Systems Modeling Approaches to Physical Resource Management

An Industrial Ecology Perspective

Rajib Sinha

Doctoral Thesis

Industrial Ecology Department of Sustainable Development, Environmental Science and Engineering (SEED) KTH Royal Institute of Technology Stockholm, Sweden

2016 Title: Systems Modeling Approaches to Physical Resource Management: An Industrial Ecology Perspective

Author: Rajib Sinha

Copyright: Copyright c 2016 Rajib Sinha All rights reserved.

Papers I-V are reprinted with permission. Paper I c 2015 John Wiley & Sons, Ltd. and ERP Environment Paper II c 2016 Elsevier B.V. Paper III c 2010 Elsevier B.V. Paper IV c 2015 Elsevier Ltd. Paper V c 2016 Elsevier Ltd.

Registration: TRITA-IM-PHD 2016:04 ISBN: 978-91-7729-077-3

Contact information: Industrial Ecology Department of Sustainable Development, Environmental Science and Engineering (SEED) School of Architecture and the Built Environment, KTH Royal Institute of Technology Teknikringen 34, SE-100 44 Stockholm, Sweden Email: [email protected] www.kth.se

Printed by: Universitetetsservice US-AB, Stockholm, Sweden, 2016

ii Preface

This doctoral thesis1 attempts to capture specific, but also diverse, research interests in the field of industrial ecology. It therefore has a broader scope than a normal PhD thesis. Here, I would like to share my academic journey in composing this thesis to guide the reader in understanding the content and the background.

At secondary and higher secondary school (9th-12th grade), I was most interested in mathematics and physics. This led me to study engineering and I completed my BSc in Civil Engineering at BUET, Dhaka, Bangladesh. During my undergraduate studies, I developed a great interest in structural engineering. As a result, I chose finite element analysis of shear stress as my Bachelor’s degree project. The title of the dissertation was Analysis of Steel-Concrete Composite Bridges with Special Reference to Shear Connectors.

With my interest in environmental engineering and sustainable development, I com- pleted my MSc in Sustainable Technology at the Division of Industrial Ecology, KTH, Stockholm, Sweden. The title of my Master’s thesis was Modelling Source and Copper Fate in Lake R˚ackstaTr¨ask,Stockholm: Sediment Copper Content as an Indicator of Urban Metal Emissions. I was then given the opportunity to continue working on the study to help to publish the work, after which I worked as a research engineer at KTH (September 2009-February 2012) in consecutive small projects on: (i) radionuclide sorption onto soil samples, (ii) reactive transport modeling for radionuclide migration, (iii) modeling the effects of mixing and spreading on biodegradable benzene, toluene, ethylbenzene, and xylenes (BTEX) in groundwater, (iv) household metabolism in a global perspective, and (v) energy systems modeling.

During that period, I became interested in research methodologies rather than a specific research topic, in particular systems analysis, dynamic modeling, and computer simulation, especially systems dynamics and agent-based modeling.

I enrolled as a post-graduate student on March 2012. The project focused on modeling a decentralized energy system for future cities and evaluating how the system can respond to different optimization criteria, such as economy, renewables, greenhouse gas emissions, low carbon, resilience, etc. An agent-based modeling approach was chosen for the systems modeling.

After studying for six months, I moved to another research group, Sustainable

1A doctoral thesis can be written either as a dissertation manuscript (monograph) or as a compi- lation thesis with a summarizing chapter (kappa) followed by a compendium of articles. I chose the latter version.

iii Production and Consumption, with a broader systems perspective. This group was working with production and consumption systems including a broader systems approach from two different angles: the design perspective and the waste management perspective. Through my interest in systems modeling, I combined the two approaches by modeling energy and material metabolism. During my post-graduate studies, my research specifically focused on the modeling aspects of industrial ecology described in this thesis.

This cover essay is based on the results presented in the appended papers (Papers I-V) and briefly describes the most important findings of the work. The modeling approaches used in Papers I-V are discussed in relation to understanding complex systems and accounting to monitor environmental performance. In this synthesis of the tools and modeling approaches used in Papers I-V, it was challenging to present a consistent storyline due to the applications in different environmental management contexts. However, this multi-application background information enriched the dis- cussion of a broader systems approach and the significance of understanding complex systems and monitoring the environmental performance of management actions.

The pressure-based, driver-oriented approach that I developed in my licentiate thesis, following the work of Prof. Frostell and Dr. Song, helped me synthesize the modeling approaches into a unified narrative. In addition, the pressure-based, driver-oriented approach showed tremendous potential to disseminate proactiveness in environmental management. As a result, this approach became one of the key messages in the thesis. I hope that the reader will agree!

Rajib Sinha, Stockholm, May 2016

iv Acknowledgments

I am grateful to Industrial Ecology, KTH for financial support to conduct the research and complete this thesis. I acknowledge financial support from the City Of Stockholm to conduct the research on environmental load profile.

My deepest gratitude to my supervisor, Prof. Bj¨ornFrostell, for his inspiration and support in conducting this research and broadening my perspective on sustainability thinking. He gave me freedom to explore; at the same time, he held me on track. Without his guidance, I would not have been able to produce this thesis. I am also grateful to my co-supervisor, Assoc. Prof. Maria Malmstr¨omfor her guidance and continuous support with constructive comments and suggestions. She literally taught me how to write a good scientific article. At the end of my PhD, I had the privilege to obtain a different perspective on my research from my new co-supervisor, Assoc. Prof. Miguel Brand˜ao. I would like to thank him for his suggestions, advice, and guidance. Thank you my dear friends, colleagues, and co-authors of the appended pa- pers, Dr. Jagdeep Singh and Dr. Rafael Laurenti, for all your support in our team work.

I would like to thank my internal quality reviewer, Prof. Per Lundqvist, for taking the time to read my cover essay and the appended papers. His comments and suggestions helped me enormously to improve the quality of the thesis. I also thank Assoc. Prof. Nils Brandt (Retired), Assoc. Prof. Getachew Assefa, Assoc. Prof. Lars Olsson, Maria Lennartsson, Dr. Qing Cui, Dr. Jiechen Wu, and Guanghong Zhou, co-authors of my papers. Thank you Dr. Daniel Franzen for reviewing the sammanfattning (abstract in Swedish) of my thesis. I am also very thankful to the anonymous reviewers of the paper.

I thank Prof. Ronald Wennersten for giving me the opportunity to continue my research career by accepting me as a post-graduate student. Special thanks goes to Karin Orve and Monika Olsson for their support and inspiring conversations throughout the year. I also wish to thank Per Jakobsson, Dr. Asa˚ Moberg, Prof. Jos´e Poting, Assoc. Prof. Cecilia Sundberg, Dr. Annika Carlsson-Kanyama, Assoc. Prof. Fredrik Grndahl, Larsgran Strandberg, Dr. Olga Kordas, Kosta Wallin, Dr. Markus Rob´ert,Per Olof Persson, Prof. Saiful Amin, Assoc. Prof. Koustuv Dalal, and Manik Lal Mandal for their encouragement, support, and good discussions during my research.

I would like to express my deepest gratitude to all my teachers.

I thank all my friends and colleagues at KTH who helped me make this memorable journey: Joseph, JB, Graham, Hossain, Anders, Louise, David, Hongling, Mauricio, Olekssi, Kate, Mohammad, Aram, Olena, Linus, Sofia, and Yevgeniya. I thank my friends and class mates, Indranil, Sujan Sikder, and Sujan Dutta, for inspiring me to

v environmental engineering during my Bachelor’s degree.

Thank you all, my dear friends, for your help and support during my student life: Topu, Ranjan, Sudip, Koushik, Nayan, Bichitro, Ashis, Surajit, Rima, Lakshmi, Nipa, Biplob Da, Anup Da, Tuhin, Rubel, Fatik, Suman, Shovon, Tanjil, Rezaul, Sushmita, Subarna, Madhumita Dalal, Apu Da, and many more (please count yourself in this list!).

