain = 60 %). Manchester had lower impact than Boston due to the smaller tank size.

Unfortunately, it was not easy to generalize the observations from Boston,

Manchester, Houston, and Las Vegas. As seen in Figure 3.5, as Vr increased, so did C and percentage spillage. Vr and percentage spillage increased at a higher rate than C. (For

3 example, in Figure 3.5a, Vr increased from 5 to about 50 m (10 fold) during which C increased from about 20 to only 70 m3 (about 3-4 fold)). Yet, the correlations among these three parameters were not perfect. It was ultimately the interplay between these parameters that effect the absolute impact from the supply based analysis.

Demand based analysis

In the demand based analysis, in contrast to the supply based analysis, there was little variation in absolute impacts because the impacts were not normalized to the volume

102 of rainwater supplied and therefore was not affected by the variation in this parameter

(Figure 3.7). In addition, unlike the supply based analysis, wastewater treatment was a large contributor (66% for energy – Houston to 98% for ecotoxicity – Las Vegas) to the absolute impact in the demand based analysis and overshadowed the variation in impacts from other system components (Figure 3.7). For example, while the cistern size varied greatly (about 14 times between Houston and Las Vegas), any variation in cistern size was completely overshadowed by the large contribution from wastewater treatment for all impact categories. The impacts from potable water treatment (and energy use by pump in global warming and acidification impact categories) can be seen in Figure 3.7. It was worth noting that when rainwater was not enough to flush the toilets, additional potable water from municipal supply system was used and unlike the supply based analysis, the treatment of municipally supplied water was included in the system boundary for the demand based.

The impact by potable water treatment varied by 11 fold (e.g. 0.3m3/day of potable water for Las Vegas versus 3.6 m3/day of potable water for Boston) but its contribution to the total impact and the extent to which it changed the impact across regions remain small due to the high impact resulting from wastewater.

Unlike supply based, a clear trend was seen between Vr and absolute impacts for the demand based functional analysis (Figure 3.7). With the increase in Vr, the rainwater supply increased and therefore the cistern size increased reducing the emissions from potable water treatment and stormwater treatment. Except global warming, all other impact categories were reported to increase with the decrease in Vr. It was also worth noting that on the contrary with the supply based approach, the absolute emissions for demand based approach were reported to decrease with the decrease in spillage according to demand

103 based analysis. Fort Myers with highest spillage had highest absolute impact followed by

Seattle (spillage = 7%).

2.3.3 Percentage Contribution to Absolute Impacts (Positive bars in Figure 3.6 and

3.7)

Minor Components

In both the supply and demand based analysis, several components did not even appear on any one of the bar graphs because their impacts were very small. Examples of these components are filter, pump including replacement, toilets including replacements, dual piping including bends, tees and valves and transportation. These components contribute marginally to all the impact categories and all the regions (Figure 3.6 and 3.7).

Major components

The effect of different system components on the impacts was different for each impact category. For example, for both the supply and demand based analyses, cistern had some contribution to the absolute energy, global warming and acidification potential but had negligible effect in other impact categories (Figure 3.6 and 3.7). This result supported the findings of Bronchi et al. (1999) and Ghimire et al. (2014) who stated that major contributor to environmental impacts are cistern and energy use by pump. It is worth mentioning that Bronchi et al. (1999) and Ghimire et al. (2014) had also used supply based functional unit.

For the supply based approach, stormwater treatment contributed around 99% of the eutrophication and ecotoxicity soil and ecotoxicity water impacts (Figure 3.6). For

104 ozone depletion and air ecotoxicity impact categories, stormwater treatment and energy use by pump were the major contributors to the impact (Figure 3.6). Similar to the supply based functional unit, contribution analysis showed that the major contributory factors for demand based functional unit also varied with the impact categories (Figure 3.7). As mentioned earlier, wastewater and potable water treatment were identified as the major contributor to life cycle energy, greenhouse gas emissions, ecotoxicity water, ozone depletion potential, acidification potential and ecotoxicity air, while for impacts, such as eutrophication and ecotoxicity soil, only wastewater was dominant (Figure 3.7). This study showed that both construction and operational phase has higher impact on environmental impact of implementing RWH system based on supply based approach while only operational phase has higher impact according to demand based approach.

2.3.4 Avoided Impacts from RWH system (Negative bars in Figure 3.6 and 3.7)

The inherent benefit of RWH systems is that it can avoid treatment of stormwater in combined sewers and of potable water in both separate and combined sewers (Devkota et al., 2015). When a RWH system is used, there is less need for a municipally supplied water and if the building is connected to a combined sewer, the stormwater saved and used for toilet flushing is not sent to the wastewater treatment. These benefits or avoided burdens are shown as negative impacts in Figures 3.6 and 3.7. In both figures, for combined sewer systems, the avoided burdens were always greater than separate sewer systems as they include the effects of not treating stormwater. Previous RWH LCA studies often discussed the benefits of saving potable water supply (Proenca et al., 2011; Angrill et al., 2012 and

Vargas-Parra et al., 2013). However, saving of stormwater can have a greater benefit to the 105 environment. This can be most easily observed in Figure 3.6, where, for the supply based functional unit the avoided burdens were the same for all regions since all results were normalized to 1 m3 of rainwater which results in saving 1 m3 of potable water treatment and 1 m3 of stormwater treatment at wastewater treatment plant. The combined sewer avoided burdens for all impact categories had longer light (from stormwater treatment) grey bars than dark (from potable water treatment) grey bars. In other words, for all impact categories, the stormwater savings had greater effect than saving municipally supplied potable water.

The effect of the choice of functional unit was clearly visible in avoided burdens.

While supply based functional unit did not differentiate among regions, the demand based functional unit showed a large variation in the avoided burden. The avoided burdens were directly related with the yield in each region. The Vr which is a measure of the yield

(considering demand was kept constant across regions) varied almost 15 times across regions (i.e. Vr=5 for Las Vegas and Vr=70 for Houston) causing this amount of variation in avoided burdens. When demand based analysis was considered for the separate sewer systems Houston with highest yield had the highest avoided burden and Las Vegas with lowest yield had the lowest burden.

2.3.5 Net Impacts from RWH system (red dots in Figure 3.6 and 3.7)

Supply based analysis

Net impacts (after deducting avoided burden from the absolute impacts) for the supply based approach varied with the climatic regions as well as the impact categories

(Figure 3.6). For example, Manchester had lowest net impact in terms of energy, GHG

106 emissions, acidification, ozone depletion and ecotoxicity soil while Boston had least net impact in terms of eutrophication, ecotoxicity soil and ecotoxicity water. Las Vegas had highest net impact for all the impact categories except eutrophication and ecotoxicity water.

This was because eutrophication potential and ecotoxicity water had negative emissions from cistern manufacturing which was reducing the net impacts. Negative emissions from manufacturing cistern indicate reduction in impacts from manufacturing which was due to the recycling benefits included in the system boundary. It was worth noting that GHG emission and acidification potential had positive net impacts indicating that there was no savings from RWH system when connected to separate sewer network.

Demand based analysis

Alike supply based, the net impacts (after deducting avoided burden) for the demand based functional unit varied also with the climatic regions as well as the impact categories (Figure 3.7). However, the result of demand based analysis showed that Boston had least net impacts for all impact categories for combined sewer scenario while Houston had least net impact when connected to a separate sewer network indicating that Boston and Houston were best place to implement RWH system among all the regions when connected to combined sewer and separate sewer respectively. Manchester, Fort Myers,

Toledo, Seattle and La Crosse also had lower net impacts. Las Vegas, Salt Lake City and

Fargo had the highest net impact thus suggesting that these regions were not very attractive

(Figure 3.7). It is also worth mentioning that all the impact categories had positive net impacts indicating that there was no savings from RWH system for both the combined and separate sewer scenario signifying that implementing RWH system is environmentally taxing in the selected climatic regions according to demand based approach.

107

2500 Avoided 0Stormwater treatment Avoided Potable water treatment -2500 Energy-5000 use by pump Stormwater treatment -7500 Concrete-10000 Pad Transportation from plant to installation site -12500

Toilets with flush accessories including replacements PVC dual piping + bends valves tees

SEP SEP SEP SEP SEP SEP SEP SEP SEP SEP

COM COM COM COM COM COM Floating tank filterCOM including replacements Pump including replacements MAN - UK BOS TOL SEA L. CRO FAR F. MYE HOU SLC L. VEG Storage tank Net Impacts

Figure 3. 6 Life cycle environmental impacts per cubic meter of rainwater supplied for RWH system according to supply based approach.

108

3000000 Avoided Stormwater treatment Avoided Potable water treatment 1500000 Energy use by pump Stormwater treatment 0 Wastewater treatment Water treatment -1500000 Concrete Pad

Transportation from plant to installation site

SEP SEP SEP SEP SEP SEP SEP SEP SEP SEP

COM COM COM COM COM COM Toilets withCOM flush accessories including replacements PVC dual piping + overflow drain + vent + sewer drain + bends valves tees HOUFloating tankBOS filterMAN including - F. MYE replacementsTOL SEA L. CRO FAR SLC L. VEG Pump including replacementsUK Storage tank Net Impacts

Figure 3. 7 Plot of life cycle environmental impacts from RWH system according to demand based approach.

109

2.3.6 Discussion and Implementation

We expected the regions with lower net impacts for one functional unit would also have lower impacts for the other functional unit. However, such a trend was not observed making it clear that the regional preference for RWH depends on the specific goal to be met: using harvested rainwater as a potential supplemental source versus using it to meet the water demand needs (Figure 3.6 and Figure 3.7). Most of the RWH LCA studies done so far have only focused on one function, either supply or fulfill the demand. Knowing the fact that RWH system serve two purpose, the choice of multiple functional units helps better understand and interpret the environmental impacts associated with it. Manchester was the best region to implement RWH system when analyzed using supply based functional unit while Houston (in case of separate sewer), Boston (in case of combined sewer) and Las Vegas (in case of GHG emission) were found to be attractive when analyzed using demand based functional unit. This showed that environmental impacts from RWH system depends on the choice of functional unit thereby supporting the findings of Martínez-Blanco et al. (2011) where he stated that the functional unit related to various parameters can give considerably different environmental impacts. Knowing that these different functional units have different system boundaries, it was not possible to compare and convert the results from one functional unit to another unless and until the study was carried out from the scratch. We therefore, suggest and recommend the supply and demand comparisons before selecting functional unit for LCA of RWH systems. If the supply is less than demand, chose demand based functional unit and if the demand is less than supply, chose supply based functional unit.

110

2.4 Conclusions

RWH system serves dual purpose: supply the rainwater and fulfill the urban water demand, thereby reducing potable water demand as well as stormwater runoff. While higher amount of rainwater was available for capture during intense precipitation events, there was be more frequent spillage. It was not only the precipitation intensity that affected the size and volumetric reliability of rainwater cistern, frequency of precipitation also had significant impact. More distributed precipitation events had smaller cistern sizes with higher volumetric reliability as compared to intense precipitation events. This is the first study to access the dual nature of RWH system in terms of environmental impacts. The effect of the supply and demand based analysis on the environmental impacts of implementing RWH system was presented. Environmental impacts were assessed and compared for two functional units: supply of one cubic meter of rainwater supplied versus provision of sanitation service in the building. The environmental impacts from RWH system connected to a combined sewer network were highly affected by the percent spillage (higher spillage = higher net impacts) while the one connected to a separate sewer were affected by the volume of rainwater supply (higher supply = lower net impacts).

While net impacts were minimum for Manchester (for both combined and separate sewer) based on supply based approach, Houston (separate sewer) and Boston (combined sewer) had minimum net impacts based on demand based approach. Though net impacts give an impression of savings, absolute impact also provide a framework for comparison of RWH systems regions among climatic regions. The regions with minimum net impacts also had minimum absolute impacts except Las Vegas which had higher net but least absolute global warming potential. The absolute environmental impacts varied significantly among regions 111 for supply based functional unit while the variation was not significant when demand based functional unit was used. Unlike expected, this study concludes that environmentally sustainable region to implement RWH system for one functional unit does not necessarily mean it will be environmentally sustainable for other functional unit. This study proposes that the demand and supply perspective should be considered while analyzing environmental impacts of RWH system as supply based system does not necessarily addressed the issue of demand.

Acknowledgements

This study was funded by National Science Foundation’s Environmental Sustainability

(grant # 1236660).

112

2.5 References

Aladenola, O. O., & Adeboye, O. B. (2010). Assessing the potential for rainwater harvesting. Water Resources Management, 24(10), 2129-2137.

Alliance for Water Efficiency, 2015. Available online from: http://www.allianceforwaterefficiency.org/1Column.aspx?id=750&LangType=1033&ter ms=water+use+in+office+building. Accessed May 2015.

Anand C and Apul D S, (2010) Cost, Energy, and CO2 Emissions Analysis of Standard, High Efficiency, Rainwater Flushed, and Composting Toilets. Journal of Environmental Management, 92(3), 419-428.

Angrill S, Farreny R, Gasol C M, Gabarrell X, Viñolas B, Josa A, and Rieradevall J, (2012). Environmental analysis of rainwater harvesting infrastructures in diffuse and compact urban models of Mediterranean climate. International Journal of Life Cycle Assessment, 17(1), 25-42.

Bare, J. (2011). TRACI 2.0: the tool for the reduction and assessment of chemical and other environmental impacts 2.0. Clean Technologies and Environmental Policy, 13(5), 687-696.

Basset-Mens, C., & Van der Werf, H. M. (2005). Scenario-based environmental assessment of farming systems: the case of pig production in France.Agriculture, Ecosystems & Environment, 105(1), 127-144.

Bronchi V, Jolliet O, and Crettaz P, (1999). Life Cycle Assessment of Rainwater use for Domestic Needs. Institute of soil and water management - Ecosystem management, CH-1015 Lausanne EPFL, Switzerland.

Cengel Y, Cimbala J, (2005). Fluid Mechanics: Fundamentals and Applications. McGraw Hill Higher Education, Boston.

Cheng C L, (2002). Study of the inter-relationship between water use and energy conservation for a building. Energy Build 34 (3), 261e266.

Commercial Building Energy Consumption Survey, 2012. Available online from: http://www.eia.gov/consumption/commercial/data/2012/. Accessed May 2015.

CostLab, 2015. Available online at: http://whitestoneresearch.com/Costlab.

Crettaz P, Jolliet O, Cuanillon J M, Orlando S, (1999). Life cycle assessment of drinking water and rain water for toilets flushing. J. Water Serv. Res. Tec. 48, 73-83.

113

Crites R, and Technobanoglous G, (1998). Small and decentralized wastewater management systems. McGraw-Hill. Deru M, Field K, Studer D, Benne K, Griffith B, Torcellini P, and Crawley D, (2011). US Department of Energy commercial reference building models of the national building stock.

Devkota, J., Apul, D., Burian, J. and Davani H. (accepted - 2015) Introducing Demand to Supply Ratio as a New Metric for Understanding Life cycle Greenhouse Gas (GHG) Emissions from Rainwater Harvesting Systems. Journal of Cleaner Production.

Devkota J, Hannah S, and Apul D S, (2015) Life Cycle Based Evaluation of Harvested Rainwater Use in Toilets and for Irrigation. Journal of Clean Production. 95 (2015): 311-321.

Devkota J, Schlachter H, Anand C, Phillips R, and Apul D S, (2013). Development and application of EEAST: A life cycle based model for use of harvested rainwater and composting toilets in buildings. Journal of Environment Management 130, 397-404.

Ecoinvent, 2010. Ecoinvent v2.2 database. Swiss Centre for Life Cycle Inventories, Dübendorf, Switzerland.

Farreny R, Morales-Pinzon T, Guisasola A, Taya C, Rieradevall J, and Gabarrell X, (2011). Roof selection for rainwater harvesting: quantity and quality assessments in Spain. Water research, 45(10), 3245-3254.

Fewkes A and Butler D, (2000). Simulating the performance of rainwater collection and reuse systems using behavioural models. Building Services Engineering Research and Technology 21, no. 2: 99-106.