On a personal note, it would not be possible for me to have reached my current position without the unconditional love and support of my parents (Ranajit Sinha and Rekha Sinha) and my younger brother (Raju Sinha). I also would like to thank all my relatives for supporting and inspiring me.

Thank you ... Snigdha!

Rajib Sinha, Stockholm, May 2016

vi Abstract

Many of the present problems that we are facing arise as unanticipated side-effects of our own actions. Moreover, the solutions implemented to solve important problems often create new problems. To avoid unintended consequences, understanding complex systems is essential in devising policy instruments and in improving environmental management. Thus, this thesis investigated systems modeling approaches to under- stand complex systems and monitor the environmental performance of management actions. The overall aim of the work was to investigate the usefulness of different systems modeling approaches in supporting environmental management. A driver- based, pressure-oriented approach was adopted to investigate systems modeling tools. Material/substance flow analysis, environmental footprinting, input-output analysis, process-based dynamic modeling, and systems dynamics modeling approaches were applied in different cases to investigate strengths and weaknesses of the tools in generating an understanding of complex systems. Three modeling and accounting approaches were also tested at different systems scales to support environmental mon- itoring. Static modeling approaches were identified as fundamental to map, account, and monitor physical resource metabolism in production and consumption systems, whereas dynamic modeling showed strengths in understanding complex systems. The results suggested that dynamic modeling approaches should be conducted on top of static analysis to understand the complexity of systems when devising and testing policy instruments. To achieve proactive monitoring, a pressure-based assessment was proposed instead of the mainstream impact/state-based approach. It was also concluded that the LCA community should shift the focus of its assessments to pressures instead of impacts.

Keywords: complex systems modeling, environmental accounting and monitoring, en- vironmental footprint, industrial ecology, pressure-based driver-oriented approach

vii Sammanfattning

M˚anganuvarande utvecklingsproblem har uppst˚attsom of¨orutseddabiverkningar av m¨anniskans egna handlingar. De l¨osningarsom pr¨ovats har i sin tur ofta ska- pat nya problem. Det d¨arf¨or viktigt att f¨orst˚a hur komplexa system fungerar och att utforma styrmedel och ledningssystem som minimerar risken f¨oro¨onskade bieffekter. Den h¨aravhandling har anv¨ant olika modelleringsmetoder f¨oratt ¨oka f¨orst˚aelsenf¨orkomplexa system och bidra med kunskaper om hur milj¨oprestanda och f¨orvaltnings˚atg¨arder kan f¨oljasupp p˚a ett mer effektivt s¨att. Det ¨overgripande syftet med arbetet var att unders¨oka anv¨andbarhetenav olika modelleringsmetoder f¨or att effektivisera den fysiska resurshanteringen i samh¨allet. I arbetet har ett fl¨odesbaseratoch akt¨orsinriktatarbetss¨att(pressure-based and driver-oriented ap- proach) anv¨ants i modelleringen. Material- och substansfl¨odesanalys,milj¨ofotavtryck, input-output analys, processbaserad dynamisk modellering och systemdynamiska modelleringsmetoder studerades f¨oratt unders¨oka styrkor och svagheter hos de olika metoderna/verktygen. Tre olika modellerings- och redovisningsmetoder f¨oratt st¨odja milj¨o¨overvakning testades ocks˚a i olika systemskalor. Statiska modelleringsmetoder (r¨akenskaper) identifierades som grundl¨aggande f¨or att kartl¨agga, kontof¨ora och ¨overvaka den fysiska resursmetabolismen i produktions- och konsumtionssystem, medan dynamisk modellering visade sin styrka i att skapa f¨orst˚aelsef¨orkomplexa system. Resultaten pekar p˚a att dynamiska modelleringsmetoder b¨or anv¨andas som ett komplement till statiska analyser f¨oratt f¨orst˚a komplexiteten i systemen n¨arman utformar och testar styrmedel. F¨oratt uppn˚a proaktiv ¨overvakning b¨or fl¨odesbaserader¨akenskaper utnyttjas i st¨orreutstr¨ackning i st¨alletf¨orden vanliga tillst˚ands-och p˚averkans¨overvakningen (state/impact monitoring). En viktig slutsats ¨ar d¨arf¨oratt LCA-samfundet b¨orflytta fokus i sina bed¨omningar fr˚anp˚averkan till fl¨oden.

Nyckelord: modellering av komplexa system, milj¨or¨akenskaper och milj¨o¨overvakning, milj¨op˚averkan, industriell ekologi, fl¨odesbaserad ¨overvakningakt¨orsorienterad strategi.

viii Contents

Preface iii

Acknowledgments v

Abstract vii

Sammanfattning viii

List of Papers xi Other publications ...... xii

List of Abbreviations xiii

1 Introduction 1 1.1 Background ...... 1 1.2 Aims and objectives ...... 3

2 Theoretical background 5 2.1 Societal metabolism ...... 5 2.2 Industrial ecology ...... 5 2.3 Industrial ecology and modeling complexity ...... 7 2.4 Complex systems ...... 7 2.4.1 Leverage points ...... 9 2.5 The DPSIR framework ...... 10 2.5.1 Pressure-based, driver-oriented approach ...... 10 2.6 Tools and approaches used in industrial ecology research ...... 11 2.6.1 Accounting and static approach ...... 11 2.6.2 Dynamic approach ...... 14

3 Research design 17 3.1 Overarching research design ...... 18 3.2 Selection of environmental issues ...... 20

4 Modeling 25 4.1 Modeling in Paper I ...... 25 4.2 Modeling in Paper II ...... 25 4.3 Modeling in Paper III ...... 28 4.4 Modeling in Paper IV ...... 29 4.5 Modeling in Paper V ...... 31

ix 5 Results and discussion 35 5.1 Results of Papers I-V ...... 35 5.1.1 Paper I ...... 35 5.1.2 Paper II ...... 36 5.1.3 Paper III ...... 38 5.1.4 Paper IV ...... 39 5.1.5 Paper V ...... 40 5.2 Synthesis and key findings ...... 44 5.2.1 Pressure-based, driver-oriented approach ...... 44 5.2.2 Understanding the system to be managed: static vs dynamic . . 48 5.2.3 Environmental accounting to monitor the performance . . . . . 53 5.3 Further research ...... 54

6 Conclusions 57

References 61

x List of Papers

Paper I: Laurenti, R., Sinha, R., Singh, J., & Frostell, B., 2015. Towards Addressing Unintended Environmental Consequences: A Planning Framework. Sustainable Development, 24, 117.

I took part in the research work, contributed to the literature survey, results analysis and development of the framework, and made a significant contribution to the application example of the framework for the case of closing the material flow loop in mobile phone product systems.

Paper II: Sinha, R., Laurenti, R., Singh, J., Malmstr¨om,M. E., & Frostell, B., 2016. Identifying ways of closing the metal flow loop in the global mobile phone product system: a system dynamics modeling approach. Resources, Conservation and Recycling, 113, 65-76.

I was responsible for the research idea and the justification for the research, modeling, data collection, analysis, and drafting the manuscript. I was also responsible for publication of the paper.

Paper III: Cui, Q., Brandt, N., Sinha, R., & Malmstr¨om,M. E., 2010. Copper content in lake sediments as a tracer of urban emissions: Evaluation through a source-transport-storage model. Science of the Total Environment, 408 (13), 2714-2725.

I contributed to development of the model and simulation, data collection, anal- ysis, cross-checking with other type of modeling and results (not included in the paper) specifically for the Lake R˚ackstaTr¨ask. In addition, I cross-checked the calculations for other lakes.

Paper IV: Frostell, B., Sinha, R., Assefa, G., & Olsson, L-E., 2015. Modeling both direct and indirect environmental load of purchase decisions: A web-based tool addressing household metabolism. Environmental Modelling and Software, 71, 138-147.

I contributed to development of the model, data collection, and analysis. In addition, I was responsible for the literature review and for writing the paper. I was also responsible for publication of the paper.

xi Paper V: Sinha, R., Lennartsson, M., and Frostell, B., 2016. Environmental Footprint Assessment of Building Structures: A comparative study. Building and Environment, 104, 162-171.

I was responsible for the research idea and justification for the research, modeling, data collection, analysis, and drafting the manuscript. I was also responsible for publication of the paper.

Other publications not included in the thesis Laurenti, R., Sinha, R., Singh, J., & Frostell, B., 2015. Some pervasive challenges to sustainability by design of electronic products − a conceptual discussion. Journal of Cleaner Production, 108, 281-288.

Laurenti, R., Singh, J., Sinha, R., Potting, J., & Frostell, B., 2015. Unintended Environmental Consequences of Improvement Actions: A Qualitative Analysis of Systems’ Structure and Behavior. Systems Research and Behavioral Science, 33, 381399.

Singh, J., Laurenti, R., Sinha, R. & Frostell, B. 2014. Progress and challenges to the global waste management system. Waste Management and Research, 32(9), 800-812.

Zhou, G., Singh, J., Wu, J., Sinha, R., Laurenti, R., & Frostell, B., 2015. Evalu- ating low-carbon city initiatives from the DPSIR framework perspective. Habitat International, 50, 289-299.

Licentiate thesis

Sinha, R., 2014. Industrial Ecology approaches to improve metal management: Three modeling experiments. Licentiate thesis at Industrial Ecology, KTH, Stockholm, Sweden.

xii List of Abbreviations ABM agent-based modeling ALCA attributional life cycle assessment BAU business as usual BTA total floor area BTEX benzene, toluene, ethylbenzene, and xylenes CGE computable general equilibrium CLCA consequential life cycle assessment CLD causal loop diagram CO2eqv carbon dioxide equivalent COICOP classification of individual consumption by purpose DC developing countries DfE design for environment DPSIR drivers-pressures-state-impact-response EEA European Environmental Agency EFA element flow analysis ELP environmental load profile EPA European Protection Agency EoL end-of-life EU European Union e-waste electronic waste GHG greenhouse gas IE industrial ecology IC industrialized countries ICT information and communication technology IOA input-output analysis IOT input-output table ISO international organization for standardization MFA material flow analysis MRIO multi-regional input-output NOxeqv nitrogen oxides equivalent LCA life cycle analysis LCI life cycle inventory PB process-based PDCA plan-do-check-act R1 research question 1 R2 research question 2 SD system dynamics SCB Statistics Sweden SFA substance flow analysis

xiii xiv 1 Introduction

1.1 Background

Economic growth, urbanization, and rapid population growth have given rise to material-intensive production and consumption in modern society (Frosch and Gal- lopoulos, 1989; Fischer-Kowalski and H¨uttler,1998; Fischer-Kowalski and Haberl, 1998; Tukker et al., 2008; Jackson, 2011). As a result, a linear model of consumption (produce-use-dispose) has been endorsed, with constantly increasing pressures on the environment from waste, hazardous pollutant emissions, resource depletion, resource waste, carbon emissions to the atmosphere, and many more (Meadows et al., 1972, 2004; Wenheng and Shuwen, 2008; Jackson, 2011). These environmental pressures lead to impacts on society, the economy, and the environment. In a response to mitigate these impacts, sustainability efforts have emerged focusing on reducing unsustainability. According to Ehrenfeld and Hoffman (2013), reducing unsustainability is not the same as creating sustainability, and there is a need for a profound shift in values in sustainability thinking.

There is a great belief that technology will provide solutions (Ehrenfeld and Hoffman, 2013). However, technology alone cannot solve problems (Ehrenfeld and Hoffman, 2013). We need to find the root causes of problems (Ruth and Davidsdottir, 2008, 2009; Jackson, 2011; Ehrenfeld and Hoffman, 2013). A root cause can be defined as the initial point that generates the problem following a chain of cause-effects. The drivers that operate or control the root causes and the chain of causal relations are the subject of this thesis.

To investigate drivers in complex production and consumption systems for proactive management, systems thinkers, and modelers (e.g., researchers, managers, policy makers, etc.) have to understand interconnectedness/interactions and dynamics among different components (Checkland, 1981; Laszlo, 1996; Meadows, 1999; Checkland, 2000; Bossel, 2007b,a; Meadows and Wright, 2008). Many of the present problems that we are facing arise as unanticipated side-effects of our own actions (Meadows, 1999; Bossel, 2007b; Laurenti et al., 2015a). Furthermore, the solutions implemented to solve important problems often create new problems (Laurenti et al., 2015a). Effective decision-making and learning in a world of growing complexity requires us to become systems thinkers to expand the boundaries of our mental models and develop tools to understand how the structure of complex systems creates their behavior (Churchman, 1968; Checkland, 1981; Laszlo, 1996; Bossel, 2007b; Meadows and Wright, 2008).

Industrial ecology (IE) provides several powerful analytical tools (e.g., life cycle as- sessment (LCA), material/substance flow analysis (MFA/SFA)) that provoke thinking about the holistic system (Ayres and Ayres, 2002; Graedel and Allenby, 2003; Bollinger

1 et al., 2012; Clift and Druckman, 2015). However, this type of modeling tends to focus primarily on static analysis (Ruth and Davidsdottir, 2008, 2009) although it covers a wide range of time scales and organizational interactions (Ayres and Ayres, 2002; Graedel and Allenby, 2003). For example, substance flow analysis is a frequently used industrial ecology tool for studying material flows, but it is unable to provide information about future developments in material flow patterns and how to find leverage points (i.e., drivers or root causes) to intervene in the system. These tools also often do not include the socio-economic drivers that create complexity (Ruth and Davidsdottir, 2008, 2009; Dijkema and Basson, 2009; Boons and Howard-Grenville, 2009).

Ruth and Davidsdottir (2008); Ruth (2009); Baynes (2009) advise the industrial ecology community to use dynamic systems modeling in order to overcome the limita- tions of the existing linear models and include complex systems thinking for improved decision support. Many dynamic systems modeling tools have been used to date in the research field of industrial ecology, e.g., agent-based modeling (Chappin and Dijkema, 2010; Bollinger et al., 2012), systems dynamics (Forrester, 1958; Meadows et al., 1972, 2004), and process-based dynamic modeling (McGuire et al., 2001). However, there is a gap in the existing literature as regards synthesizing static and dynamic tools used in the field of industrial ecology to address complexity. Therefore, different systems modeling approaches and tools used in the research field of industrial ecology were studied in this thesis to examine the importance of understanding complexity in production and consumption systems, by asking the following research question:

R1. What systems modeling approaches can foster an understanding of complex sys- tems, in order to explore and improve physical resource management in production and consumption systems?

Monitoring the environmental performance of physical resource management in production and consumption systems or any kind of societal metabolism is very important in understanding the development trend, identifying upcoming threats, and checking management strategies and the understanding of complexity (Burstr¨om, 1999). Mainstream monitoring approaches mostly cover concentrations of contaminants in the environment and impacts on human health, the economy, and the environment (Simms and Thomas, 1982; Slocombe, 1992; Kumar et al., 2013). End-of-pipe discharges or emissions are also frequently monitored under environmental regulations, e.g., industrial waste disposal. However, there is a lack of an established system to monitor physical resource flows (and stocks) in production and consumption systems in a life cycle perspective, in order to provide feedback to the drivers and to policy makers.