Fewkes A, (1998). The use of rainwater for WC flushing: the field testing of a collection system. Building and Environment 34(6), 765-772.

Fewkes A, and Wam P, (2000). Method of modelling the performance of rainwater collection systems in the United Kingdom. Building Services Engineering Research and Technology, 21(4), 257-265.

Furumai H, (2008) Rainwater and reclaimed wastewater for sustainable urban water use. Physics and Chemistry of the Earth, 33(5), 340–346

Georgia Rainwater Harvesting Guidelines, 2009. Available online from: http://www.conservewatergeorgia.net/resources/Georgia_Rainwater_Harvesting_Guideli nes.pdf. Accessed May 2015.

114

Ghimire S R, Johnston J M, Ingwersen W W, and Hawkins T R, (2014). Life Cycle Assessment of Domestic and Agricultural Rainwater harvesting Systems. Environmental science & technology, 48(7), 4069-4077.

Ghisi E, and Mengotti de Oliveira S, (2007). Potential for potable water savings by combining the use of rainwater and greywater in houses in southern Brazil.Building and Environment, 42(4), 1731-1742.

Gleick P H, (1996). Basic water requirements for human activities: meeting basic needs. Water Int. 21 (2), 83.

Haas, G., Wetterich, F., & Köpke, U. (2001). Comparing intensive, extensified and organic grassland farming in southern Germany by process life cycle assessment. Agriculture, Ecosystems & Environment, 83(1), 43-53.

Hayashi, K. (2006). Environmental indicators for agricultural management: integration and decision making. Int J Mater Struct Reliab, 4(2), 115-127.

Imteaz M A, Shanableh A, Rahman A, and Ahsan A, (2011). Optimisation of rainwater tank design from large roofs: A case study in Melbourne, Australia.Resources, Conservation and Recycling, 55(11), 1022-1029.

Khastagir A, and Jayasuriya N, (2011). Investment evaluation of rainwater tanks. Water resources management, 25(14), 3769-3784.

Kirk S J, Dell’Isola A J, (1995). Life Cycle Costing for Design Professionals. McGraw- Hill, New York.

Krishna HJ, (2005). The Texas manual on rainwater harvesting, edition 3, Texas water development board. Available online from: http://www.ecy.wa.gov/programs/wr/hq/pdf/texas_rw_harvestmanual_3rdedition.pdf. Accessed May 2015.

Li X Y, and Gong J D, (2002). Effects of different ridge: furrow ratios and supplemental irrigation on crop production in ridge and furrow rainfall harvesting system with mulches. Agricultural Water Management, 54(3), 243-254.

Liang X, Van Dijk M P, (2011). Economic and financial analysis on rainwater harvesting for agricultural irrigation in the rural areas of Beijing. Resources, Conservation and Recycling, 55, 1100e1108.

Martínez-Blanco, J., Antón, A., Rieradevall, J., Castellari, M., & Muñoz, P. (2011). Comparing nutritional value and yield as functional units in the environmental assessment of horticultural production with organic or mineral fertilization. The International Journal of Life Cycle Assessment, 16(1), 12-26.

115

Martínez-Blanco, J., Lehmann, A., Muñoz, P., Antón, A., Traverso, M., Rieradevall, J., & Finkbeiner, M. (2014). Application challenges for the social Life Cycle Assessment of fertilizers within life cycle sustainability assessment.Journal of Cleaner Production, 69, 34-48. Mayer P W, William B D, (1999). Residential End Uses of Water. American Water Works Association Research Foundation (AWWA), Denver.

McCuen R, (2004). Hydrologic Analysis and Design, (third ed.)Pearson Education Inc, Upper Saddle River, NJ.

Mitchell V G, (2007). How important is the selection of computational analysis method to the accuracy of rainwater tank behaviour modelling?. Hydrological processes, 21(21), 2850-2861.

Morales-Pinzón, T., Lurueña, R., Rieradevall, J., Gasol, C. M., & Gabarrell, X. (2012). Financial feasibility and environmental analysis of potential rainwater harvesting systems: a case study in Spain. Resources, Conservation and Recycling, 69, 130- 140.

National Climatic Data Center, 2015 (NCDC). Available online from: http://www.ncdc.noaa.gov/monitoring-references/maps/us-climate- regions.php#references. Accessed April 2015.

National Green Building Adoption Index, 2014. Available online from: http://www.cbre.com/o/international/AssetLibrary/Green-Building-Adoption-Index- 2014_Final.pdf

PE International, 2014. Gabi database. http://www.pe-international.com.

Proenca L C, Ghisi E, Tavares DF, Coelho G M, (2011). Potential for electricity savings by reducing potable water consumption in a city scale. Resources, Conservation and Recycling, 55, 960-965.

Racoviceanu A I, and Karney B W, (2010). Life-cycle perspective on residential water conservation strategies. Journal of Infrastructure Systems,16 (1), 40-49.

Roy, P., Nei, D., Orikasa, T., Xu, Q., Okadome, H., Nakamura, N., & Shiina, T. (2009). A review of life cycle assessment (LCA) on some food products.Journal of food engineering, 90(1), 1-10.

Saving Water in Office Building, EPA, 2012. Available online from: http://www.epa.gov/watersense/commercial/docs/factsheets/offices_fact_sheet_508.pdf. Last accessed July 2015.

116

Tavakol-Davani, H., Burian, S., Shugart-Schmidt, W., Chaney, M., Apul, D., Devkota, J., (accepted) Comparing Long-Term Performance and Associated Life-Cycle Costs of Green and Gray Infrastructures to Control Combined Sewer Overflow.

TxDOT, Hydraulic Design Manual, Texas, (2009) Available online from: http://onlinemanuals.txdot.gov/txdotmanuals/hyd/hyd.pdf (accessed April 2015)

US Census, 2000. Commercial Buildings -- Number and Size, by Principal Activity. Available online from: http://www.allcountries.org/uscensus/1227_commercial_buildings_number_and_s ize_by.html

U.S. Environmental Protection Agency, eGRID Ninth edition with year 2010 data (Version 1.0). 2014.

Vargas-Parra M V, Villalba G, and Gabarrell X, (2013). Applying exergy analysis to rainwater harvesting systems to assess resource efficiency.Resources, Conservation and Recycling, 72, 50-59.

Vickers A, (2001). A Handbook of water use and conservation. Water Flow Press. Amherst: MA.

Viessman W and Lewis G L, (2003) Introduction to Hydrology, (fifthth ed.)Prentice Hall, Upper Saddle River, New Jersey

Villarreal E L, and Dixon A, (2005). Analysis of a rainwater collection system for domestic water supply in Ringdansen, Norrköping, Sweden. Building and Environment, 40(9), 1174-1184.

Wang R and Zimmerman J B, (2015). Economic and Environmental Assessment of Office Building Rainwater Harvesting Systems in Various US Cities. Environmental science & technology, 49(3), 1768-1778.

Ward S, Memon F, and Butler D, (2010). Rainwater harvesting: model-based design evaluation.

Yuan T, Fengmin L, and Puhai L, (2003). Economic analysis of rainwater harvesting and irrigation methods, with an example from China. Agricultural water management, 60(3), 217-226.

117

2.6 Supporting information: Environmental impacts from harvesting rainwater:

A comparison of supply versus demand based analysis for different climatic

regions.

Jay P. Devkota1, Harish Jeswani2, Adisa Azapagic2, Defne S. Apul1*,

1 2801 W. Bancroft, Department of Civil Engineering, University of Toledo, Toledo, OH,

43606.

2 Oxford Rd, Manchester M13 9PL, School of Chemical Engineering and Analytical

Science, University of Manchester, United Kingdom.

*Corresponding author tel: (419)-530-8132; email: [email protected]

SI – 1. Equations used to estimate the volume of potable water, wastewater and stormwater for both the functional units used in the model.

Demand based functional unit:

Potable water treatment = demand to flush the toilets – volume of rainfall capture

Wastewater treatment = wastewater from toilets to the wastewater treatment plant

= demand to flush the toilets

Stormwater treatment = roof runoff after capture in cistern (only entertained when the building is connected to a combined sewer network).

= volume of rainfall on the roof – volume of rainfall capture in the

rainwater cistern.

Avoided stormwater treatment = volume of stormwater runoff avoided by implementing rainwater harvesting system

118

= volume of rainfall capture

Avoided potable water treatment = volume of potable water avoided by implementing rainwater harvesting system

= volume of rainfall capture

Supply based functional unit:

Stormwater treatment = roof runoff after capture in cistern (only entertained when the building is connected to a combined sewer network).

= volume of rainfall on the roof – volume of rainfall capture in the

rainwater cistern.

Avoided stormwater treatment = volume of stormwater runoff avoided by implementing rainwater harvesting system

= volume of rainfall capture

Avoided potable water treatment = volume of potable water avoided by implementing rainwater harvesting system

= volume of rainfall capture.

SI – 2. Equations used to estimate the environmental impacts for both the functional units used in the model.

Demand based functional unit:

Emission from each component are estimated as the product of the mass of that components times per unit emission (from GaBi or Ecoinvent database). After estimating

119 the emissions for all the components in the construction and operational phase, the life time emissions are estimated as:

Life time emissions = emissions from construction phase + [emission from operational phase (daily)] * 365 * 75

Supply based functional unit:

Emission from each component are estimated as the product of the mass of that components times per unit emission (from GaBi or Ecoinvent database). After estimating the emissions for all the components in the construction and operational phase, the life time emissions per cubic meter of rainwater supplied are estimated as:

Life time emissions per cubic meter = {emissions from construction phase + [emission from operational phase (daily)] * 365 * 75)} / volume of rainwater supplied.

120

SI -3. Tank size versus daily precipitation

60000 y = 4162.4x2 - 9972.9x + 11072 R² = 0.8873 50000

40000

30000

20000

10000 Rainwater cistern size (gallons) size cistern Rainwater 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Average Daily Precipitation (mm)

Figure SI - 3. 1 Rainwater tank size for different regions with respective average daily precipitation. A second degree polynomial fit the data fairly well with an R square value of 0.88. A R square value close to 1 represents the best fit.

121

Table SI - 3. 1 Inventory for Supply based functional unit when the building is connected to a separate sewer network.

Regions Quantity

Scenario Item Unit TOL SLC SEA MAN L.VEG L.CRO HOU F.MYE FAR BOS Storage tank (galvanized steel) Kg 3631 2633 3631 2633 2296 4086 9326 6503 3150 4520 Pumps (steel + copper) number 1 1 1 1 1 1 1 1 1 1 Floating tank filter (plastic

cover + media) number 1 1 1 1 1 1 1 1 1 1 PVC Pipes (dual piping) ft 222 222 222 222 222 222 222 222 222 222 Overflow drain (PVC pipe) ft 5 5 5 5 5 5 5 5 5 5 Bends valves and tees (PVC) number 20 20 20 20 20 20 20 20 20 20 Transportation from plant to installation point kg-km 377974 377974 377974 377974 377974 377974 377974 377974 377974 377974

Rainwater Harvesting Concrete Pad ft3 56 35 56 35 29 66 205 126 45 76 Energy use by pump kwh 23 10 21 26 3 21 35 26 14 28 Pump replacement (every 20 years) number 3 3 3 3 3 3 3 3 3 3 Filter replacement (every 5 years) number 5 5 5 5 5 5 5 5 5 5

122

Table SI - 3. 2 Inventory for Supply based functional unit when the building is connected to a combined sewer network.

Regions Quantity

PVC Pipes (dual piping) ft 222 222 222 222 222 222 222 222 222 222 Overflow drain (PVC pipe) ft 5 5 5 5 5 5 5 5 5 5 Bends valves and tees (PVC) number 20 20 20 20 20 20 20 20 20 20 Transportation from plant to installation point kg-km 377974 377974 377974 377974 377974 377974 377974 377974 377974 377974

Rainwater Harvesting Concrete Pad ft3 56 35 56 35 29 66 205 126 45 76 Stormwater treatment gal 2088 522 3654 2610 522 2610 7830 7308 2610 2610 Energy use by pump kwh 23 10 21 26 3 21 35 26 14 28 Pump replacement (every 20 years) number 3 3 3 3 3 3 3 3 3 3 Filter replacement (every 5 years) number 5 5 5 5 5 5 5 5 5 5

123

Table SI - 3. 3 Inventory for Demand based functional unit when the building is connected to a separate sewer network.

Regions

Scenario Item Unit TOL SLC SEA MAN L.VEG L.CRO HOU F.MYE FAR BOS Storage tank (galvanized steel) Kg 3631 2633 3631 2633 2296 4086 9326 6503 3150 4520 Pumps (steel + copper) No. 1 1 1 1 1 1 1 1 1 1 Floating tank filter (plastic cover + media) No. 1 1 1 1 1 1 1 1 1 1 PVC Pipes (dual piping) ft 222 222 222 222 222 222 222 222 222 222 Overflow drain (PVC pipe) ft 5 5 5 5 5 5 5 5 5 5 Vent for RWH tank (PVC) ft 5 5 5 5 5 5 5 5 5 5 Toilets (vitrified clay) lbs 223 223 223 223 223 223 223 223 223 223

Toilet flush tank with accessories (vitrified clay) lbs 101 101 101 101 101 101 101 101 101 101 Sewer drain pipe (PVC) ft 100 100 100 100 100 100 100 100 100 100 Bends valves and tees (PVC) No. 20 20 20 20 20 20 20 20 20 20 Transportation from plant to kg- installation point km 377974 377974 377974 377974 377974 377974 377974 377974 377974 377974

Rainwater Harvesting Concrete Pad ft3 56 35 56 35 29 66 205 126 45 76 Potable water treatment gal 28709 41758 30275 25055 49588 30275 15659 25577 37583 23489 Wastewater treatment gal 52198 52198 52198 52198 52198 52198 52198 52198 52198 52198 Energy use by pump kwh 23 10 21 26 3 21 35 26 14 28 Pump replacement (every 20 years) No. 3 3 3 3 3 3 3 3 3 3 Toilet replacement (every 35 years) Kg 446 446 446 446 446 446 446 446 446 446 Filter replacement (every 5 years) No. 5 5 5 5 5 5 5 5 5 5 124

Table SI - 3. 4 Inventory for Demand based functional unit when the building is connected to a combined sewer network.

Regions

Toilets (vitrified clay) lbs 223 223 223 223 223 223 223 223 223 223 Toilet flush tank with accessories (vitrified clay) lbs 101 101 101 101 101 101 101 101 101 101 Sewer drain pipe (PVC) ft 100 100 100 100 100 100 100 100 100 100 Bends valves and tees (PVC) No. 20 20 20 20 20 20 20 20 20 20 Transportation from plant to kg- installation point km 377974 377974 377974 377974 377974 377974 377974 377974 377974 377974

Rainwater Harvesting Concrete Pad ft3 56 35 56 35 29 66 205 126 45 76 Potable water treatment gal 28709 41758 30275 25055 49588 30275 15659 25577 37583 23489 Wastewater treatment gal 52198 52198 52198 52198 52198 52198 52198 52198 52198 52198 Stormwater treatment gal 2088 522 3654 2610 522 2610 7830 7308 2610 2610 Energy use by pump kwh 23 10 21 26 3 21 35 26 14 28 Pump replacement (every 20 yrs) No. 3 3 3 3 3 3 3 3 3 3 Toilet replacement (every 35 yrs) Kg 446 446 446 446 446 446 446 446 446 446 Filter replacement (every 5 yrs) No. 5 5 5 5 5 5 5 5 5 5

125

Table SI - 3. 5 Percentage change in life cycle environmental impacts due to percentage change in precipitation across various geographical locations – supply based functional unit. Las Vegas (with lowest mean monthly ppt) was taken as reference; percentage change in impacts were calculated with respect to increase from Las Vegas. A positive number indicates decrease.