2 The European Environmental Agency (EEA) uses the DPSIR framework (drivers- pressures-state-impact-response) to organize information for monitoring and reporting the state of the environment and impacts for policy measures (Kristensen, 2004). This framework encourages researchers to look for the drivers utilizing the causal chain from pressures, states, and impacts (Kristensen, 2004; Ness et al., 2010; Sinha, 2014; Zhou et al., 2015). Society and policy makers generally take necessary measures on a problem when an impact to human health, an ecosystem, or an economy is perceived (Smeets and Weterings, 1999; Kristensen, 2004). Thus, mainstream approaches to find the drivers are generally impact-based, i.e., start from observation of environmental and/or biodiversity change impacts by monitoring state changes (Kristensen, 2004; Song and Frostell, 2012). Employing the impact-based mainstream approach to combat a problem is considered a reactive response to finding the drivers (Song and Frostell, 2012; Sinha, 2014). In addition, the impact-based approach can overlook other problems created by the same drivers of a potential problem. Therefore, the existing mainstream approach should be considered a predominantly end-of-pipe response. Rather than mainstream approaches, this thesis employed a pressure-based, driver- oriented approach (based on a proactive approach adopted from Sinha (2014)) to foster environmental accounting and monitoring by addressing the following research question:

R2. What systems modeling approaches can foster environmental accounting to monitor the environmental performance of production and consumption systems for proactive physical resource management?

1.2 Aims and objectives The overall aim of the thesis was to investigate the advantage of different systems modeling approaches to support environmental management. The systems modeling work in the thesis mainly focused on a driver-oriented approach to understand physical resource flows in order to foster management of production and consumption dynamics. Thus the thesis attempted to outline how can we influence the drivers and how can modeling help in this.

To address the research questions, the thesis investigated systems modeling ap- proaches to understand complex systems in order to help devise improved management strategies. In order to follow up physical resource management in production and consumption systems, the thesis also investigated systems modeling approaches for improved environmental management to support pressure-based monitoring systems. Two specific objectives of the thesis were set, to respond to the two research questions (R1 and R2) 2:

2The thesis objectives directly reflect the research questions.

3 1. Assess the strengths and weaknesses of different systems modeling approaches to understand complex systems and explore and improve physical resource manage- ment in production and consumption systems (Papers I-V).

2. Demonstrate three systems modeling approaches at different systems scales to fos- ter environmental accounting to monitor the environmental performance of produc- tion and consumption systems for proactive physical resource management (Papers III-V).

A pressure-based, driver-oriented approach, considered proactive management (Sinha, 2014), was adopted to investigate systems modeling approaches. The approach is discussed in Section 2.5, illustrating the DPSIR framework. The significance of the approach is further discussed in Chapter 5. This approach was employed in Papers I-V to perform different systems modeling techniques applied in different suitable cases. The case selections with their systemic scales and levels are discussed in Chapter 3.

The overall aims and objectives of the thesis attempted to address two fundamentally important issues of environmental management: i) understanding complex systems and ii) how to monitor progress in environmental management. These issues are often generic to methodology, rather than case-specific. Thus, the present thesis is based on cases and the intention was to make generalizations based on these and lift them to a generic level. The analysis did not focus on a single case, because no case covered all the issues considered in the thesis, due to lack of data/information and necessary systems scales and levels. On the other hand, since a single systems modeling approach is an essential method for many cases (Keirstead, 2014), applications of different tools in different cases bring multi-disciplinary learning and discussion, which was very important for this thesis as well as in the study field of Industrial Ecology and sustainability science.

One might argue that it is necessary to assess the same case study with different tools in order to make a proper comparison to fulfill the first objective of the thesis. While acknowledging the benefit of same case studies, this thesis argues that complexity is not case-specific, but is rather generic to many cases and depends on the intrinsic characteristics of the system, e.g., feedback dynamics and non-linearities (Meadows and Wright, 2008; Balestrini Robinson, 2009). These characteristics that constitute complexity are discussed in Chapter 2. The strengths and weaknesses of the systems modeling approaches are compared based on the criteria of complex systems in Chapter 5.

4 2 Theoretical background

The thesis synthesized five modeling studies with different scopes in the research field of industrial ecology conducted using overlapping industrial ecology approaches (Papers I-V). The work adopted a pressure-based approach to environmental problems with the aim of improving physical resource management. The DPSIR framework was utilized to illustrate the pressure-based approach. This chapter describes some terms, concepts, and framework approaches which are essential to understanding the results presented in subsequent chapters.

2.1 Societal metabolism The literal meaning of the Greek word metabolism is ‘transformation’ or ‘conversion’ and the term is mainly used in biology (Baccini and Brunner, 2012). A metabolic approach to comprehending complex systems was first studied in environmental science and technology focusing on ecology (Odum and Odum, 1959; Odum, 1989; Baccini and Brunner, 2012). Later, Wolman (1965) coined the term metabolism of cities. Now, the word metabolism is widely used in physical resource management in the human system to comprehend all physical flows and stocks of matter and energy within the anthroposphere (Baccini and Brunner, 1991, 2012).

The study of societal metabolism (Fischer-Kowalski and H¨uttler,1998) in human systems is continuously increasing in sustainable growth discussions, analytical think- ing, and policy formulation (Fischer-Kowalski and H¨uttler,1998; Fischer-Kowalski and Haberl, 1998). Research on societal metabolism investigates physical resource flows in order to identify hot-spots and provide suggestions for measures promoting sustain- ability (Wolman, 1965). The sustainable development concept basically comprises the ‘triple bottom line approach’, where three partly overlapping circles (economic, social, environmental) have equal weight. Following Frostell (2013), this thesis argues that environmental sustainability should be the most strongly weighted factor, because of the inter-dependency and physical metabolic interaction of the formal economy and human activities with the global system.

2.2 Industrial ecology Industrial ecology was born from the studies of the eco-efficiency of industrial processes (Baccini and Brunner, 2012) under the umbrella of a dedicated journal. Later, the scope of the journal and the study field of industrial ecology evolved by integrating human systems into the environment, focusing on a broader systems approach to production and consumption systems. Societal metabolism (Fischer-Kowalski and

5 H¨uttler,1998) and industrial metabolism (Ayres, 1989) are the core research topics of industrial ecology.

Within the sustainability debate, industrial ecology evolved by asking why industrial systems do not imitate ecosystems(Frosch and Gallopoulos, 1989; Graedel and Allenby, 2003; Kapur and Graedel, 2004). Industrial ecology employs a metaphor analogous to the biological ecosystem, where the study focuses on material and energy flows through industrial metabolism (Graedel and Allenby, 2003; Kapur and Graedel, 2004). In ecosystems, trophic levels are primary producers, herbivores, carnivores, and detritus decomposers (which receive residues from other trophic levels and regenerate materials that can again flow out to primary producers). Similarly, industrial trophic levels are extractors, producers, consumers (often at several levels), and decomposers. However, the significant difference between biological and industrial systems is that overall loss of resources in biological systems is very low, whereas in industrial systems the losses are enormous. Consequently, industrial systems extract substantial resources outside the industrial system (Graedel and Allenby, 2003, see also Chapter 4).

In modern society, people generally use products and discard them, while nature closes the material loop by recycling the resources. In this societal linear consumption model (Frostell, 2013), decomposers could create a circular flow of resources by reuse and recycling. There are (relatively) few decomposers (compared with producers) in the current production and consumption system. By boosting decomposers, economic incentives and policy instruments could play an important role in creating a circular economy (Geng and Doberstein, 2008). However, within modern socio-technical systems, it is very difficult to assess the outcome from a deliberate action, since it can lead to unintended consequences, for example, rebound effects (Laurenti, 2013).