% % change in environmental impact – supply based functional unit change Ozone in ppt Average Acidificatio Eutrophicati Depletion Ecotox Daily Energy GWP ) n Potential on Potential Potential Ecotox Air Ecotox Soil Water Region ppt Com Sep Com Sep Com Sep Com Sep Com Sep Com Sep Com Sep Com Sep Toledo 2.5 89 70.9 75.6 46.7 45.7 49.5 48.6 39.9 85.7 63.0 77.6 39.8 14.5 55.3 12.4 40.0 84.5 Salt Lake City 1.1 75 65.2 39.5 42.0 74.2 67.1 12.6 10.7 73.0 Seattle 2.56 89 61.0 74.5 42.2 45.1 43.4 47.9 18.4 84.4 36.8 76.4 15.7 14.2 16.6 12.2 19.2 83.2 Manchester - UK 2.97 91 74.2 81.0 49.5 49.2 52.2 52.3 30.9 92.5 62.5 83.3 38.0 15.8 51.7 13.3 30.8 91.0 Las Vegas 0.28 0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 La Crosse 2.46 89 65.1 72.7 43.5 43.8 45.6 46.5 19.0 82.0 51.9 74.3 30.2 13.7 40.3 11.8 18.7 80.9 Houston 4.43 94 66.8 39.7 42.1 74.1 66.0 11.1 10.1 73.6 Fort Myers 3.38 92 43.0 67.6 32.7 40.4 31.5 42.8 95.5 75.3 1.9 68.0 18.4 11.9 37.0 10.6 97.6 74.6 Fargo 1.72 84 69.4 41.9 44.5 78.5 71.2 13.3 11.4 77.4 Boston 3.1 91 70.7 75.6 46.5 45.6 49.2 48.4 38.6 85.4 62.2 77.3 39.1 14.3 54.3 12.3 38.5 84.2 Maximum % change 74.2 81.0 49.5 49.2 52.2 52.3 95.5 92.5 62.5 83.3 39.1 15.8 55.3 13.3 97.6 91.0

126

Table SI - 3. 6 Percentage change in life cycle environmental impacts due to percentage change in precipitation across various geographical locations – demand based functional unit. Las Vegas (with lowest mean monthly ppt) was taken as reference; percentage change in impacts were calculated with respect to increase from Las Vegas. A positive number indicates decrease.

% change in environmental impact – demand based functional unit Ozone % Average Acidificatio Eutrophicati Depletion Ecotox Ecotox change Daily Energy GWP n Potential on Potential Potential Ecotox Air) Soil) Water in ppt Region ppt Com Sep Com Sep Com Sep Com Sep Com Sep Com Sep Com Sep Com Sep Toledo 2.5 89 14.2 16.0 11.0 9.2 1.3 0.5 0.9 2.0 11.0 13.1 9.2 11.2 2.7 0.3 8.3 10.6 Salt Lake City 1.1 75 6.1 3.2 0.4 0.7 5.0 4.2 0.1 3.9 Seattle 2.56 89 11.3 14.7 12.3 8.6 3.3 0.4 3.9 1.9 8.0 12.1 6.3 10.4 5.7 0.2 5.2 9.8 Manchester - UK 2.97 91 17.2 19.5 11.6 9.2 0.7 1.8 1.7 2.2 12.8 15.6 10.5 13.2 3.6 0.3 9.2 12.3 Las Vegas 0.28 0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 La Crosse 2.46 89 12.2 14.4 11.7 9.2 2.5 0.0 1.9 1.9 9.3 12.1 7.6 10.4 3.7 0.2 6.8 9.8 Houston 4.43 94 22.9 21.4 3.8 3.9 20.6 18.0 0.4 17.7 Fort Myers 3.38 92 9.5 16.7 22.1 14.1 9.9 1.9 9.7 2.7 5.9 14.7 4.1 12.8 12.5 0.3 2.6 12.5 Fargo 1.72 84 9.1 5.4 0.2 1.2 7.5 6.4 0.2 6.1 Boston 3.1 91 17.3 19.6 14.6 12.2 2.3 0.1 1.3 2.6 13.6 16.4 11.3 14.0 3.6 0.3 10.2 13.3 Maximum % change 17.3 22.9 22.1 21.4 9.9 3.8 9.7 3.9 13.6 20.6 11.3 18.0 12.5 0.4 10.2 17.7

127

Table SI - 3. 7 Main characteristics of the precipitation and its effect on RWH system for different climatic regions.

daily

y (m3)

Region

Maximum daily ppt (mm) Avg daily ppt Daily demand (gallons per day) Vr (%) % spillage Volume of rainwater suppl Volume of spillage (m3) tank size (m3) Tank size/rwh supply (unit less) Volume of spillage/volume of rainwater supply (unit less) % days without rain D/S ratio Toledo 800 2.54 1716 45 4 2.9 0.3 47 16 0.09 63 2 SLC 472 1.1 1716 20 1 1.3 0.1 28 22 0.05 73 4.8 Seattle 1026 2.56 1716 42 7 2.7 0.5 47 17 0.17 56 2 Manchester 516.2 2.97 1716 52 5 3.4 0.3 28 8 0.10 10 1.8 Las Vegas 419 0.28 1716 5 1 0.3 0.1 19 58 0.20 94 16.7 La Crosse 1928 2.46 1716 42 5 2.7 0.3 57 21 0.12 65 2.1 Houston 4973 4.43 1716 70 15 4.5 1.0 265 58 0.21 75 1.2 Fort Myers 1570 3.38 1716 51 14 3.3 0.9 76 23 0.27 65 1.5 Fargo 1179 1.72 1716 28 5 1.8 0.3 38 21 0.18 70 3 Boston 1090 3.1 1716 55 5 3.6 0.3 66 18 0.09 65 1.7

128

Chapter 4

Environmental impacts from Implementing Rainwater Harvesting System at a Suburban Neighborhood: A comparison of centralized versus decentralized tank location and pervious to impervious area ratio

Jay P. Devkota1, Anthony Dietrich1, Rahul Yarlagadda2, Cyndee Gruden1, Steven J.

Burian3, Defne S. Apul1*

1 2801 W. Bancroft, Department of Civil Engineering, University of Toledo, Toledo, OH,

43606

2 2801 W. Bancroft, Department of Mechanical Engineering, University of Toledo,

Toledo, OH, 43606

3110 S. Central Campus Dr, Department of Civil & Environmental Engineering,

University of Utah, Salt Lake City, UT, 84112

*Corresponding author tel: (419)-530-8132; email: [email protected]

129

4.1 Introduction

Most of the Northeastern, Pacific Northwest and the Great Lakes regions of the

United States have combined sewer systems to transmit sewage to the treatment facilities

(U.S. EPA 2008). In communities served by combined sewer systems, an estimated 3.2 billion cubic meters of diluted sewage in the form of CSO (combined sewer overflow) is released to receiving waters in the US every year (USEPA, 2004). CSOs are one of the greatest challenges to meet water quality standards (U.S. EPA 2014a). Since the very beginning, stormwater infrastructures have been defined and constructed as a part of the built environment (Choguill 1996; Hanson 1984) to store and then detain the runoff (Burns et al., 2012). Nowadays with the increasing awareness towards green infrastructures, rainwater harvesting (RWH) systems are becoming a partial alternative to green building construction (Reidy, 2008). RWH, which is designed to capture and store the runoff from the roof, follows low-impact development based stormwater management principles seeking to recreate the natural hydrologic cycle (Dietz, 2007). RWH systems have been used to reduce stormwater runoff volumes. For example, Steffen et al. (2013) reported the potential for a reasonable implementation of rainwater harvesting to reduce up to 20% of stormwater runoff in most regions of the US.

Most RWH system studies have been done at the building scale, and very few have been done at catchment or city scale. While building scale RWH system research has focused on toilet flushing (Bronchi et al., 1999; Fewkes, 1998 and Furumai, 2008; Anand and Apul, 2010; Ghimire et al., 2014; Devkota et al., 2013; Angrill et al., 2012), lawn irrigation (Li and Gong, 2002 and Stout et al., 2015), laundry (Bronchi et al., 1999 and

Angrill et al., 2012) and vehicle cleaning (Ghisi et al., 2009 and Villarreal et al., 2005), 130 larger scales were mostly focused on agriculture and irrigation (Bruins et al, 1986; Yuan et al 2003; Li and Gong, 2002; Pachpute et al., 2009; Helmreich and Horn, 2009 and Zhu et al, 2004). The scale of the RWH installation affects its feasibility, both economically and environmentally. For example, Farreny et al., (2011) reported that RWH systems installed at a neighborhood scale have proven to be more economically feasible as compared to the building scale. This was because city scale RWH systems perform better when combined with other grey infrastructures. For example, RWH systems combined with a storage tunnel were reported to be economically and hydrologically efficient when implemented at the city scale in city of Toledo, Ohio (Tavakol-Davani et al., 2015 - accepted).

The capacity of cistern used in the analysis has significant effect in stormwater reduction. RWH cisterns installed in one third of properties within a 23 hectare watershed were able to prevent combined sewer overflow during normal rain events but were too small to prevent overflow during heavy rain events (Petrucci et al., 2011). The stormwater volume diverted from the combined sewer was also not significant with the randomly selected size of the cistern for the city of Toledo (Tavakol-Davani et al., 2015 - accepted).

Therefore it is important to use the accurately designed rainwater harvesting cistern in the analysis. We argue that the environmental feasibility of RWH systems depends mainly on the volume of stormwater diverted from the combined sewer because our previous studies have shown that stormwater treatment and potable water treatment were the major contributors to the environmental impacts (Devkota et al., 2015 – accepted). We therefore used long term precipitation to incorporate the effect of extreme precipitation on cistern sizing to better perform during extreme rainfall events.

131

The savings, in environmental emissions, from implementing RWH system depends on the density of the neighborhoods. Density of neighborhood can be defined both in terms of spacing of the houses as well as building occupation. Angrill et al., (2012) reported that the environmental impacts of using RWHS in a compact urban density neighborhood were lower than those in the diffuse density neighborhood. Morales-Pinzón et al. (2012) also stated that optimum scale of RWH systems was at high density urban developments. The definition of compact density neighborhood varies with the country and region. In United States, compact density neighborhood refers to a minimum of four units per acre for single family development and eight units per acre for multifamily (EOHED,

2015). Densely occupied building was also found to be environmentally attractive. For example, savings in environmental impacts were higher when the demand to supply ratio was equal to one or greater (Devkota et al., 2015 - accepted).

In United States, apartments consist of around 27.8% of the total housing supply in urban settlements and 40.3% of the housing supply in suburban settlements of more than

100,000 residents (APA, 1960). With the increase in urban population, this percent is expected to increase. Indoor water use in residential apartment buildings is high as compared to commercial and educational buildings (Vickers, 2001). Additionally, the decision to implement RWH system in these neighborhoods would be easier as there would be less people involved in the management as compared to individual residential houses.

Therefore, this research is aimed at identifying the potential benefits of RWH system at suburban neighborhoods consisting of multifamily apartment complex.

Since multifamily apartment complexes are in close proximity with each other, one

RWH cistern could be used for one building or for multiple buildings. The choice of

132 centralized and decentralized rainwater cistern also depends on the length of pipes and space availability. Previous research has shown that centralized tanks were more attractive than decentralized from a water saving perspective, when RWH system was implemented in a high density neighborhood (Vargas-Parra et al., 2013). We argue that the use of RWH system not only saves the potable water but also diverts stormwater from entering the combined sewer network. If not diverted, the stormwater would require treatment at the wastewater treatment plant thereby increasing the environmental emissions. We hypothesize that the volume of stormwater diverted from the combined sewer has huge impact on the emissions. Though the centralized scenario is attractive from a water savings perspective, it would be interesting to see the nature of savings from both water savings as well as stormwater savings (diverted stormwater) perspective for centralized and decentralized scenario.

Another parameter that affects the centralized or decentralized cistern is the mass of the cistern. The greater the mass of the rainwater harvesting components, greater would be the emissions from the transportation. Though transportation has less impact on global warming, acidification and eutrophication potential, it has higher contribution to ozone depletion potential. We argue that the weight of the cistern from decentralized scenario would be higher than the centralized scenario provided that the two scenarios have same capacity.

Therefore, the objective of this research was to analyze the effect of centralized and decentralized RWH cistern on the stormwater reduction at suburban neighborhood by accessing both the hydrologic and environmental performances. Knowing that the soil types are site specific, a sensitivity analysis was performed by varying the soil type across

133 the neighborhood to make the results representative on a larger scale. Also, a compact density neighborhood would have lesser pervious area while a diffuse density neighborhood would have higher pervious area. To incorporate the differences in results caused by the compact and diffuse density neighborhoods, a sensitivity of pervious area to impervious area was also performed. Our research focuses on toilet flushing as the only end use because water use in buildings accounts for about 11% of fresh water withdrawals in the US (Barber, 2009) and 27% of indoor water use goes to toilet flushing (Mayer et al

1999 and Gleick, 1996). In addition, the toilet usage does not show large daily variability

(Fewkes, 2000).

4.2 Methodology

Previous research has reported that the methodology of integrating GIS and stormwater model to estimate stormwater runoff can be applied to large watersheds (Barco et al., 2008). Using this concept, the first task in developing a suburban neighborhood to implement RWH system was to develop a calibrated model in PCSWMM. The result from this model (stormwater runoff) was then entered in the LCA model (Devkota et al., 2013) to estimate environmental emissions. Obtained results were then compared with pre-RWH implementation scenario (referred as Business as Usual or BAU) to access its environmental feasibility.

A schematic representation of the hydrologic-LCA model is presented in Figure

4.1. The land use characteristics data were gathered from the United States Department of

Agriculture: National Resource Conservation Service’s Web Soil Survey. Topographic

134 data such as roof area of the buildings, parking lot area, and pervious (grass) area were obtained from Auditors Real Estate Information System developed by the Lucas County

Auditor. Using the building characteristics (building type, roof area and number of expected occupancy) and Toledo precipitation data, the size of the rainwater cistern was estimated using Yield After Spillage approach (explained in section “4.2.4” of methodology). Along with the cistern, other rainwater harvesting components such as filter, pump, dual piping and concrete pad were estimated according to Devkota et al.

(2013) and Devkota et al. (2015). Estimated cistern size was then used in PCSWMM to estimate the runoff after implementing a RWH system in the Oak Hill Court neighborhood.

Three scenarios were developed for decentralized, partly centralized and centralized scenarios, respectively. The results from hydrologic simulation (stormwater runoff) from each respective scenario were then used in the LCA model to estimate the environmental impacts from RWH system at neighborhood scale for various centralized and decentralized cistern location combinations.

135

Precipitation of Toledo

Watershed Buildings selection

classification Pervious and impervious area Roof Area Occupancy

Hydrologic simulation using SWMM Tank Sizing based on daily Storage node used to represent rainfall cistern

Tank location Demand specified using negative time series

Runoff before and after RWH System design – water used installing RWHS to flush the toilets

Quantity estimation (cistern, concrete pad, potable water, wastewater, stormwater runoff, energy use and others...)

LCA (using GaBi database)

Construction phase Operational phase - Cistern -Energy use by pump -Pump -Potable water treatment -Filter -Wastewater and -Dual pipes stormwater treatment -Concrete pad -Replacements: toilet, -Bends valves, tees filter, pump. -Vent pipe -Under drain -Toilet accessories

Life cycle environmental impacts (GWP, AP, EP, ODP, ET, and Energy) Figure 4. 1 Schematic representation of conceptual hydrologic – LCA model used for RWH system at neighborhood scale

136

4.2.1 Base watershed

For this study, a suburban residential apartment complex from Toledo, Ohio was chosen (Figure 4.1). This area is well suited for explaining the environmental feasibility of

RWH system in a suburban neighborhood. The 81,580 m2 Oak Hill Court neighborhood is located next to the University of Toledo Medical Center and Southland Shopping Center.