Imitating nature or natural ecosystems in an industrial system requires eco-cycling, for instance closed loop material flows in the industrial system. According to Ravetz (2000) and Eco-Cycle (2014), the concept of eco-cycling illustrates the metabolism in human systems, whereby substances or resources continuously recirculate within the metabolic industrial system (Fischer-Kowalski and Haberl, 1998; Geels, 2005, 2012). Synonymous concepts in other perspectives are described in the literature, e.g., zero waste (Eco-Cycle, 2014), cradle-to-cradle (McDonough and Braungart, 2010), closed loop system, industrial symbiosis, eco-park (Graedel and Allenby, 2003), and the circular economy (Geng and Doberstein, 2008). In this thesis, an eco-cycle scenario is synonymously used as an industrial ecology principle of a closed loop system (Paper II), where technical nutrients/resources (e.g., metals) circulate in the socio-economic system without entering the lithosphere (Graedel and Allenby, 2003; Preston, 2012; Ranhagen and Frostell, 2014).

6 2.3 Industrial ecology and modeling complexity Industrial ecology uses some powerful tools, e.g., life cycle assessment (LCA), design for environment (DfE), material flow analysis (MFA), and substance flow analysis (SFA), and concepts, e.g., eco-efficiency, dematerialization, and decarbonization, to assess and improve products and activities (Graedel and Allenby, 2003; Kapur and Graedel, 2004). In a systems perspective, the traditional tools used in the industrial ecology community (e.g., LCA, MFA) mainly account for a static system. This could be regarded as simple systems modeling, because it cannot include feedback loops in the modeling approach (Ruth and Davidsdottir, 2008; Dijkema and Basson, 2009; Ruth and Davidsdottir, 2009; Davis et al., 2010; Bollinger et al., 2012). Practical systems − whether environmental, economic, technical, cultural, or societal − are complex and change with time and space. DfE and LCA analyses undoubtedly help in making a product or service better for the environment. However, their lack of inclusion of the systems linking technology, society, and the environment may counterbalance the benefits achieved by these mainstream analytical approaches in industrial ecology. Therefore, industrial ecology tools should incorporate additional approaches to understand complexity in order to take better decisions in environmental management.

2.4 Complex systems For understanding complex systems, a comparison between simple systems and complex systems could give an insight into system behavior. According to Graedel and Allenby (2003), simple systems behave in a linear way, while complex systems are generally nonlinear, sometimes show an abrupt response to a minor change (i.e., leverage), and often evolve with time and space. In a simple system, action can easily be detected by following the cause and its associated effects. On the other hand, the link between a cause and its effect is often difficult to establish in complex systems because of their complex interrelations and dynamics (e.g., positive/negative feedback effects).

In a number of books and articles, various authors attempted to describe systems characteristics that constitute complexity (Nelson, 1976; Edmonds, 1995; Bar-Yam, 1997; Anderson, 1999; Sterman, 2000; Meadows and Wright, 2008; Cotsaftis, 2009; Boccara, 2010). Each publication uses a slightly different set of criteria to describe complexity. Based on the commonly accepted characteristics of complex systems by researchers in complexity science, Balestrini Robinson (2009) summarized a list of important criteria for assessing complexity. These are:

Dynamism: One of the most important characteristics of complex systems is

7 systemic behavior over time. If a static system (i.e., frozen in time) is analyzed, it is not possible to observe complexity as time progress. Complexity arises with its behavior and the behavior can only be observed if systems are analyzed over time.

Inter-dependencies: Complex systems exhibit inter-dependencies between differ- ent subsystems. They are often time-dependent, i.e., they may appear and disappear as a result of changing conditions. This thesis therefore recognizes inter-dependencies in terms of dynamic interactions between systems.

Non-linearities: The interactions in complex systems are not proportional to a cause or an action. As explained above, complex systems exhibit characteristics whereby a small input change may cause a large change in output, and vice versa. These non-linearities happen because of both macro-level effects (e.g., feedback loops) and micro-level effects (e.g., if-then rules).

Emergent, self organizing, and adaptation: Complex systems exhibit no central control and instead have an ability to respond to a change and structure themselves, and often evolve with time and space.

Delays: Delays can be defined as the length of time relative to the rate of systems change (Meadows and Wright, 2008). Delays generally show a strong influence on systems behavior and sensitive leverage points (see below) for policy. Outcome of an action greatly depends on how much longer or shorter the delay would be.

Feedback dynamics: Collective dynamics are often observed in complex systems, exhibiting continuous feedback dynamics between different systems levels.

Number of subsystems: A complex system consists of a set of subsystems. It is important to consider the whole set of subsystems in any analysis of the complexity.

This thesis employed the above criteria in research on sustainable production and consumption to manage physical resources proactively. It is important to set a list of criteria for physical resource management in order to assess the systems modeling tools for understanding complexity. Finnveden and Moberg (2005) assessed systems analysis tools to support decision making in environmental management by establishing two criteria: natural resources and environmental impacts. Based on the literature in the field of industrial ecology (Ayres and Ayres, 2002; Wrisberg et al., 2002; Graedel and Allenby, 2003; Ehrenfeld, 2004; Dijkema and Basson, 2009; Baccini and Brunner, 2012; Keirstead, 2014), in this thesis the systems modeling approaches were evaluated together with the criterion for complex systems of whether the tool was capable of mapping physical resources, assessing environmental concerns, accounting and monitoring, and building characterization potential. The systems modeling approaches

8 were also assessed for data requirements and ease of modeling. A short description of the terms assessed is presented below:

Mapping physical resources: This included mapping physical resource stocks and flows in order to manage them.

Environmental assessment: This included assessment of indicators based on environmental concerns.

Accounting and monitoring: This included accounting physical resources in order to monitor the environmental performance of societal metabolism or any kind of development, in order to understand the development trend and identify upcoming threats.

Scenario characterization potential: This included the potential to illustrate possible alternative futures. The purposes of futures studies are to discover or invent, examine and evaluate, and propose possible, probable, and preferable futures for the purpose of maintaining and improving management of physical resources in production consumption systems. Scenarios can be predictive, explorative, or normative (B¨orjeson et al., 2006).

Data requirements: This included the amount of data required (as well as data availability) to conduct the systems modeling.

Ease of modeling: This assessed the ease of performing the modeling task.

2.4.1 Leverage points

A leverage point in a complex system can be described as a point when a small change in a parameter value leads to a large shift towards the goal or a target parameter (Meadows, 1999). A leverage point approach is generally used in systems dynamics modeling and simulation to achieve better systems performance. In a complex system, this approach explores various possible scenarios and strategies to intervene in the system and is useful in testing policies and investments (Meadows, 1999; Sterman, 2000). In this thesis, the leverage point approach was used to find the most significant potential drivers for proactive physical resource management (Paper II).

9 2.5 The DPSIR framework The DPSIR framework (Figure 1) is a conceptual model of a chain of causal rela- tions between human systems and the environment (Smeets and Weterings, 1999; Kristensen, 2004). The European Environmental Agency (EEA) uses this framework to organize information for monitoring and report the state of the environment and impacts for policy measures (Kristensen, 2004). Indicators are used as a commu- nication tool to communicate the seriousness of an environmental issue (Smeets and Weterings, 1999). The mainstream approach to the use of the DPSIR frame- work is generally to monitor the indicators based on the state and the potential impacts.

Figure 1: The DPSIR framework (driver-pressures-state-impact-response) (Smeets and Weterings, 1999; Kristensen, 2004). The dark grey rectangles represent different DPSIR phases. The thick arrows show causal relations, the thin arrows illustrate the societal responses to the different DPSI phases, and the dashed arrows (additional recommendations to the original DPSIR framework developed by the EU (Smeets and Weterings, 1999)) are proposed to understand and monitor the pressures and state (i.e., not waiting until the impact is perceived) and take a proactive response. The thick-bordered rect- angle represents a pressure-based approach focusing on the driver-pressure-response (DPR) framework (Song, 2012).