The small watershed discharges its runoff into a combined sewer system. The soil types in this neighborhood varied from sand to sandy loam (Table SI-4.1 and Figure SI-4.1) (Web

Soil Survey, United States Department of Agriculture: National Resource Conservation

Service). The soils classification varied from sisson loam to colwood loam as classified by

Web Soil Survey. The neighborhood had a slope of 0 to 2% on average with the capacity of the most limiting layer to transmit water varying from moderately high to high (0.60 to

2.00 in/hr). Small to large depressions storages were also observed in the neighborhood with the depth ranging from 0 to 2 inches. This neighborhood was selected as a representative of the suburban neighborhood in the United States. Knowing that the soil types are site specific, a sensitivity analysis was performed by varying the soil type across the neighborhood to account for the spatial variability of soil type. The results were generated for both sand (poor soil) and clay (good soil). Also, a compact density neighborhood would have lesser pervious area while a diffuse density neighborhood would have higher pervious area. To incorporate the differences in result caused by the compact and diffuse density neighborhoods, a sensitivity of pervious area to impervious area was also performed.

137

4.2.2 Building classification

There were 14 residential buildings in the watershed consisting of 292 apartments with 81,580 m2 roof area and 828 numbers of total occupants. Out of 14 buildings, 9 buildings have the roof area of 743 m2 with 72 occupants (Type A) each while rest of the

5 buildings has 372 m2 roof area with 36 occupants each (Type B). The Oak Hill neighborhood consists of large grass areas (49,907 m2), significant parking lot area (19,055 m2), one swimming pool (which does not contributes to the drainage), and one tennis court, in addition to the 14 buildings. The ratio of impervious (buildings and parking lots) to total area was 64%, while the ratio of building rooftop to the total area was found to be 11%.

The ratio of pervious to impervious area ratio is 1.8. Building characteristics were collected from satellite imagery and Oak Hill Court Apartment management.

4.2.3 Demand and supply

The water demand in the neighborhood was assumed to equal the water needed to flush the toilets. The water required to flush the toilets was estimated using the building occupancy, flushes per person per day (5.1 flush per person per day for residential buildings) (Vickers 2001), and standard flush volume for standard toilets (1.6 gallons per flush). The daily water supply was estimated by multiplying the daily precipitation in

Toledo by the roof area of the building assuming that 25% (roof runoff coefficient – 0.75) of the rainwater might be lost before entering the cistern (Krishna 2005). Since rainwater is variable with time and space, the supply of rainwater is not always sufficient to meet the

138 water demand to flush the toilets. Therefore, additional water was assumed to be supplied from the municipal supply in that case.

4.2.4 Cistern sizing and cistern location

The rainwater cistern was sized using Yield After Spillage (YAS) approach and long term daily precipitation. YAS approach was used as it gives a conservative estimate of tank yield (Mitchell 2007 and Fewkes and Butler 2000). The use of one year rainfall data reduces the accuracy of the estimated yield of the estimated rainwater cistern

(Mitchell, 2007). Mitchell further stated that use of 10 year precipitation data or more does not significantly affect the yield. Therefore, in this research, we used 15 years of daily precipitation (from the year 2000 to 2015) data to size the rainwater cistern.

Volumetric reliability (Vr) of a rainwater cistern, which is also a measure of the water saving efficiency of the tank, is used to size the tank. Vr can be calculated using:

Vr = volumetric reliability = ∑Yt / ∑Dt……………………………….….. 4.1

Where,

Yt = Volume of rainwater supplied (m3)

Dt = Daily water demand (m3).

In the YAS approach, Yt and Dt are calculated using:

Yt = MIN ( Dt , Vt-1 + It ) ……………………………………………….... 4.2

Vt = MIN (Vt-1 + It – Yt , C – Yt ) ...……………………………………… 4.3

Where,

Vt = volume of rainwater in the tank at the end of time step t (m3)

139

Vt-1 = volume of rainwater in the tank at the end of previous time step t (m3)

It = inflow or roof runoff (m3)

C = Capacity of the rainwater tank (m3)

Daily spillage (St) is calculated from the daily mass balance equation (equation 4)

Vt = Vt-1 + It + Pt – Et – St – Lt - Yt ……………………………...…….. .4.4

Where:

Pt = incident precipitation received by the tank (m3)

Et = evaporation (m3)

St = amount of spillage due to overflow from tank (m3)

Lt = seepage or leakage (m3)

Evaporation (Et), Incident precipitation (Pt), and Leakage (Lt) are very small and are not considered in this study.

A series of the rainwater cisterns were assumed. Based on the initial storage volume

(Vt−1), rainfall inflow on the first day (It) and daily water demand (Dt), the volume of rainwater supply (Demand (Yt)) was calculated for the first time step t using equation 4.2.

After usage on first day, the volume of water remaining in the cistern at the end of time step t also called Vt was calculated using equation 4.3. Daily spillage was calculated by comparing the daily demand and supply using Equation 4.4. The initial storage volume

(Vt−1) for the second time step would be the volume of water at the end of first time step

(Vt). Similar to the first time step, Yt and Vt for the second time step were then estimated using Equations 4.2 and 4.3 respectively and so on. Volumetric reliability of the rainwater tank was estimated by dividing long term daily supply with long term daily demand. The volumetric reliabilities of a series of rainwater tanks assumed in the first step were then

140 assessed using Equation 4.1. The size of the rainwater tank was considered optimum when any increment in the tank size only changes the volumetric reliability by 1% or less. A cistern made of galvanized steel was selected for this study. Considering the large cistern size that might result from buildings with large roof areas, a series of interconnected rainwater cisterns were assumed in the case when storage capacity exceeded 100 m3.

Three scenarios were considered to optimize the location and tank size. The first approach considered one tank per building (decentralized approach) (Figure 4.2 – a) while the second approach was partially centralized where two or more buildings were connected to one rainwater cistern depending on their location (Figure 4.2 - b). The third approach considered one centralized tank (Figure 4.2 – c). In each of the scenarios, the tanks were connected to the toilets of each building by additional piping networks. The length of dual piping, cistern and the energy used by pump in these three approaches were different and were calculated separately. Though the dual piping does not have significant impact on life cycle GHG emissions, the cistern and energy used by the pump will have a significant impact. An optimization was carried out to observe which scenarios had the least life cycle environmental impacts.

141

Figure 4. 2 Schematic representation of (a) decentralized RWH tank, (b) partly centralized RWH tank and (c) centralized RWH tank locations.

4.2.5 Dual piping calculation

Dual piping is required to supply the rainwater from the rainwater cistern to the toilets. As the location of tank was different, length of dual piping was also different for each scenario. For the decentralized scenario, length of dual piping was calculated by assuming that one primary pipe runs all the way to the top floor and that each toilet was approximately 3 ft away from this pipe. In addition, the horizontal length of pipe from the cistern to the main pipe was assumed to equal to the width of the building. In the case of partly centralized and centralized scenario (Figure 4.2 b and 4.2 c), the length of dual piping was equal to the length of piping inside the buildings (same as decentralized) plus the

142 length of piping from the building to the cistern location. The pipe made of PVC was assumed for this study.

4.2.6 Pumping design

The cisterns for all scenarios were assumed to be located on the ground. Therefore a pump was required to transport the rainwater from the cistern to the toilets in respective floors. Energy delivered to pump was estimated using following equation:

P = ϒ x Q x (he + hp) x [1+α]/η ……………………………………………4.5

Then the pump’s life cycle energy requirements are calculated as:

E = P x 365 [days/year] x 24 [hrs/day] x 0.001 [kW/W] x 75 years………..4.6

Where,

P = Power delivered to Pump (W)

E = Annual energy required by Pump (kWh)

Q = Flow rate (m3/sec). h = he + hp = sum of elevation head and pressure head provided by the pump (m).

η = combined mechanical and electrical efficiency of the pump (assumed 65 %; Cengel and Cimbala 2005).

ϒ = Specific weight of water (N/m3)

α = Percentage of energy lost due to friction (assumed 0.3%; Cheng 2002).

The above equations (equation 4.5 and 4.6) are valid for all the centralized, partly centralized and decentralized scenarios. Pressure head provided by the pump (hp) was set

143 equal to the pressure provided by the city (assumed 35.2m of water or 50 psi). Elevation head provided by pump (he) was set equal to the building height.

The energy use by pump was different for different rainwater supply. For example, in the decentralized scenario, type A and type B building had different rainwater supply and had different energy use. In the case of partly centralized and centralized scenario, energy use was also different because each scenario had different rainwater supply from the group of buildings serving the storage node.

4.2.7 Hydrologic simulation

The hydrologic analysis in watershed scale requires simulating the impacts of implementing RWH system on potable water reduction and stormwater reduction at the urban watershed scale. A Storm Water Management Model (PCSWMM) was used to simulate the urban runoff considering infiltration, evapotranspiration, and stormwater runoff. PCSWMM simulates hydrologic processes and hydraulic transport in urban environments. PCSWMM calculates the infiltration and surface storage of water at a sub- hourly time step and routes the rest as sheet flow using the non-linear reservoir algorithm.

The sheet flow is then routed to storm drain inlets and then to the discharge point using an implicit solution to the coupled one dimensional unsteady Saint- Venant equations (Chow et al 1998). A schematic of how PCSWMM fits in the analysis is presented in Figure 4.3.

Before implementing RWH system, all the rainwater collected from the roof goes to the downspout and either gets infiltrated or discharged into the storm sewer. After implementing RWH system, the rainwater collected from the roof is collected into the

144 rainwater cistern and only the overflow is discharged into the parking lot or grass. If the overflow is discharged into the parking lot, then it enters the storm sewer while the overflow discharged into the grass gets infiltrated and, depending on the infiltration capacity of the soil, some will be converted to runoff.

Figure 4. 3 Schematic of how SWMM fits in the analysis.

The satellite image was first digitized in ArcGIS and imported to PCSWMM with respective attribute data (ie. area of the sub-catchments, manning’s roughness, slope, conductivity and depression storages). After importing, the flow paths were defined in such a way that the overall neighborhood discharges the runoff to the combined sewer running parallel to the road in front of the neighborhood. The rainfall runoff in the neighborhood

145 was then estimated using the long term daily rainfall data from 2000 to 2014. Recent precipitation was included to incorporate any possible effect of climate change in the precipitation. To estimate the stormwater reduction by implementing RWH system, a storage node was assigned to each building. We used storage nodes to represent the rainwater cistern with the provision of overflow at the height equal to the height of the cistern (size of cistern obtained using YAS approach – refer section “4.2.4” of methodology). The overflow was then routed back to the sub-catchment adjacent to the building. With the idea that, change in height of water in the storage node changes the volume of water coming out of the orifice located at the bottom of the storage node, the storage node is not able to supply constant demand even if the volume of water is enough to meet the demand. Therefore, a constant negative time series was used to represent the demand to flush the toilets. This estimated negative time series was assigned to the storage node.

4.2.8 Centralized vs decentralized

Three scenarios were defined in order to optimize the size and location of rainwater cistern: decentralized, partly centralized and centralized. In the case of decentralized scenario (Figure 4.2 a), an individual storage node was assigned to each building with the demand equal to water demand to flush the toilets for the occupants of that particular building. For partly centralized (Figure 4.2 b), three storage node locations were assumed.

The first storage node served eight buildings out of which six were type A and two were type B. The second storage served four buildings out of which three were type A and one

146 was type B. The third node served two type B buildings (Figure 4.2 b). All the three storage nodes were sized separately using the daily precipitation, accumulated roof area, and occupancies of the served buildings. One storage node was assumed for the centralized scenario that served 9 type A buildings and 5 type B buildings (Figure 4.2 c). Water demand for each storage node was estimated by multiplying the total number of occupants served, by that particular storage node, by the per person water use for residential buildings (5.1 flush per person per day) (Vickers, 2001).

4.2.9 Sensitivity of soil type

Due to the spatial variation of soil, different neighborhoods have different soil types.

The results from this neighborhood might not be applicable to other neighborhoods in the

United States. Therefore, a sensitivity analysis was performed using the poor (low drainage) and good (high drainage) soil type to better represent other neighborhoods. Sand and Clay were assumed along with the existing soil type to estimate the effect of soil type on the stormwater runoff in a suburban neighborhood.

4.2.10 Sensitivity of pervious to impervious area ratio

Allocation of pervious (parking lots) and impervious areas (grass) in any suburban neighborhood depends mostly on the space availability and less on the design guidelines.

The runoff from the neighborhood depends on the pervious and impervious area. Pervious areas (grass) are highly likely to either hold the water or infiltrate while impervious areas

147 convert rainfall directly to surface runoff. Higher impervious areas produce higher runoff after the rainfall event.

International Building Code (IBC, 2006) and Deru et al. (2011) provided a guideline for building design based on space requirement, ventilation requirement, and energy efficiency. However, such designed building roof areas are not always sufficient to supply enough rainwater to meet urban water demands. Therefore, a hypothetical scenario was assumed where the building roof area was increased keeping the occupancy constant.

The roof area was increased until the supply of rainwater met toilet flushing demand. Once the demand was met, the rest of the collected rainwater overflowed from the cistern as surface runoff. It is important to estimate the effect of pervious and impervious area proportion in the neighborhood to minimize the surface water runoff as well as the environmental impacts associated with it. Therefore a sensitivity analysis was performed by increasing the roof area and decreasing the grass area to access the relation between surface runoff and the pervious to impervious area ratio. Parking lot area was kept constant considering the constant number of occupants would need the same parking space independent of the roof area. Only decentralized scenario was considered in the sensitivity analysis because our preliminary result showed no significant difference in runoff between decentralized and centralized scenario (see section “b” of results and discussion section).

4.2.11 Life cycle assessment

In this research, environmental impacts of implementing RWH systems in a suburban neighborhood was estimated and compared with standard municipal based water supply system using life cycle assessment (LCA). The modeling approach was partly

148 adapted from the EEAST model (Devkota et al. 2013). This model first designs the RWH system based on building characteristics and long term daily precipitation using Yield After

Spillage approach. Using the size of RWH components and the runoff from PCSWMM model (explained in section “4.2.7” of methodology), environmental impacts (energy use, global warming, acidification potential, eutrophication potential and ozone depletion potential) were estimated using LCA data from the GaBi (PE International 2014) and

Ecoinvent (2010) databases (Figure 4.1) assuming 75 years as the life time of the RWH system. TRACI was used as impact assessment method. US data were used whenever available for the RWH components. In cases where US data were not available (potable water treatment, wastewater treatment and concrete pad), the original electricity mix

(European) was replaced with the Eastern US grid mix with an assumption that the processes were similar.

The system boundary was considered from “cradle to grave” consisting of construction, operation, maintenance and decommissioning phase of RWH system and was selected in accordance to Devkota et al (2015 - accepted). The functional unit of this study was “stormwater runoff from a suburban neighborhood from implementing RWH system to meet the toilet flushing demand”. Additional potable water from municipal supply is required if the supply of rainwater is not enough to meet the toilet flushing demand. The construction phase includes the construction, installation of RWH components (cistern, concrete pad, dual piping, filter, pump, bends-tees-valves, toilets, flush accessories, vent pipe, overflow pipe and underdrain) as well as transportation of those components to the installation point. The operational phase includes the energy use by pump, potable water required (in case when rainwater supply is not enough), wastewater treatment, stormwater

149 treatment and the replacement of RWH components such as filter, pump, toilets and flush accessories. A detailed inventory of items in the construction and operational phase of

RWH system for different building types and scenarios are presented in supporting information (Table SI – 4.2, 4.3, 4.4, 4.5, 4.6 and 4.7. The replacement components such as filter, pumps, and toilets were assumed to be replaced every 5, 20, and 35 years, respectively (Kirk and Dell’Isola 1995).

4.3 Results and Discussion

4.3.1 Cistern Sizing

Table 4. 1 Rainwater cistern sizes for different scenarios. There were nine, type A; five, type B; eight, location 1; four, location 2; two; location 3 and one, centralized cisterns.

Scenario Decentralized Partly Centralized Centralized Type Type Total Loc 1 Loc 2 Loc 3 Total Cent (1) A (9) B (5) (8) (4) (2) Cistern size 38 28 483 189 151 38 370 473 (m3) Volumetric 61 63 61.7 59 62 61 60.1 62 reliability (Vr) (%) Stormwater 27 8 27 reduction (%)

With the volumetric reliabilities of all the scenarios being less than 100%, the installed RWH system in the neighborhood was not able to supply enough rainwater to

150 meet toilet flushing demand (Table 4.1). The total capacity of the decentralized scenario was 483 cubic meters with the cistern for type A building being 38 cubic meters each and for type B being 28 cubic meters. Type A buildings (with roof area 743 m2) resulted in cistern size equal to 38 m3 with volumetric reliability of 61% whereas type B (with roof area 371 m3) resulted in cistern size equal to 28 m3 and 63% volumetric reliability (Table

4.1). The result showed that doubling the roof area did not necessarily double the rainwater cistern size and volumetric reliability. This was because, with the increase in roof area by two times, the daily capture of rainwater from the roof as well as the demand to flush the toilets also increased by twice (because occupancy increased by twice). However, more daily spillage was observed (as a result of doubled roof area) rather than increases in the rainwater supply resulting from the larger size of cistern. Higher volumes of rainwater could be captured during intense rain events but frequent spillage was also observed because the cistern could not hold rainwater higher than its storage capacity. This result showed that cistern size did not vary linearly with the increase in roof area or occupation

(Figure SI – 4.2, and SI – 4.3). On the other hand, volumetric reliability increased with the increase in roof area or population until 100% and becomes constant after that. Any increase in roof area or occupancy beyond 100% did not increase the volumetric reliability.

However, the cistern capacity was found to decrease with the increase in roof area (Figure

SI – 4.2) or population (Figure SI – 4.3) beyond 100% reliability. Similarly, the capacities of the cistern at location 1, 2 and 3 (Figure 4.2), for the partly centralized scenario, were found to be 189 m3, 151 m3 and 38 m3 with volumetric reliabilities of 57, 59 and 61% respectively (Table 4.1). The total capacity of the cistern for the partly centralized scenario was 370 m3 which was about two third of the cistern capacity for decentralized scenario.

151

This was because of the nonlinear behavior of the cistern capacity with the increase in roof area and population (Figure SI 4.4). However, the overall volumetric reliability of the

RWH system did not change significantly within the decentralized and partly centralized scenarios (Figure 4.1). No defined pattern was found between the roof area, population, and the cistern capacity indicating that a linear increase in roof area or population does not necessarily increase the cistern capacity linearly. Finally, the cistern size for the centralized scenario (473 m3) was found to be approximately equal to the total cistern size from the decentralized scenario (483 m3) with similar volumetric reliability (Table 4.1).

4.3.2 Stormwater Runoff and associated Environmental Impacts

4.3.2.1 Comparison with BAU

The result showed that the volume of stormwater runoff before and after implementing RWH system in a neighborhood depends on the building downspout connectivity. A downspout may be connected either to the pervious areas, such as grass, the impervious areas or a storm drain. Different results were reported when the building downspout was connected to pervious or impervious areas.

4.3.2.2 Effect of downspout connection

Downspout connected to pervious areas

(a) Clay type – sand/loam

With the current soil type (mostly loam), the volume of stormwater runoff in the

Oak Hill Court neighborhood before and after implementing RWH system was equal

152

(Figure SI – 4.5 a). This was because all the rainfall captured from the roof was infiltrated into the ground when the downspout discharged into the surrounding sub-catchment. This result showed that there was no stormwater reduction benefit from implementing RWH system if the infiltration capacity of the soil was very good.

Though there was no stormwater reduction in a neighborhood from implementing

RWH system, there was reduction in potable water demand equal to the volume of rainwater supply from the cistern. Environmental impacts from implementing RWH system in a suburban neighborhood were lower than municipal water supply system

(except for ozone depletion potential) indicating that RWH systems were attractive in most cases (Figure SI – 4.5). As expected, the decentralized scenario had higher ozone depletion potential. This was because the mass of the cistern was higher in the decentralized scenario even though the total capacity of cistern was approximately equal for centralized and decentralized. Transportation of larger masses caused higher ozone depletion potential.

The result of this study suggests that only the potable water saving benefit was sufficient for the RWH system to be environmentally attractive. This study reported that the environmental impacts for decentralized, partly centralized and centralized scenario were not significantly different except ozone depletion potential indicating that any scenario would be beneficial provided that ozone depletion was not a deciding factor.

(b) Clay type – clay

On the other hand when the downspout of the building discharged the runoff to the pervious areas with soil type clay, reduction in stormwater runoff was noticed from implementing RWH system. Life cycle stormwater runoff in the oak hill court

153 neighborhood before installing RWH system was found to be 2.2*106 m3 (Figure SI – 4.6 a). Use of RWH system to flush the toilets consumed the collected rainwater and thereby reduced the volume of stormwater runoff entering the combined sewer. For all the RWH scenarios, the volume of stormwater runoff reported was 1.9*106 m3 (Figure SI – 4.6 a) which was 8% less than the BAU case. Though the capacity of cistern was lower in the partly centralized scenario, the volume of stormwater runoff after implementing RWH system was found to be similar. This was because of the excess overflow from the cistern infiltrated into the ground. It was not only the soil type that affected infiltration, depression storages also had significant effects on stormwater runoff. Larger depression storages have the capability to store water for longer duration within which the collected rainwater either gets infiltrated or evaporated. This result suggested that stormwater benefit of RWH system entirely depends on the soil type. Soil with high infiltration capacity would have lower stormwater reduction while the soil with low infiltration capacities would have higher stormwater reduction from RWH system.

The use of RWH systems not only saved potable water and reduced stormwater runoff but also saved energy and emissions associated with the treatment of those. Like the sand soil type scenario, the environmental impacts from implementing RWH system with clay soil type in the neighborhood for the decentralized, partly centralized and centralized scenarios were approximately equal and lower than that of BAU scenario, except for ozone depletion potential. This indicates that RWH systems were attractive in most of the cases

(Figure SI – 4.6). Ozone depletion potential for the decentralized scenario was higher as compared to the partly centralized and centralized scenarios because of the transportation of higher mass of cistern in decentralized case. All RWH scenarios had higher ozone

154 depletion than BAU indicating that RWH system was not attractive if ozone depletion potential was considered as deciding factor.

Downspout connected to impervious areas or storm drain

When the downspout of the building was connected to the impervious areas, all the rainfall captured from the roof would be discharged to the storm drain and any consumptive use of the captured rainwater would directly reduce the stormwater runoff in the neighborhood. Life cycle stormwater runoff in the Oak Hill Court neighborhood before installing RWH system was found to be 1.92*106 m3 (Figure 4.4 a). The use of RWH system to flush the toilets consumed the collected rainwater and thereby reduced the volume of stormwater runoff. In the decentralized RWH scenario, the volume of surface runoff (stormwater runoff) reported was 1.39*106 m3 (Figure 4 a) which was 27% less than the BAU case. Similarly, in the case of partly centralized and centralized scenario, the volume of stormwater runoff was 1.78*106 m3 and 1.39*106 m3 (Figure 4.4 a), which was about 8% and 27% less than BAU case respectively. Decentralized and centralized scenario had highest reduction in stormwater runoff. This was because the total capacity of the rainwater cistern for the decentralized and centralized scenarios were similar while partly centralized had almost two third of the cistern capacity of decentralized or centralized. In another set of analysis, a hypothetical scenario was assumed when the total capacity of the cistern in the partly centralized scenario was kept equal to the decentralized and centralized scenarios. In that case, the volume of runoff noticed was equal to the decentralized and centralized scenarios. This study showed that the stormwater runoff from any neighborhood after implementing RWH systems depends on the capacity of the cistern.

155

Higher capacity would have higher reduction. It was worth mentioning that cistern size was reported to vary nonlinearly with the increase in roof area and or population. Therefore, it is important to accurately size the cistern before hydrologic analysis. However, one could also notice that linear increase in roof area and or population did not increase the cistern size linearly, but the stormwater runoff generated from the neighborhood was found to vary linearly with the cistern capacity. This study showed that the hydrologic results obtained using PCSWMM are highly susceptible to the cistern size selected in the study.

Considering maximum reduction in stormwater runoff, decentralized and centralized RWH scenarios were found to be attractive to implement in a suburban neighborhood.

The ratio of roof area of the buildings to the total area of the neighborhood was found to be 11% (Figure 4.2). No reduction in stormwater runoff was noticed for sand and loam soil type however, stormwater runoff reduction for clay type soil was found to be 8%.

Similarly, reduction in runoff for the downspout connected to impervious area was found to be 27% (Figure 4.4). A nonlinear relation was observed between the total roof area of the neighborhood and reduction in stormwater runoff. This was because, a large proportion of pervious areas (grass) were not observed to route the surface runoff to the storm drain.

In addition to the good soil type (sand, loam), several depression storages also played significant role in holding the runoff as well as infiltrating it.

Environmental impacts from implementing RWH systems in a suburban neighborhood were lower than the municipal water supply system (except decentralized scenario for ozone depletion potential) indicating that RWH systems were attractive in most of the cases (Figure 4.4). This was because of the transportation of RWH components from manufacturer to the installation point. Transportation is highly ozone depletion

156 intensive as compared to other RWH components. As the mass of cistern was increased in the decentralized scenario, the ozone depletion was higher among others. Unlike stormwater runoff, life cycle environmental impact did not always have the least environmental impact for the decentralized or centralized RWH scenarios (Figure 4.4).

Though differences were small between centralized and decentralized in relative comparison, the absolute values were significant.

The environmental feasibility of RWH scenarios in a suburban neighborhood varied with the impact categories. For example, centralized RWH system had minimum emissions in case of global warming, acidification and ozone depletion potential while decentralized RWH system had minimum emissions for energy use and eutrophication potential (Figure 4.4). Though the percentage reductions in stormwater runoff were similar, the environmental impacts were different among the centralized and decentralized scenarios. This was because the mass of the cistern (although the cisterns had the same capacity, the mass was higher in decentralized scenario), dual piping length (higher in centralized scenario), energy use by pump (higher in centralized scenario), and concrete pad (higher in decentralized scenario) were different among decentralized and centralized scenario. The decision to implement RWH system therefore is based on the impact categories of interest. This study suggested that emissions from centralized RWH scenario are always lower than BAU scenario, indicating that centralized scenario is always attractive. This finding supplements he findings of Vargas-Parra et al. 2013 where she stated that centralized cistern are environmentally attractive from water saving perspective.

Supplement in the sense that centralized scenarios are also environmentally attractive from both water savings and stormwater runoff reduction perspectives.

157

The percentage contribution to the life cycle environmental impacts also varied with the impact categories (Figure 4.4). Stormwater treatment was found to be major contributor to all the impact categories under consideration. For energy, global warming, acidification potential, wastewater treatment and potable water treatment; cistern and dual piping were found as other major contributors to life cycle emissions besides stormwater treatment. Eutrophication was solely due to the stormwater and wastewater treatment. For ozone depletion potential, transportation from plant to installation point was observed to be the second greatest contributor after stormwater treatment. The result of this study verified the findings of Devkota et al. 2015 (in prep) that the percentage contributions to life cycle emissions vary with the impact categories. It was also worth noting that eutrophication potential for all the downspout connectivity and all centralized and decentralized scenarios had negative emissions from the manufacturing of cistern and pump. This was because recycling of these materials was included in the system boundary.

158

2.5E+06

(a) 2.E+07 (b) ) 3 2.0E+06 1.5E+06 1.E+07

1.0E+06 (kWh) 5.E+06

5.0E+05 Life cycle Energycycle Life 0.0E+00 Life cycle runoff (m runoffcycle Life 0.E+00 Decentralized Partly Centralized Decentralized Partly Centralized Centralized Centralized BAU RWHS BAU RWHS

8.E+05 3.E+06 (c) (d) 2.E+06 6.E+05 2.E+06 4.E+05 1.E+06

5.E+05 2.E+05

Life cycle GHGcycle Life emission (Kg CO2 eq) CO2 (Kg emission 0.E+00 eq) H+ (Kg Potential 0.E+00 Decentralized Partly Centralized Acidification cycle Life Decentralized Partly Centralized Centralized Centralized BAU RWHS BAU RWHS

9.0E+04 (e) 1.6E-01 (f) 7.5E+04 1.2E-01 6.0E+04

4.5E+04 8.0E-02 3.0E+04

1.5E+04 4.0E-02eq) 11 CFC Life cycle Ozonecycle Life

0.0E+00 Potential (Kg N eq) N (Kg Potential Decentralized Partly Centralized (Kg Potential Depletion 0.0E+00

Life cycle Eutrophicationcycle Life -1.5E+04 Centralized Decentralized Partly Centralized Centralized BAU RWHS BAU RWHS

PVC Dual Piping 0.20000 Toilet flush tank with accessories Toilet replacement Wastewater treatment Storage tank Pumps Floating tank filter Overflow drain (PVC pipe) Vent for RWH tank (PVC) Sewer drain pipe (PVC)0.00000 Bends valves and tees (PVC) Toilets Transportation from plant to installation point Concrete Pad Potable water treatment Decentralized Partly Centralized Centralized Stormwater treatment Energy use by pump Pump replacement (every 20 years) Toilet flush tank with accessories replacementBAU Filter replacement (every 5 years) RWHS

Figure 4. 4 Life cycle (a) runoff (gallons), (b) Energy (kWh), (c) GWP (Kg CO2 eq), (d) Acidification (Kg H+ eq), (e) Eutrophication (Kg N eq) and (f) Ozone depletion (Kg CFC 11 eq) of RWH system at a suburban neighborhood when downspout is connected to storm- sewer.

159

4.3.3 Effect of pervious to impervious area ratio

50 Existing building Demand = supply case 40

stormwater runoff stormwater Percentage reduction in in reductionPercentage 10

0 5.8 5.2 4.7 4.3 3.9 3.6 3.3 3.0 2.8 2.6

Grass area to roof area ratio

Figure 4. 5 Percentage reduction in stormwater runoff due to change in roof area to grass area ratio.

The current oak hill neighborhood had a grass to roof area ratio of 5.8 which was high as compared to other urban neighborhoods. By knowing that the existing buildings in the neighborhood could not supply enough water to meet toilet flushing demand, the roof area was increased until the supply of rainwater meet the demand. The grass to roof area ratio at which the demand could be met by using harvested rainwater was found to be 2.6.

With the current soil type in the neighborhood, when the roof area increased or the ratio of grass to roof area decreased, the reduction in stormwater runoff by implementing RWH system also increased (Figure 4.5). A linear trend was observed with an R square value

0.99 (Figure SI-4.7). An R square value close to 1 represents the best fit. This was because the runoff was directly related to the impervious areas (buildings and parking lots), as grass

(pervious area) was not contributing to runoff at all. So any linear increase in roof area was responsible for the linear increase in stormwater runoff. The result of this analysis could 160 be useful to the building designers and practitioners to estimate stormwater runoff reductions for the respective grass to roof area ratio. In real practice, the roof area could not be increased beyond the design guidelines. However, increase in building roof area was found to be largely responsible for higher reduction in stormwater runoff. The increase in roof area has been limited to the point where supply equates the toilet flushing demand.

Any increase in building area after this point will only contribute to the stormwater runoff and is not considered in this study.

4.3.4 Selection framework for implementing RWH in a neighborhood

Based on the results observed in this study, a framework was developed to aid designers and policy makers to make a decision among different RWH scenarios (Figure 4.6).

Knowing that multifamily apartment complexes consist of around 27.8% of the total housing supply in urban settlements and 40.3% of the housing supply in suburban settlements, a designer or practitioner may be interested in designing RWH system for a neighborhood. Due to non-potable water use in residential buildings being the highest among commercial and institutional buildings. Since the downspout of the buildings entirely depend on the location, policy of the city and available space, a framework was developed in such a way that any practitioner could chose the downspout connectivity and the soil type to estimate the environmental impacts from implementing RWH system in a neighborhood. With the established downspout connectivity, soil type, impervious to pervious area, the designer would be able to estimate the reduction in stormwater runoff from the neighborhood. The decision to choose best scenario entirely depends on the

161 impact categories of interest. However, with the centralized RWH scenario always being less than BAU scenario, the centralized scenario was always attractive. It was worth noting that the decentralized scenario had less environmental impacts than the centralized for some impact categories, such as energy and eutrophication potential, but had higher ozone depletion potential than BAU case.