2.5.1 Pressure-based, driver-oriented approach Instead of following the mainstream approach, the pressure-based approach was adopted in this thesis work, which was based on the principle on modeling pressures to take proactive responses to drivers and pressures, rather than reactive responses

10 to states or impacts. However, additional emphasis was placed on identifying drivers based on the pressures. Therefore, a pressure-based, driver-oriented approach was employed. The pressure-based approach was first introduced by Song and Frostell (2012) focusing on the drivers-pressure-response framework (Figure 1). The approach was further developed to the pressure-based, driver-oriented approach by Sinha (2014).

2.6 Tools and approaches used in industrial ecology research 2.6.1 Accounting and static approach

Material/substance/element flow analysis (MFA/SFA/EFA) In physical resource management, material flow analysis (MFA) is one of the tools used to understand the societal metabolism. Furthermore, MFA is often used as a communication and decision support tool for environmental management (Brunner, 2004; M˚anssonet al., 2009). It supports environmental governance for sustainable polices (Hashimoto and Moriguchi, 2004; Fischer-Kowalski et al., 2011), for example by examining existing systems to support efficient resource management by establishing mass balances and finding hotspots. In addition, an economy-wide material flow analysis can address material flows with the national economy (Fischer-Kowalski et al., 2011). However, the economy-wide MFA is very aggregated and requires further disaggregation for actual material balance calculation (Fischer-Kowalski et al., 2011). Furthermore, incorporating economic factors in MFA is not sufficient, and integration of social aspects and identification of a proactive way of managing physical resources in a complex system are also required (Binder, 2007a,b).

Substance flow analysis (SFA) of chemical elements and compounds is based on the same principle as MFA and deals with the actual mass balance of chemical elements and compounds (Ayres and Ayres, 2002; Brunner, 2004). Substance flow analysis is mainly used in accounting of pollutants (Ayres and Ayres, 2002) and is not a widely used tool for physical resource management in a broader systems perspective (Ayres and Ayres, 2002; Wrisberg et al., 2002; Brunner, 2004). In addition, SFA represents an aggregated semantic of substances, including elements (i.e., it does not distinguish between chemical compounds and elements). Chemical compounds may transform into other compounds or elements during different processes in their life cycle stages. Therefore, an SFA focusing on chemical compounds is sometimes not applicable in a broader system. In this thesis, element flow analysis (EFA) was therefore used to model complex systems and to promote proactive physical resource management. Element flow analysis was introduced and used in my previous licentiate work (Sinha, 2014), which described it as a complementary tool to MFA and SFA.

Element flow analysis is a systematic assessment of the flows and stocks of chemical

11 elements based on the principle of mass balance (see Appendix). Material and substance flow analysis (MFA/SFA) are also based on the same principle of mass balance. Table 1 compares the main principles of MFA, SFA and EFA, where SFA deals with the flows consisting of chemical compounds and elements, MFA deals with the flows consisting of goods, substances, and elements, and EFA is a subset of SFA and MFA. Thus, EFA acts as a building block that complements MFA and SFA, which allows the metabolism of the periodic table in human systems to be understood. In addition, EFA can link socio-economic aspects with the complex flows in the human system to identify drivers of a complex issue. Furthermore, it permits identification of the flows between the human and environmental systems (i.e., pressure, cf. Figure 3) to make responses to the drivers of the flows.

Table 1: Comparison of the principles of material flow analysis (MFA), substance flow analysis (SFA), and element flow analysis (EFA).

MFA SFA EFA Goods (e.g., computers) Compounds (e.g., carbon- Compounds di-oxide) Elements (e.g., carbon) Elements Elements

In this thesis, element flow analysis (EFA) is proposed for both the modeling of complex systems and the accounting and monitoring of management performance. An element is not generally destroyed or transformed throughout the whole system and, therefore, it can be considered a conservative unit. Furthermore, EFA can integrate comprehensive systems by taking a broader systems perspective of an industrial value chain, following the principle of conservation. The elements can then be connected to tradable substances or materials, for example carbon flow analysis can be expressed in terms of carbon dioxide (CO2) or methane (CH4) or carbon dioxide equivalent (CO2eqv). Thus, EFA enhances the power of MFA and SFA in a complex system aimed at improved resource management by allowing a clearer analysis of industrial metabolism.

Life cycle assessment (LCA) Life cycle assessment (LCA) provides an holistic environmental perspective on a product by assessing potential impacts and resources used throughout the life cycle of the product from raw material acquisition, production and consumption phases, to waste management (ISO 14044, 2006). LCA is an analytical tool that is used in decision-making for products and services by researchers, manufacturers, suppliers, customers, policy-makers, and other stakeholders (Tillman and Baumann, 2004). LCA generally assesses various impact categories that can be defined for environmental

12 impacts, human health, and ecological consequences (Tillman and Baumann, 2004; ISO 14044, 2006).

Conventional LCA generally refers to attributional LCA (ALCA), which accounts for immediate physical flows in the life cycle of a product and takes average data for each unit process (Tillman and Baumann, 2004). On the other hand, consequential LCA (CLCA) broadens the systems boundaries and includes unit processes from inside and outside the product’s immediate system boundary (Weidema, 2003; Earles and Halog, 2011). It also includes the consequence of a change in one flow for the other flows and considers the shifting of burdens between different countries and between different stages in the life cycle (Brand˜ao,2012). The CLCA approach has received increasing attention in the past two decades in the discussion of allocation and economic modeling for policy making and strategic planning (Weidema, 2003; Ekvall et al., 2004; Earles and Halog, 2011; Brand˜ao,2012; Ekvall et al., 2016). The CLCA approach also provides an advantage in decision making by addressing broader systems (Earles and Halog, 2011). A CLCA is therefore capable of revealing a greater systems understanding than an ALCA.

Input-output Input-output table (IOT) or input-output analysis (IOA) generally establishes a relationship between the economic sector and other sectors (Leontief, 1986). Envi- ronmentally extended IOT is widely used in industrial ecology to analyze economic activities in terms of environmental pressures/impacts (Suh, 2009). Material flow analysis is the foundation of the IOT used in industrial ecology and is known as economy-wide material flow accounting (Giljum and Hubacek, 2009; Fischer-Kowalski et al., 2011; Kneese et al., 2015). Conceptually, IOT, environmentally extended IOT, and economy-wide material flow accounting are synonymously used in different studies, and the approach mainly investigates the relationships between sectors.

Researchers in industrial ecology have long had an interest in IOT, for two main reasons: (i) easily accessible national-level databases to analyze networks and (ii) the existing relationship between material flow accounting and economic indicators (Kytzia, 2009). The IOT is an important data source for hybrid LCA models (Joshi, 1999; Suh and Huppes, 2002; Suh and Kagawa, 2009) and material flow accountings (Daniels and Moore, 2001; Daniels, 2002; Bringezu, 2002). An IOT usually assumes that imports from other countries use the same technology as the domestic economy (Suh, 2009). When the technology and energy mix show a significant divergence in the production country compared with the consumption country, the IOT gives an erroneous relationship between different indicators. Therefore, multi-regional input-output (MRIO) tables have attracted the interest of industrial ecologists to solve the above-mentioned issues (Peters and Hertwich, 2009). However, the main shortcoming of input-output models is that flow-driven analysis requires a large

13 aggregation of materials to make a proper mass balance (Suh, 2009). Furthermore, it generally assumes linear relationships between the outputs and the final deliveries (Duchin et al., 2004; Duchin, 2009). Therefore, industrial ecologist need to analyze stock-driven phenomena with a dynamic non-linear model that can adequately address the link between stocks, flows, and drivers.