Figure 4. 6 Selection framework for environmental assessment of RWH system in a neighborhood.

162

4.3.5 Limitation

Capturing the rainwater onsite and using it to flush the toilets caused delay in the peak runoff that was supposed to be treated at the wastewater treatment plant. One of the limitations of LCA is that it does not account for the time delay in the treatment of stormwater runoff. Only the emissions related to the treatment of the captured rainwater can be captured in LCA. We recommend further study to capture the effect of time delay in the treatment of stormwater runoff.

4.4 Conclusions

This study is the first to assess both the hydrologic as well as the environmental benefits of implementing RWH system in a suburban neighborhood. This study presented surface runoff and environmental emissions from a wide range of RWH cistern location scenarios as well as the sensitivity of grass to roof area ratio. The results of this study are expected to help designers and policy makers choose environmentally attractive RWH scenario by looking into effect of downspout connectivity and soil type. Several conclusions were drawn from this study:

1. The cistern size increased with the increase in roof area but the relationship was

not linear. This study suggested that YAS approach was not linearly scalable with

demand (occupancy) and supply (roof area). Partly centralized RWH scenario had

the smallest cistern as compared to centralized and decentralized RWH scenario.

163

2. Stormwater reductions from RWH system depend mostly on the building

downspout connectivity. No reduction in stormwater was noticed when the

downspout was connected to pervious areas with soil type such as sand, loam and

silt. In such cases the savings from implementing RWH systems was only potable

water savings. Significant stormwater reduction (8%) was observed for the clay

and was similar across the centralized, decentralized and partly centralized

scenarios. The reductions were different among RWH scenarios when downspouts

directly discharged to impervious areas or storm drain. 27% reduction was noticed

for the centralized and decentralized, while 8% was observed for the partly

centralized scenario.

3. The stormwater reduction from implementing RWH system was directly

proportional to the cistern capacity when the downspout was connected to storm

drain. Bigger cisterns had higher stormwater reduction. However, reduction in

stormwater runoff did not depend on the cistern capacity when the downspout was

connected to pervious areas.

4. Regardless of the downspout connectivity, the environmental emissions from the

decentralized, partly centralized and centralized scenarios were lower than BAU

scenario, except ozone depletion potential, indicating that RWH systems were

attractive in most of the cases. When the downspout was connected to impervious

areas, few environmental impact categories (global warming, acidification and

ozone depletion potential) reported that centralized RWH scenario was attractive,

while others (energy use and eutrophication potential) reported that the

decentralized scenario was attractive. It is therefore important to pick the impact

164

category of interest to decide the environmental feasibility of RWH system. Also,

the decentralized scenario was not attractive according to ozone depletion

potential due to larger transportation distance and heavy weight of cistern material.

Reduction in transportation distance could not only reduce ozone depletion

potential but also make it attractive.

5. RWH systems could supply enough water to meet the toilet flushing demand when

the grass to roof area ratio was 2.6. At this ratio, the roof area was twice the

existing roof area which might not be practical according to building design

guidelines, but the result from this study would be beneficial for building

practitioners and designers to consider pervious area to impervious area ratio to

reduce stormwater runoff from a neighborhood.

Acknowledgements

This study was funded by National Science Foundation’s Environmental Sustainability

(grant # 1236660).

165

4.5 References

Anand, C., Apul, D.S., 2010., Economic and environmental analysis of standard, high efficiency, rainwater flushed, and composting toilets. Journal of Environment Management 92, 419-428.

American Planning Association. (APA, 2015). Available online from: https://www.planning.org/pas/at60/report139.htm. Last accessed: September 2015.

Barco, J., Wong, K. M., & Stenstrom, M. K. (2008). Automatic calibration of the US EPA SWMM model for a large urban catchment. Journal of Hydraulic Engineering.

Bruins, H. J., Evenari, M., & Nessler, U. (1986). Rainwater-harvesting agriculture for food production in arid zones: the challenge of the African famine.Applied Geography, 6(1), 13-32.

Burns, M. J., Fletcher, T. D., Walsh, C. J., Ladson, A. R., & Hatt, B. E. (2012). Hydrologic shortcomings of conventional urban stormwater management and opportunities for reform. Landscape and Urban Planning, 105(3), 230-240.

Cengel Y, Cimbala J, (2005). Fluid Mechanics: Fundamentals and Applications. McGraw Hill Higher Education, Boston.

Cheng C L, (2002). Study of the inter-relationship between water use and energy conservation for a building. Energy Build 34 (3), 261e266.

Choguill CL (1996) Ten steps to sustainable infrastructure. Habitat Int 20(3):389–404

Chow, V. T., Maidment, D. R., & Mays, L. W. (1988). Applied hydrology.

Deru, M., Field, K., Studer, D., Benne, K., Griffith, B., Torcellini, P., ... and Crawley, D. (2011). US Department of Energy commercial reference building models of the national building stock.

166

Devkota, J., Apul, D., Burian, J. and Davani H. (accepted) Introducing Demand to Supply Ratio as a New Metric for Understanding Life cycle Greenhouse Gas (GHG) Emissions from Rainwater Harvesting Systems. Journal of Cleaner Production. Devkota, J., Schlachter, H., Anand, C., Phillips, R., and Apul, D., 2013. Development and application of EEAST: A life cycle based model for use of harvested rainwater and composting toilets in buildings. Journal of Environment Management 130, 397-404.

Dietz, M.E. (2007). “Low impact development practices: A review of current research and recommendations for future directions.” Water Air Soil Pollution, 186 (1-3), 351- 363.

Ecoinvent, 2010. Ecoinvent v2.2 database. Swiss Centre for Life Cycle Inventories, Dübendorf, Switzerland.

Executive Office of Housing and Economic Development. Mass.Gov. Compact Neighborhoods. Available online from: http://www.mass.gov/hed/community/planning/compact-neighborhoods.html. Last Accessed Sep 2015.

Farreny, R., Gabarrell, X., & Rieradevall, J. (2011). Cost-efficiency of rainwater harvesting strategies in dense Mediterranean neighbourhoods. Resources, conservation and recycling, 55(7), 686-694.

Fewkes A and Butler D, (2000). Simulating the performance of rainwater collection and reuse systems using behavioural models. Building Services Engineering Research and Technology 21, no. 2: 99-106.

Fewkes, A., 1998. The use of rainwater for WC flushing: the field testing of a collection system. Building and Environment 34(6), 765-772.

Furumai, H., 2008. Rainwater and reclaimed wastewater for sustainable urban water use. Phys. Chem. Earth - Parts A/B/C; 33(5), 340-346.

Gleick, P.H., 1996. Basic water requirements for human activities: meeting basic needs. Water Int. 21 (2), 83.

167

Hanson R (ed) (1984) Perspectives on urban infrastructure. Committee on National Urban Policy, National Research Council, Washington, DC. National Academies Press, 216 pp

Helmreich, B., & Horn, H. (2009). Opportunities in rainwater harvesting. Desalination, 248(1), 118-124.

IBC, I. (2006). International building code. International Code Council, Inc.(formerly BOCA, ICBO and SBCCI), 4051, 60478-5795.

Kirk S J, Dell’Isola A J, (1995). Life Cycle Costing for Design Professionals. McGraw- Hill, New York.

Mayer, P.W., William, B.D., 1999. Residential End Uses of Water. American Water Works Association Research Foundation (AWWA), Denver.

Mitchell V G, (2007). How important is the selection of computational analysis method to the accuracy of rainwater tank behaviour modelling?. Hydrological processes, 21(21), 2850-2861.

Pachpute, J. S., Tumbo, S. D., Sally, H., & Mul, M. L. (2009). Sustainability of rainwater harvesting systems in rural catchment of Sub-Saharan Africa. Water Resources Management, 23(13), 2815-2839.

PE International, 2014. Gabi database. http://www.pe-international.com.

Petrucci, G., Deroubaix, J. F., De Gouvello, B., Deutsch, J. C., Bompard, P., & Tassin, B. (2012). Rainwater harvesting to control stormwater runoff in suburban areas. An experimental case-study. Urban Water Journal, 9(1), 45-55.

168

Reidy, P. C. (2008). Integrating rainwater harvesting and stormwater management infrastructure: Double benefit–single cost. In 2008 International Low Impact Development Conference.

Steffen, J., Jensen, M., Pomeroy, C. A., & Burian, S. J., 2013. Water Supply and Stormwater Management Benefits of Residential Rainwater Harvesting in US Cities. JAWRA Journal of the American Water Resources Association, 49(4), 810-824. Stout, D. T., Walsh, T. C., & Burian, S. J. (2015). Ecosystem services from rainwater harvesting in India. Urban Water Journal, (ahead-of-print), 1-13.

U.S. EPA. (2014a). “Greening CSO plans: Planning and modeling green infrastructure for combined sewer overflow (CSO) control.” EPA-832-R-14-001, Washington, DC.

U.S. EPA. (2008). “Combined sewer overflows: Demographics.” 〈http://cfpub.epa.gov/npdes/cso/demo.cfm〉 (Jul. 5, 2014).

Vargas-Parra, M. V., Villalba, G., & Gabarrell, X. (2013). Applying exergy analysis to rainwater harvesting systems to assess resource efficiency.Resources, Conservation and Recycling, 72, 50-59.

Vickers A, (2001). A Handbook of water use and conservation. Water Flow Press. Amherst: MA.

Villarreal, E. L., and Dixon, A., 2005. Analysis of a rainwater collection system for domestic water supply in Ringdansen, Norrköping, Sweden. Building and Environment, 40(9), 1174-1184.

Web Soil Survey. United States Department of Agriculture. Natural Resource Conservation Service. Available online from: http://websoilsurvey.sc.egov.usda.gov/App/HomePage.htm

Yuan, T., Fengmin, L., & Puhai, L. (2003). Economic analysis of rainwater harvesting and irrigation methods, with an example from China. Agricultural water management, 60(3), 217-226.

169

Zhu, K., Zhang, L., Hart, W., Liu, M., & Chen, H. (2004). Quality issues in harvested rainwater in arid and semi-arid Loess Plateau of northern China.Journal of Arid Environments, 57(4), 487-505.

170

4.6 Supporting information: Environmental impacts from Implementing

Rainwater Harvesting System at a suburban neighborhood: A comparison of

centralized versus decentralized tank location and pervious to impervious area

ratio.

Jay P. Devkota1, Anthony Dietrich1, Rahul Yarlagadda2, Cyndee Gruden1, Steven J.

Burian3, Defne S. Apul1*

1 2801 W. Bancroft, Department of Civil Engineering, University of Toledo, Toledo, OH,

43606

2 2801 W. Bancroft, Department of Mechanical Engineering, University of Toledo,

Toledo, OH, 43606

3110 S. Central Campus Dr, Department of Civil & Environmental Engineering,

University of Utah, Salt Lake City, UT, 84112

*Corresponding author tel: (419)-530-8132; email: [email protected]

171

Figure SI - 4. 1 Soil characteristics for the suburban neighborhood (collected from web soil survey, Natural Resource Conservation Service, United States Department of Agriculture).

120 (1) Occupancy is constant (72 no. of occupants) 30000

100 25000

80 20000

60 15000

40 10000 Tank(gallons) size

Volumetric Volumetric reliability(%) 20 Volumetric reliability (%) Tank size (gallons) 5000

0 0 0 5000 10000 15000 20000 25000 Roof area (sq ft)

Figure SI - 4. 2 Variation of cistern size with the increase in roof area keeping occupancy constant.

172

120 (2) Roof area is constant (8000 sq ft roof area) 25000

100 20000 80 15000 60 10000

Tank(gallons) size Volumetric Volumetric reliability(%) 20 5000

0 0 0 20 40 60 80 100

Occupancy Volumetric reliability (%) Tank size (gallons)

Figure SI - 4. 3 Variation of cistern size with the increase in occupancy keeping roof area constant.

100 (3) Both roof area and occupancy are increasing linearly 40000 provided that roof area to occupancy ratio is constant throughout 75 30000

50 20000

25 10000 Tank(gallons) size Volumetric Volumetric reliability(%)

0 0 0 2 4 6 8 10 12 14

Occupancy Volumetric reliability (%) Tank size (gallons)

Figure SI - 4. 4 Variation of cistern size with the increase in roof area and occupancy keeping roof area to occupancy ratio constant.

173

2.E+06 (a) 1.E+07 (b) 1.E+07 1.E+06 8.E+06 6.E+06 5.E+05 4.E+06 2.E+06

Life cycle runoff (m3) runoffcycle Life 0.E+00 0.E+00 Decentralized Partly Centralized (kWh) Energycycle Life Decentralized Partly Centralized Centralized Centralized BAU RWHS BAU RWHS 2.E+06 (c) 6.E+05 (d) 2.E+06 5.E+05 4.E+05 1.E+06 3.E+05 8.E+05 2.E+05 4.E+05

1.E+05 Life cycle GHGcycle Life 0.E+00

emission (Kg CO2 eq) CO2 (Kg emission 0.E+00 DecentralizedPartly CentralizedCentralized eq) H+ (Kg Potential

Life cycle Acidification cycle Life Decentralized Partly Centralized BAU RWHS Centralized BAU RWHS 0.15 8.E+4 (e) (f) 6.E+4 0.12

5.E+4 0.09

3.E+4 0.06 CFC 11 eq) 11 CFC 2.E+4

Life cycle Ozonecycle Life 0.03 Depletion Potential (Kg Potential Depletion

Potential (Kg N eq) N (Kg Potential 0.E+0 0.00 Decentralized Partly Centralized Decentralized Partly Centralized Life cycle Eutrophicationcycle Life Centralized -2.E+4 Centralized BAU RWHS BAU RWHS PVC Dual0.20000 Piping Toilet flush tank with accessories Toilet replacement Wastewater treatment Storage tank Pumps Floating tank filter Overflow drain (PVC pipe) Vent for RWH tank (PVC) Sewer drain pipe (PVC) Bends0.00000 valves and tees (PVC) Toilets Transportation from plant to installation point Concrete Pad Potable water treatment Decentralized StormwaterPartly Centralized treatment Centralized Energy use by pump Pump replacement (every 20 years) Toilet flush tank with accessoriesBAU replacement Filter replacementRWHS (every 5 years)

Figure SI - 4. 5 Life cycle (a) runoff (gallons), (b) Energy (kWh), (c) GWP (Kg CO2 eq), (d) Acidification (Kg H+ eq), (e) Eutrophication (Kg N eq) and (f) Ozone depletion (Kg CFC 11 eq) of RWH system at a suburban neighborhood when the downspout is connected to grass and the soil type is sand or loam.