Computable general equilibrium model Computable general equilibrium (CGE) models are economic models generally used in international trade (e.g., the US international trade commission) to investigate how changes in one sector affect other sectors for policy recommendations (Rose, 1995; Dixon and Jorgenson, 2012). The models assume a market in equilibrium (i.e., demand=supply, and consumption=output), and the exogenous parameters are fixed (Wing, 2004; Sundberg, 2005; Dixon and Jorgenson, 2012). The inter-relations between the parameters and equations in CGE are closely related to IOT. However, CGE models are in many cases considered a better alternative than IOT to optimize policy implications (Nordhaus, 2011), e.g., modeling carbon tax effects (Mongelli et al., 2009). An IOT embedded within the CGE model provides the basis for optimization of policy support (Hanson et al., 2009). Similarly, linking MFA/SFA/EFA with CGE models (as well as IOT) can produce important insights in identifying drivers in complex systems.

2.6.2 Dynamic approach Due to the complexity behind the relations between different parameters in different sectors analyzed in IOA and CGE, dynamic analysis of those models has received increasing interest (Yang, 1999; Shui Zhang, 2001; Sundberg, 2005; Suh, 2009; Dixon and Jorgenson, 2012). However, the primary objectives of those modeling approaches are static-based accounting, and the dynamic modeling of those tools is challenging to construct and solve (Yang, 1999; Shui Zhang, 2001; Suh, 2009). Many types of dynamic and complex systems modeling have been used to assess environmental challenges and sustainability issues, e.g., agent-based modeling, systems dynamics, process-based dynamic modeling, Monte Carlo simulation, discrete event modeling, and gaming-based modeling and simulation. The present thesis predominantly examines three dynamic modeling approaches: process-based modeling3, systems dynamics, and agent-based modeling, which are often used in environmental assessment and management for forecasting/prediction, systems understanding, and social learning (Kelly et al., 2013).

Process-based modeling A process-based dynamic model is a computer-based (generally) mathematical model consisting of a set of ordinary or partial differential equations expressing one or more

3Process-based modeling refers to process-based dynamic modeling in this thesis. The word dynamic is omitted in most instances to avoid complexity, monotony and cumbersome presentation.

14 processes (Buck-Sorlin, 2013). A model of population dynamics (Raftery et al., 1995) and a crop model (Goudriaan and Van Laar, 2012) are examples of process-based models in ecology. In Paper III, a process-based approach was used to model lake dynamics and predict the fate of pollutant in lakes.

This modeling approach is appropriate for exploring natural systems (H˚akanson, 2004) and making predictions for improving management (Cui et al., 2010; Wong et al., 2015; Cuddington et al., 2013). It has also been applied in modeling carbon balance, climate change prediction, and responses (Cao and Woodward, 1998; McGuire et al., 2001).

System dynamics System dynamics (SD) is the study of interconnections between complex systems (Forrester, 1958, 1997). The SD approach is grounded in feedback control theory and non-linear dynamics, which allows the behavior of complex systems to be understood (Forrester, 1997; Sterman, 2000). It also teaches systems thinking and helps users take decisions and develop effective interventions in complex systems. The approach deals with an aggregated view of stocks and flows.

System dynamics has a wide range of applications, from business decision support to sustainability assessments (Sterman, 2000; Achachlouei, 2015). A quintessential example of system dynamics modeling is the book The Limits To Growth (Meadows et al., 1972), which presents a world model explaining physical limits to growth, consequences of short-term policies and theoretical alternatives to transition to an equilibrium (i.e., sustainable) system. After 30 years, the authors revisited their models to validate their findings and explore other future scenarios (Meadows et al., 2004).

Agent-based modeling Agent-based modeling (ABM) is a computational method for simulating interactions and actions between autonomous entities, e.g., humans (both individual and groups), animals, and biophysical entities (Gilbert, 2008; Kelly et al., 2013). This approach couples elements of complex systems, game theory, evolutionary programming, compu- tational sociology, and emergence (Gilbert, 2008). It also allows Monte Carlo methods to address randomness. Agent-based modeling is a bottom-up modeling approach and, usually, applied in social learning to understand the emergent behavior from simple interactions and actions between the agents.

One of the first agent-based models was a dynamic model of segregation by Schelling (1971). Moreover, ABM has also been applied in many environmental management modeling experiments, e.g., consumer behavior in e-commerce to understand envi- ronmental consequences (Xu et al., 2009), energy systems modeling for transition

15 (Chappin and Dijkema, 2010), exploring the effects of information and communication technology (ICT) (Achachlouei, 2015), land use models (Matthews et al., 2007), sustainability assessment of small farms in natural resource management (Astier et al., 2012), and integrated sustainability assessment of water systems (Tabara et al., 2008).

Computable general equilibrium modeling and agent-based modeling were not applied in any of Papers I-V, but these tools are very important in understanding complexity to manage physical resources. Thus this thesis considered these tools to discuss what they can offer compared with the tools actually applied in Papers I-V. The importance and application of these two modeling approaches (in relation to other approaches applied in the papers) to manage physical resources are discussed in the synthesis and key findings of the thesis (Chapter 5).

16 3 Research design

The mental model of production and consumption systems within our research group is illustrated in Figure 2. Today’s stocks are tomorrows resources, wastes, or emissions (Kleijn et al., 2000). Therefore the research group conducts studies on manage not only the flows, but also the stocks. The group has examined production and consumption systems from two different perspectives, a design perspective4 and a waste management perspective5. Through my primary interests in systems modeling, I combined the two approaches by modeling energy and material metabolism in production and consumption systems.

Figure 2: Mental model of production and consumption systems within our research group. The white rectangle represents human society, where production and consumption take place. The outer black rectangle represents the environment. The thicker of the two black solid arrows shows the physical resource flows from the environment to the production and consumption systems in human society, while the thinner solid black arrow represents the outflow from the production and consumption systems to the environment. The inner black rectangle represents the stocks of physical resources in human society. The dashed line with an arrow head shows the reuse and/or recycling of resources within human society. The open white arrows show energy flow (coming in from the sun and going out from reflection of sunlight and from emission of infrared radiation (NASA, 2009)), illustrating that the environment and our planet Earth are open to energy flows.

Instead of following the mainstream impact-based approaches, the pressure-based approach was adopted in this thesis to model metabolic approaches to monitoring pressures in order to make proactive responses to the drivers of the pressures. The thesis also acknowledges that mainstream state/impact-based management is the main approach used currently to manage physical resources in human society. However, in the thesis the pressure-based approach is viewed as a proactive response to the drivers.

4The study was compiled in a PhD thesis - The Karma of Products: Exploring the Causality of Environmental Pressure with Causal Loop Diagram and Environmental Footprint (Laurenti, 2016). 5The study was also compiled in a PhD thesis - Beyond Waste Management: Challenges to Sus- tainable Physical Resource Management (Singh, 2016).

17 3.1 Overarching research design

Figure 3 presents the research design of the thesis mapping the interaction between human systems and the environmental system within the DPSIR framework (cf. Figure 1). In the thesis, human systems represent human society (Figure 2), entailing the socio-technical and economic systems related to production and consumption systems. The drivers and responses in the human systems were identified, and the pressures were designated as the flows between the human systems and the environmental systems (see the dark grey and dotted single head straight arrows in Figure 3), e.g., raw material extraction from the environment and emissions to the environment. The state represents the quality of the environment. Finally, the impact is perceived on both the human systems and the environmental systems.

Figure 3: Research design of the thesis, showing the interaction between human systems and the environment incorporating the DPSIR framework. The upper thick bordered rectangles represent human systems, and the lower one represents the environment. Grey-filled rectangles represent the DPSIR phases. Solid lines/curves with single arrow heads show the direction of the physical resource flows. The arrows from the response suggest taking necessary measures on drivers, pressures, state, and impacts. Dotted lines with two arrowheads show the inter-connections between systems. Dotted rectangles and dotted arrows were not analyzed deeply in the thesis.