174

2,500,000 16000000 ) (b) 3 2.2E+6 (a) 1.9E+6 1.9E+6 1.9E+6 2,000,000 12000000 1,500,000 8000000 1,000,000 500,000 4000000

Life cycle runoff (m runoffcycle Life 0 0 Decentralized Partly Centralized Decentralized Partly Centralized

Centralized (kWh) Energycycle Life Centralized BAU RWHS BAU RWHS

3000000 (c) 800000 (d) 2500000 700000 600000 2000000 500000 1500000 400000 1000000 300000 (Kg CO2 eq) CO2 (Kg 200000 500000 eq) H+ (Kg Potential Life cycle Acidification cycle Life 100000 Life cycle GHG emission emission GHGcycle Life 0 Decentralized Partly Centralized 0 Centralized Decentralized Partly Centralized Centralized BAU RWHS BAU RWHS 100000 (e) 0.20 (f) 80000 0.15 60000

40000 0.10

20000 0.05

Potential (Kg N eq) N (Kg Potential 0

Life cycle Eutrophicationcycle Life Decentralized Partly Centralized

Potential (Kg CFC 11 eq) 11 CFC (Kg Potential 0.00

Centralized Depletion Ozonecycle Life -20000 Decentralized Partly Centralized BAU RWHS Centralized BAU RWHS PVC Dual0.20000 Piping Toilet flush tank with accessories Toilet replacement Wastewater treatment Storage tank Pumps Floating tank filter Overflow drain (PVC pipe) Vent for RWH tank (PVC) Sewer drain pipe (PVC) Bends0.00000 valves and tees (PVC) Toilets Transportation from plant to installation point Concrete Pad Potable water treatment Decentralized StormwaterPartly Centralized treatment Centralized Energy use by pump BAU Pump replacementRWHS (every 20 years) Toilet flush tank with accessories replacement Filter replacement (every 5 years)

Figure SI - 4. 6 Life cycle (a) runoff (gallons), (b) Energy (kWh), (c) GWP (Kg CO2 eq), (d) Acidification (Kg H+ eq), (e) Eutrophication (Kg N eq) and (f) Ozone depletion (Kg CFC 11 eq) of RWH system at a suburban neighborhood when the downspout is connected to grass and the soil type is clay.

175

runoff 20 y = -2.8625x + 43.88 R² = 0.9939 10

Percentage change in stormwater stormwater in change Percentage 0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Grass to roof area ratio

Figure SI - 4. 7 Percentage change in stormwater runoff due to change in roof area to grass area ratio.

Table SI - 4. 1 Soil characteristics for the suburban neighborhood presented in Figure SI-1 (collected from web soil survey, Natural Resource Conservation Service, United States Department of Agriculture).

Map Unit Map Unit Name Acres in Percent of Symbol AOI AOI Co Colwood loam 4.1 18.40% Cp Colwood-Urban land complex 11.3 51.00%

DgA Digby sandy loam, 0 to 2 0.4 1.90% percent slopes

DsA Dixboro fine sandy loam, 0 to 2 5.2 23.60% percent slopes

SmC Sisson loam, 6 to 12 percent 1.1 5.10% slopes Totals for Area of Interest 22.1 100.00%

176

177

Table SI - 4. 2 Inventory of RWH system components for building type A.

Phase Scenario Item Quantity Units PVC Pipe (total) 79.48 kg

Sewer drain pipe (PVC) 7.05 kg

Bends valves and tees (PVC) 0.65 kg Transportation from plant to installation point 6306.98 kg-km

Construction Toilets 297.56 kg Toilet flush tank with accessories 134.26 kg

Toilet replacement 595.11 kg

Business as Usual Business as Potable water treatment (per day) 2,224 kg Wastewater treatment (per day) 2.2 m3

Operation Stormwater treatment (per day) 5.0 m3 Storage tank 3150.43 kg Pumps 10.70 kg Floating tank filter 3.63 kg PVC Pipes (dual piping) 170.05 kg

Overflow drain (PVC pipe) 0.96 kg Vent for RWH tank (PVC) 0.96 kg Sewer drain pipe (PVC) 29.39 kg

Construction Bends valves and tees (PVC) 0.65 kg Toilets 297.56 kg Toilet flush tank with accessories 134.26 kg Transportation from plant to installation point 327814.90 kg-km Concrete Pad 1.28 m3 Potable water treatment (per day) 867 kg Wastewater treatment (per day) 2.2 m3

Stormwater treatment (per day) 3.6 m3

Rainwater Harvesting / Toilet Flushing /Rainwater Toilet Harvesting Energy use by pump (per day) 0.01236 MJ Pump replacement (every 20 years) 32.10 kg Operation Toilet replacement 595.11 kg Toilet flush tank with accessories replacement 268.53 Kg Filter replacement (every 5 years) 18.14 kg

178

Table SI - 4. 3 Inventory of RWH system components for building type B.

Phase Scenario Item Quantity Units PVC Pipe (total) 53.10 kg

Sewer drain pipe (PVC) 3.53 kg Bends valves and tees (PVC) 0.65 kg Transportation from plant to installation point 5954.26 kg-km

Construction Toilets 148.78 kg Toilet flush tank with accessories 67.13 kg Toilet replacement 297.56 kg

Business as Usual Business as Potable water treatment (per day) 1,112 kg

Operation Wastewater treatment (per day) 1.1 m3 Stormwater treatment (per day) 5.0 m3 Storage tank 2632.58 kg Pumps 10.70 kg Floating tank filter 3.63 kg PVC Pipes (dual piping) 113.62 kg

Overflow drain (PVC pipe) 0.96 kg Vent for RWH tank (PVC) 0.96 kg Sewer drain pipe (PVC) 29.39 kg

Construction Bends valves and tees (PVC) 0.65 kg Toilets 148.78 kg Toilet flush tank with accessories 67.13 kg Transportation from plant to installation point 275085.63 kg-km Concrete Pad 0.99 m3 Potable water treatment (per day) 411 kg Wastewater treatment (per day) 1.1 m3

Stormwater treatment (per day) 3.6 m3

Rainwater Harvesting / Toilet Flushing /Rainwater Toilet Harvesting Energy use by pump (per day) 0.00638 MJ Pump replacement (every 20 years) 32.10 kg Operation Toilet replacement 297.56 kg Toilet flush tank with accessories replacement 134.26 Kg Filter replacement (every 5 years) 18.14 kg

179

Table SI - 4. 4 Inventory of RWH system components for cistern location 1.

Phase Scenario Item Quantity Units PVC Pipe (total) 274.75 kg

Sewer drain pipe (PVC) 49.38 kg Bends valves and tees (PVC) 0.65 kg Transportation from plant to installation point 10539.54 kg-km

Construction Toilets 2082.90 kg Toilet flush tank with accessories 939.84 kg Toilet replacement 4165.80 kg

Business as Usual Business as Potable water treatment (per day) 15,568 kg

Operation Wastewater treatment (per day) 15.6 m3 Stormwater treatment (per day) 23.4 m3 Storage tank 7951.74 kg Pumps 10.70 kg Floating tank filter 3.63 kg PVC Pipes (dual piping) 6338.12 kg

Overflow drain (PVC pipe) 0.96 kg Vent for RWH tank (PVC) 0.96 kg Sewer drain pipe (PVC) 235.14 kg

Construction Bends valves and tees (PVC) 0.65 kg Toilets 2082.90 kg Toilet flush tank with accessories 939.84 kg Transportation from plant to installation point 960144.69 kg-km Concrete Pad 4.70 m3 Potable water treatment (per day) 6,694 kg Wastewater treatment (per day) 15.6 m3

Stormwater treatment (per day) 21.7 m3

Rainwater Harvesting / Toilet Flushing /Rainwater Toilet Harvesting Energy use by pump (per day) 0.08084 MJ Pump replacement (every 20 years) 32.10 kg Operation Toilet replacement 4165.80 kg Toilet flush tank with accessories replacement 1879.69 Kg Filter replacement (every 5 years) 18.14 kg

180

Table SI - 4. 5 Inventory of RWH system components for cistern location 2.

Phase Scenario Item Quantity Units PVC Pipe (total) 153.06 kg

Sewer drain pipe (PVC) 24.69 kg Bends valves and tees (PVC) 0.65 kg Transportation from plant to installation point 8070.55 kg-km

Construction Toilets 1041.45 kg Toilet flush tank with accessories 469.92 kg Toilet replacement 2082.90 kg

Business as Usual Business as Potable water treatment (per day) 7,784 kg

Operation Wastewater treatment (per day) 7.8 m3 Stormwater treatment (per day) 23.4 m3 Storage tank 4938.91 kg Pumps 10.70 kg Floating tank filter 3.63 kg PVC Pipes (dual piping) 1859.43 kg

Overflow drain (PVC pipe) 0.96 kg Vent for RWH tank (PVC) 0.96 kg Sewer drain pipe (PVC) 117.57 kg

Construction Bends valves and tees (PVC) 0.65 kg

Toilet Flushing Toilet Toilets 1041.45 kg

Toilet flush tank with accessories 469.92 kg Transportation from plant to installation point 552156.11 kg-km Concrete Pad 2.44 m3 Potable water treatment (per day) 3,347 kg Wastewater treatment (per day) 7.8 m3

Stormwater treatment (per day) 21.7 m3

Rainwater Harvesting / Rainwater Harvesting Energy use by pump (per day) 0.04042 MJ Pump replacement (every 20 years) 32.10 kg Operation Toilet replacement 2082.90 kg Toilet flush tank with accessories replacement 939.84 Kg Filter replacement (every 5 years) 18.14 kg

181

Table SI - 4. 6 Inventory of RWH system components for cistern location 3.

Phase Scenario Item Quantity Units PVC Pipe (total) 60.21 kg

Sewer drain pipe (PVC) 7.05 kg Bends valves and tees (PVC) 0.65 kg Transportation from plant to installation point 6306.98 kg-km

Construction Toilets 297.56 kg Toilet flush tank with accessories 134.26 kg Toilet replacement 595.11 kg

Business as Usual Business as Potable water treatment (per day) 2,224 kg

Operation Wastewater treatment (per day) 2.2 m3 Stormwater treatment (per day) 23.4 m3 Storage tank 3150.43 kg Pumps 10.70 kg Floating tank filter 3.63 kg PVC Pipes (dual piping) 514.78 kg

Overflow drain (PVC pipe) 0.96 kg Vent for RWH tank (PVC) 0.96 kg Sewer drain pipe (PVC) 58.79 kg

Construction Bends valves and tees (PVC) 0.65 kg Toilets 297.56 kg Toilet flush tank with accessories 134.26 kg Transportation from plant to installation point 338922.57 kg-km Concrete Pad 1.28 m3 Potable water treatment (per day) 867 kg Wastewater treatment (per day) 2.2 m3

Stormwater treatment (per day) 21.7 m3

182

Table SI - 4. 7 Inventory of RWH system components for centralized cistern location.

Phase Scenario Item Quantity Units PVC Pipe (total) 842.91 kg

Sewer drain pipe (PVC) 81.12 kg Bends valves and tees (PVC) 0.65 kg Transportation from plant to installation point 13713.97 kg-km

Construction Toilets 3421.90 kg Toilet flush tank with accessories 1544.03 kg Toilet replacement 6843.81 kg

Business as Usual Business as Potable water treatment (per day) 25,576 kg

Operation Wastewater treatment (per day) 25.6 m3 Stormwater treatment (per day) 70.2 m3 Storage tank 17959.31 kg Pumps 10.70 kg Floating tank filter 3.63 kg PVC Pipes (dual piping) 17362.82 kg

Overflow drain (PVC pipe) 0.96 kg Vent for RWH tank (PVC) 0.96 kg Sewer drain pipe (PVC) 411.50 kg

Construction Bends valves and tees (PVC) 0.65 kg Toilets 3421.90 kg Toilet flush tank with accessories 1544.03 kg Transportation from plant to installation point 2212259.99 kg-km Concrete Pad 13.56 m3 Potable water treatment (per day) 9,975 kg Wastewater treatment (per day) 25.6 m3

Stormwater treatment (per day) 50.9 m3

Rainwater Harvesting / Toilet Flushing /Rainwater Toilet Harvesting Energy use by pump (per day) 0.14213 MJ Pump replacement (every 20 years) 32.10 kg Operation Toilet replacement 6843.81 kg Toilet flush tank with accessories replacement 3088.06 Kg Filter replacement (every 5 years) 18.14 kg

183

Chapter 5

Conclusions

This study presented a wide range of parameters affecting environmental sustainability of rainwater harvesting systems. Two parameters, sewer type and demand to supply ratio were introduced for the first time to estimate environmental impacts from rainwater harvesting systems and are discussed in Chapter 2. The result showed that the life cycle GHG emissions of using rainwater harvesting system in an office building depends whether the building is connected to a combined sewer or separate sewer. D/S ratio which is a measure of building occupancy to roof was introduced to estimate life cycle greenhouse gas emissions from any commercial building. Based on the results, a decision tree was proposed to estimate life cycle greenhouse gas emissions based on the building roof area and occupancy.

Chapter 3 presented a unique concept when the rainwater harvesting system was analyzed using two different approach; supply and demand based. The effect of the supply and demand based analysis on the environmental impacts of implementing RWH system was presented. Environmental impacts were assessed and compared for two functional units: supply of one cubic meter of rainwater supplied versus provision of sanitation service in the building. Nine climatic regions of United States as well as one climatic region from

184

United Kingdom are selected to best describe the effect of regional precipitation pattern in the environmental sustainability of rainwater harvesting system. We expected the regions with lower net impacts for one functional unit would also have lower impacts for the other functional unit. However, such a trend was not observed making it clear that the regional preference for RWH depends on the specific goal to be met: using harvested rainwater as a potential supplemental source versus using it to meet the water demand needs.

Chapter 4 is the first to assess both the hydrologic as well as the environmental benefits of implementing RWH system in a suburban neighborhood. This study presented surface runoff and environmental emissions from a wide range of RWH cistern implementation strategies as well as the sensitivity of grass to roof area ratio. The result showed that the environmental impacts from rainwater harvesting system depends mainly on the downspout connection, soil type and cistern implementation strategy. A selection framework was therefore proposed to select the system with lower environmental impact based on downspout connectivity, soil type and the centralized partly centralized or decentralized strategy.

Recommendations for Future Work

This study demonstrated the effect of D/S ratio in the life cycle GHG emissions of implementing RWH systems. The result of this research and the trends observed are only applicable to GHG emissions. While GHG emissions played a major role in decision making in commercial sectors, other environmental impact categories such as eutrophication potential, ecotoxicity, acidification potential and ozone depletion also would be critical. It is further recommended to investigate whether similar trends would be

185 achieved for other impact categories such as eutrophication potential, acidification potential, ecotoxicity and others. Capturing the rainwater onsite and using it to flush the toilets caused delay in the peak runoff that was supposed to be treated at the wastewater treatment plant. One of the limitations of LCA is that it does not account for the time delay in the treatment of stormwater runoff. Only the emissions related to the treatment of the captured rainwater can be captured in LCA. We recommend further study to capture the effect of time delay in the treatment of stormwater runoff. In this study, we assumed all excess roof runoff to be treated at a wastewater treatment plant in combined sewer communities. However, in reality, only some of this storm water and some of the toilet flushed wastewater are treated at the wastewater treatment plant during combined sewer overflows. The percentages that would be treated depend on many factors such as sewer network and capacity, rain intensity, and wastewater treatment plant capacity. Future research is recommended to account these factors to accurately estimate the environmental impacts associated with the combined sewer overflow. There are areas of study of rainwater harvesting systems that have been neglected but are critical for evaluating sustainability. One notable area is the social impacts and benefits. Rainwater harvesting is dependent on the human intervention and oversight and few studies comment on the amount of maintenance required by a building owner, the amount of training, different models for distributed infrastructure management and maintenance, and other important issues. Further study is recommended to investigate social implications of implementing harvested rainwater.

186

References

Aladenola, O. O., & Adeboye, O. B. (2010). Assessing the potential for rainwater harvesting. Water Resources Management, 24(10), 2129-2137.

Alam, R., Munna, G., Chowdhury, M.A.I, Sarkar, M.S.K.A., Ahmed, M., Rahman, M.T., Jesmin, F., and Taimoor M.A. (2012),"Feasibility study of rainwater harvesting system in Sylhet City, Environmental Monitoring and Assessment", 184(1),573– 580.

Alliance for Water Efficiency, 2015. Available online from: http://www.allianceforwaterefficiency.org/1Column.aspx?id=750&LangType=1033&ter ms=water+use+in+office+building. Accessed May 2015.

American Planning Association. (APA, 2015). Available online from: https://www.planning.org/pas/at60/report139.htm. Last accessed: September 2015.

Anand C and Apul D S, (2010) Cost, Energy, and CO2 Emissions Analysis of Standard, High Efficiency, Rainwater Flushed, and Composting Toilets. Journal of Environmental Management, 92(3), 419-428.

Barco, J., Wong, K. M., & Stenstrom, M. K. (2008). Automatic calibration of the US EPA SWMM model for a large urban catchment. Journal of Hydraulic Engineering. Bare, J. (2011). TRACI 2.0: the tool for the reduction and assessment of chemical and other environmental impacts 2.0. Clean Technologies and Environmental Policy, 13(5), 687-696.

Basset-Mens, C., & Van der Werf, H. M. (2005). Scenario-based environmental assessment of farming systems: the case of pig production in France.Agriculture, Ecosystems & Environment, 105(1), 127-144.

187

Blengini, G. A., & Di Carlo, T., 2010. The changing role of life cycle phases, subsystems and materials in the LCA of low energy buildings. Energy and Buildings, 42(6), 869-880.

Bruins, H. J., Evenari, M., & Nessler, U. (1986). Rainwater-harvesting agriculture for food production in arid zones: the challenge of the African famine.Applied Geography, 6(1), 13-32.

Burian, S.J. and Jones, D. (2010), “National assessment of rainwater harvesting as a stormwater best management practice: Challenges, needs, and recommendations.” Low Impact Development 2010: Redefining Water in the City, Proceedings of the 2010 International Low Impact Development Conference, ASCE, Reston, VA.

California Energy Commission (2005), “California’s Water-Energy Relationship.” Integrated energy policy report proceeding, (04-IEPR-01E), Sacramento, Ca, http://www.energy.ca.gov/2005publications/CEC-700-2005-011/CEC-700-2005- 011-SF.PDF.

Cengel Y, Cimbala J, (2005). Fluid Mechanics: Fundamentals and Applications. McGraw Hill Higher Education, Boston.

Cheng C L, (2002). Study of the inter-relationship between water use and energy conservation for a building. Energy Build 34 (3), 261e266.

Chiu, Y. R., Liaw, C. H., & Chen, L. C., 2009. Optimizing rainwater harvesting systems as an innovative approach to saving energy in hilly communities. Renewable energy, 34(3), 492-498.

Choguill CL (1996) Ten steps to sustainable infrastructure. Habitat Int 20(3):389–404 Chow, V. T., Maidment, D. R., & Mays, L. W. (1988). Applied hydrology.

Commercial Building Energy Consumption Survey, 2012. Available online from: http://www.eia.gov/consumption/commercial/data/2012/. Accessed May 2015.

188

CostLab, 2015. Available online at: http://whitestoneresearch.com/Costlab.

Crettaz P, Jolliet O, Cuanillon J M, Orlando S, (1999). Life cycle assessment of drinking water and rain water for toilets flushing. J. Water Serv. Res. Tec. 48, 73-83.

Crites R, and Technobanoglous G, (1998). Small and decentralized wastewater management systems. McGraw-Hill.

Crowley, B.J. (2005), "A neighborhood level analysis of rainwater catchment in Portland, OR”. Research paper in partial fulfillment of Master of Science Degree in Geography, Portland State University.

Deru M, Field K, Studer D, Benne K, Griffith B, Torcellini P, and Crawley D, (2011). US Department of Energy commercial reference building models of the national building stock.

Devkota J, Hannah S, and Apul D S, (2015) Life Cycle Based Evaluation of Harvested Rainwater Use in Toilets and for Irrigation. Journal of Clean Production. 95 (2015): 311-321.

Devkota, J., Apul, D., Burian, J. and Davani H. (2015) Introducing Demand to Supply Ratio as a New Metric for Understanding Life cycle Greenhouse Gas (GHG) Emissions from Rainwater Harvesting Systems. Journal of Cleaner Production.

Devkota, J., Apul, D., Burian, J. and Davani H. (accepted with revisions) Review of Rainwater Harvesting Benefits for Urban Water Infrastructure Sustainability. Critical Reviews in Environmental Science and Technology.

Dietz, M.E. (2007). “Low impact development practices: A review of current research and recommendations for future directions.” Water Air Soil Pollution, 186 (1-3), 351- 363.

Ecoinvent, 2010. Ecoinvent v2.2 database. Swiss Centre for Life Cycle Inventories, Dübendorf, Switzerland.

189

Fewkes A and Butler D, (2000). Simulating the performance of rainwater collection and reuse systems using behavioural models. Building Services Engineering Research and Technology 21, no. 2: 99-106.

Fewkes A, (1998). The use of rainwater for WC flushing: the field testing of a collection system. Building and Environment 34(6), 765-772.

Fewkes A, and Wam P, (2000). Method of modelling the performance of rainwater collection systems in the United Kingdom. Building Services Engineering Research and Technology, 21(4), 257-265. Furumai, H. (2008),"Rainwater and reclaimed wastewater for sustainable urban water use." Physics and Chemistry of the Earth, 33(5), 340–346

Georgia Rainwater Harvesting Guidelines, 2009. Available online from: http://www.conservewatergeorgia.net/resources/Georgia_Rainwater_Harvesting_ Guidelines.pdf Accessed Dec 2014.

GHG protocol. Available online from: http://www.ghgprotocol.org/files/ghgp/public/FAQ.pdf Accessed May 14 2014.

Ghisi, E., Rupp, R. F., & Triska, Y. (2014). Comparing indicators to rank strategies to save potable water in buildings. Resources, Conservation and Recycling, 87, 137- 144.

Gleick P H, (1996). Basic water requirements for human activities: meeting basic needs. Water Int. 21 (2), 83.

Gleick, P.H. (2010),"Roadmap for sustainable water resources in southwestern North America." Proceedings of the National Academy of Sciences of the United States of America (PNAS), 107(50), 21300-21305. 190

Glendenning, C.J., and Vervoort, R.W. (2010),"Hydrological impacts of rainwater harvesting (rainwater harvesting) in a case study catchment: The Arvari River, Rajasthan, India. Part 1: Field-scale impacts, Agricultural Water Management." 98(2), 331-340.

Gomez-Ullate, E., Novo, A.V., Bayon, J.V., Hernandez, J.R., Castro-Fresno, D., (2011),"Design and construction of an experimental pervious paved parking area to harvest reusable rainwater", Water Science and Technology, 64(9), 1942-1950.

Gould, J., and Nissen-Petersen, E. (2008),"Rainwater catchment systems for domestic supply: Design, construction and implementation." Intermediate Technology Publications Ltd., UK.

Han, M.Y., Mun, J.S., 2008. Particle behavior consideration to maximize the settling capacity of rainwater storage tanks. Water Sci. Technol. 56 (11), 73e79.

Hanson R (ed) (1984) Perspectives on urban infrastructure. Committee on National Urban Policy, National Research Council, Washington, DC. National Academies Press, 216 pp

Hayashi, K. (2006). Environmental indicators for agricultural management: integration and decision making. Int J Mater Struct Reliab, 4(2), 115-127.

Helmreich, B., & Horn, H. (2009). Opportunities in rainwater harvesting. Desalination, 248(1), 118-124.

Herrmann, T., Schimda, U., 2000. Rainwater utilization in Germany: efficiency, dimensioning, hydraulic and environmental aspects, Urban Water Journal 1, 307- 316.

Huston, R., Chan, Y.C., Chapman, H., Gardner, T., Shaw, G., 2012. Source apportionment of heavy metals and ionic contaminants in rainwater tanks in a subtropical urban area in Australia. Water Res. 46 (4), 1121e1132.

IBC, I. (2006). International building code. International Code Council, Inc.(formerly BOCA, ICBO and SBCCI), 4051, 60478-5795.

191

Jones, M. P., and Hunt, W. F., 2010. Performance of rainwater harvesting systems in the southeastern United States. Resources, Conservation and Recycling, 54(10), 623- 629.

Kahinda, J.M., and Taigbenu, A.E. (2011),"Rainwater harvesting in South Africa: Challenges and opportunities." Physics and Chemistry of the Earth 36(14-15), 968– 976.

Karim, M.R., 2010. Assessment of rainwater harvesting for drinking water supply in Bangladesh. Water Sci. Technol. Water Supply 10 (2), 243e249.

Khastagir A, and Jayasuriya N, (2011). Investment evaluation of rainwater tanks. Water resources management, 25(14), 3769-3784. Kinkade-Levario, Heather. (2007). Design for Water. Gabriola Island, British Columbia, Canada: New Society Publishers.

Kirk, S. J., & Dell'Isola, A. J., 1995. Life cycle costing for design professionals (pp. 1-3). New York: McGraw-Hill.

Larsen, S.G., and Burian, S.J. (2011), “Energy requirements for water supply in Utah.” In: The water-energy nexus in the American West, Edited by Doug Kenney, Edward Elgar.

Lassaux, S., Renzoni R, Germanin A (2007),"Life cycle assessment of water: From the pumping station to the wastewater treatment plant." International Journal of Life cycle assessment, 12(2), 118-126.

192

Liang X, Van Dijk M P, (2011). Economic and financial analysis on rainwater harvesting for agricultural irrigation in the rural areas of Beijing. Resources, Conservation and Recycling, 55, 1100e1108.

Lynch, D., Deborah, D.K. (2010),"Water efficiency measures at Emory University, Journal of Green Building." 5(2), 41-54.

Mayer P W, William B D, (1999). Residential End Uses of Water. American Water Works Association Research Foundation (AWWA), Denver.

McCuen R, (2004). Hydrologic Analysis and Design, (third ed.)Pearson Education Inc, Upper Saddle River, NJ.

Mitchell V G, (2007). How important is the selection of computational analysis method to the accuracy of rainwater tank behaviour modelling?. Hydrological processes, 21(21), 2850-2861.

Myers, L.E., 1975. Water Harvesting-2000 B.C. to 1974 AD. IN: G. Frasier Editor, Proc. Water Harvesting Symposium. Phoenix, Arizona 26-28 Mar 1974. ARS W-22 USDA Agricultural Research Service. Pgs 1-5.

National Climatic Data Center, 2015 (NCDC). Available online from: http://www.ncdc.noaa.gov/monitoring-references/maps/us-climate- regions.php#references. Accessed April 2015.

National Green Building Adoption Index, 2014. Available online from: http://www.cbre.com/o/international/AssetLibrary/Green-Building-Adoption- Index-2014_Final.pdf

Nolde, E. (2007),"Possibilities of rainwater utilisation in densely populated areas including precipitation runoffs from traffic surfaces." Desalination, 215(1-3), 1-11. 193

Pachpute, J. S., Tumbo, S. D., Sally, H., & Mul, M. L. (2009). Sustainability of rainwater harvesting systems in rural catchment of Sub-Saharan Africa. Water Resources Management, 23(13), 2815-2839.

Palla, A., Gnecco, I., & Lanza, L. G., 2011. Non-dimensional design parameters and performance assessment of rainwater harvesting systems. Journal of Hydrology, 401(1), 65-76.

PE International, 2014. Gabi database. http://www.pe-international.com.

Peuportier, B. L. P., 2001. Life cycle assessment applied to the comparative evaluation of single family houses in the French context. Energy and buildings, 33(5), 443-450.

Proenca L C, Ghisi E, Tavares DF, Coelho G M, (2011). Potential for electricity savings by reducing potable water consumption in a city scale. Resources, Conservation and Recycling, 55, 960-965.

Racoviceanu, A. I., & Karney, B. W., 2010. Life-cycle perspective on residential water conservation strategies. Journal of Infrastructure Systems,16 (1), 40-49.

Rahman, A., Dbais, J., Imteaz, M. (2010),"Sustainability of Rainwater harvesting Systems in Multistorey Residential Buildings."American J. of Engineering and Applied Sciences, 3 (1), 889-898.

Reid, M. (1982), “Lessons of history in the design and acceptance of rainwater cistern systems.” Proceedings of the International Conference on Rainwater Catchment Systems, Honolulu, Hawaii.

Reidy, P. C. (2008). Integrating rainwater harvesting and stormwater management infrastructure: Double benefit–single cost. In 2008 International Low Impact Development Conference.

Roy, P., Nei, D., Orikasa, T., Xu, Q., Okadome, H., Nakamura, N., & Shiina, T. (2009). A review of life cycle assessment (LCA) on some food products.Journal of food engineering, 90(1), 1-10.

194

Sample, D.J., Liu, J (2014), “Optimizing rainwater harvesting systems for the dual purposes of water supply and runoff capture” Journal of Cleaner Production 75, 174-194

Saving Water in Office Building, EPA, 2012. Available online from: http://www.epa.gov/watersense/commercial/docs/factsheets/offices_fact_sheet_50 8.pdf. Last accessed July 2015.

Stout, D. T., Walsh, T. C., & Burian, S. J. (2015). Ecosystem services from rainwater harvesting in India. Urban Water Journal, (ahead-of-print), 1-13.

Tam, V.W.Y., Tam. L., and Zeng, S.X. (2010),"Cost effectiveness and tradeoff on the use of rainwater tank: An empirical study in Australian residential decision-making." Resources Conservation and Recycling, 54, 178-186 Tavakol-Davani, H., Burian, S., Shugart-Schmidt, W., Chaney, M., Apul, D., Devkota, J., (2015) Comparing Long-Term Performance and Associated Life-Cycle Costs of Green and Gray Infrastructures to Control Combined Sewer Overflow.

TxDOT, Hydraulic Design Manual, Texas, (2009) Available online from: http://onlinemanuals.txdot.gov/txdotmanuals/hyd/hyd.pdf (accessed April 2015)

U.S. Environmental Protection Agency, eGRID Ninth edition with year 2010 data (Version 1.0). 2014.

U.S. EPA. (2002)," The clean water and drinking water infrastructure gap analysis." EPA- 816-R-02-020, U.S. EPA Office of Water, Washington, DC.

U.S. EPA. (2008). “Combined sewer overflows: Demographics.” 〈http://cfpub.epa.gov/npdes/cso/demo.cfm〉 (Jul. 5, 2014).

U.S. EPA. (2014a). “Greening CSO plans: Planning and modeling green infrastructure for combined sewer overflow (CSO) control.” EPA-832-R-14-001, Washington, DC.

U.S.EPA. (2009),"Pipe leak detection technology development."EPA/600/F-09/019, http://www.epa.gov/awi/pdf/600f09019.pdf.

US Census, 2000. Commercial Buildings -- Number and Size, by Principal Activity. Available online from: http://www.allcountries.org/uscensus/1227_commercial_buildings_number_and_s ize_by.html

USEPA, Water Conservation Plan Guidelines. Available online from: http://www.epa.gov/WaterSense/docs/app_b508.pdf. Accessed May 2015.

Vargas-Parra M V, Villalba G, and Gabarrell X, (2013). Applying exergy analysis to rainwater harvesting systems to assess resource efficiency.Resources, Conservation and Recycling, 72, 50-59.

Vickers A. 2001. A Handbook of water use and conservation. Water Flow Press. Amherst: MA.

Viessman W and Lewis G L, (2003) Introduction to Hydrology, (fifthth ed.)Prentice Hall, Upper Saddle River, New Jersey Villarreal, E. L., and Dixon, A., 2005. Analysis of a rainwater collection system for domestic water supply in Ringdansen, Norrköping, Sweden. Building and Environment, 40(9), 1174-1184.

Wang R and Zimmerman J B, (2015). Economic and Environmental Assessment of Office Building Rainwater Harvesting Systems in Various US Cities. Environmental science & technology, 49(3), 1768-1778.

Ward S, Memon F, and Butler D, (2010). Rainwater harvesting: model-based design evaluation.

Web Soil Survey. United States Department of Agriculture. Natural Resource Conservation Service. Available online from: http://websoilsurvey.sc.egov.usda.gov/App/HomePage.htm

Yuan T, Fengmin L, and Puhai L, (2003). Economic analysis of rainwater harvesting and irrigation methods, with an example from China. Agricultural water management, 60(3), 217-226.

Zhu, K., Zhang, L., Hart, W., Liu, M., & Chen, H. (2004). Quality issues in harvested rainwater in arid and semi-arid Loess Plateau of northern China.Journal of Arid Environments, 57(4), 487-505.

196