18 In mainstream research, impacts on the environmental systems or the human systems are perceived as the sensor to make a response to DPSI phases. The DPSIR framework encourages identification of the drivers of an impact through the causal relationship. Now the impact-based, driver-oriented approach is deeply rooted in mainstream research. Thus the impacts are mostly monitored by expanding the monitoring system with the immediate causal link, state of the environment. The impacts in the mainstream approach have been identified as main problems. However, Meadows et al. (1972, 2004) illustrated that the impacts are not exactly the problems, but rather symptoms of a problem.

According to Meadows et al. (1972, 2004), the main problem lies inside the drivers in human systems, e.g., the way of consumption or growth. In human systems, especially production and consumption systems (which are the focus of this thesis), products, services, and activities are interlinked (see the human systems in Figure 3). In addition, physical resources flow in the systems through complex societal metabolic processes. The drivers that exert pressure on the environmental system are embedded in this societal metabolic system. To make response to the drivers, the societal metabolism needs to be understood, i.e., how human systems exert pressures on the environmental systems and consequently cause the impact. Thus, this thesis opted to focus on the pressure from the societal metabolism to identify the drivers and also to make a response.

To understand the drivers of the pressures exerted by the societal metabolism, Papers I-II responded to research question R1 utilizing industrial ecology modeling approaches and tools in complex systems to explore and describe physical resource management in production and consumption systems. Paper I took a soft systems approach to comprehend the complexity in physical resource management, and indicated the drivers through a qualitative analysis. In addition, the study developed a framework to model and understand complexity. To model complex systems, the thesis recommended element flow analysis. Paper II undertook a dynamic element flow analysis and delved into a specific product system.

Paper I and Papers III-V responded to research question R2 utilizing industrial ecology modeling approaches and tools for accounting and monitoring the performance of production and consumption systems to improve physical resource management in production and consumption systems. Paper III focused on how mainstream state monitoring can be utilized to identify drivers through confirming pressures. After conducting the study in Paper III, we came to realize the importance of accounting and monitoring the pressure as a sensor to make a response. Paper IV and V developed models targeting two major drivers in human systems, households and buildings, for accounting pressure to monitor, get feedback, and create a common platform among stakeholders.

19 Table 2 shows the contributions of Papers I-V to the research questions, listing the cases and modeling approaches. Papers I-II responded primarily to questions R1 and partly to R2. Papers III-V responded solely to question R2. Case selections and modeling approaches presented in the papers are discussed in the following sections.

Table 2: Contributions of Papers I-V to the research questions R1 and R2 through cases and modeling approaches

3.2 Selection of environmental issues

Papers I and II studied physical resource flows in a conceptual production and consumption system. In addition, Paper I exemplified a case of mobile phone product systems. Paper II deepened the analysis with quantitative and dynamic modeling considering resource flows (Cu, Au, Ag, Pd) in global mobile phone product systems. Papers I and II were the seminal work, followed by a study (in our research group) by Laurenti et al. (2015a,b), on product systems of physical consumer goods. The global mobile phone was chosen among the numerous physical consumer goods for in-depth investigation due to the complexity in production, consumption, and end-of-life.

20 Important reasons for selecting the mobile phone system as a case included:

• The use of mobile phones has increased exponentially during the past 20 years and users frequently replace phones. In addition, due to rapid technological improve- ments and innovations, the life span of a mobile phone is becoming ever shorter (Basel Convention, 2008; Kuehr, 2012; Tischner, 2012; Herat and Agamuthu, 2012). At the same time, interest in using second hand (or/and refurbished) phones has emerged. Similar activities have been perceived in both industrialized and devel- oping countries (Paper II). In addition, there is a great interest in e-wastes for informal recycling in developing countries. As a result, industrialized countries export e-wastes to developing countries. The presence of emerging economies in mobile phone product systems make the end-of-life very complex; at the same time, it makes the whole mobile phone system more challenging in terms of achieving sustainability (Paper II).

• In industrialized countries in particular, active phones predominantly end up either in hibernation (e.g., unused in a drawer) or in the waste stream (Basel Convention, 2008; Bollinger, 2010; Kuehr, 2012; Herat and Agamuthu, 2012). In addition, the mobile phone waste stream is rapidly increasing in these countries (Bollinger, 2010; Panambunan-Ferse and Breiter, 2013; Umair et al., 2013). These waste streams sometimes end through a cheap route in developing countries, e.g., Ghana or Pakistan (Umair et al., 2013); a practice that is sometimes illegal, cf., BBC (2011). In developing countries, these e-wastes are often managed by informal recycling, for instance by open burning or manual dismantling with bare hands (Panambunan-Ferse and Breiter, 2013; Umair et al., 2013).

• Mobile phones are manufactured with more than 40 elements from the periodic table (Kuehr, 2012; Schluep et al., 2009). Thus, mobile phones as an e-waste contain a substantial amount of valuable resources (e.g., gold, silver, palladium), which can be recovered (Basel Convention, 2008; Geyer and Blass, 2010; Tischner, 2012; Wang et al., 2013). Among these elements, on average 23% are metals (Schluep et al., 2009), mostly copper (Cu) and a small amount of precious metals. Mobile phone circuit boards contain a higher economic value of metals than other types of circuit boards in electronic products (Williams et al., 2013). Paper I examined the recovery potential from mobile phones relative to global production of virgin materials in 2012. Around 16% of the silver (Ag) and 8% of the gold (Au) produced world-wide in 2012 were used in mobile phones (Paper I).

Papers III-V employed cases for accounting and monitoring environmental perfor- mance. Paper III focused on mainstream state monitoring to identify drivers. The City of Stockholm frequently monitors and reports sediment metal content in the lakes around Stockholm (Stockholm Stad, 2014). In most of these lakes, an elevated level of

21 copper has been observed (Sinha, 2009; Cui, 2010). In addition, S¨ormeand Lagerkvist (2002) found a significant amount of copper in the wastewater treatment system in Stockholm. Thus Paper III focused on diffuse sources of copper emissions, which is the pressure from human systems. The pressure was then linked to state monitoring, i.e., the fate of pollutant in the lake, to trace back the drivers by monitoring the sediment of a lake.

The case selection and the research approach in Paper III proved to be a reactive response to a problem. Monitoring state is certainly a strong sensor to make a response to drivers of the pressure. It can even indicate problems earlier than impact monitoring approach. However, state monitoring is still a reactive approach. This led us to consider formulating a case using a pressure-based approach to directly provide feedback to the drivers. Thus we selected two major drivers in human systems, households and buildings, to conduct research on a pressure-based, driver-oriented approach to account, monitor, and follow up environmental performance.

Most consumption activities take place within the household economy and therefore the household (rather than the individual consumer) determines a large proportion of resource consumption (McKenzie-Mohr, 1999; Moll et al., 2005; Jones and Kammen, 2011). Furthermore, around 70% of total energy use (both direct and indirect) in a country can generally be attributed to household metabolism. This formed the basis for selecting the household as a case for modeling resource flows with a pressure-based, driver oriented approach (Paper IV). In this regard, energy use is the major anthro- pocentric contributor to greenhouse gas (GHG) emissions. Since consumer awareness of pro-environmental behavior is increasing, feedback to consumers on household metabolism is necessary to achieve an environmentally benign outcome. By providing feedback on a pro-environment direction, household finances, which are one of the most important factors in consumers mind when making decisions, can be linked to environmental pressures through household metabolism. Consumers in our modern age generally get feedback on how much money is left at the end of the month in their bank account. Thus we modeled household metabolism linking to the household economy, with the aim of increasing awareness of a more pro-environment direction.

The case of buildings was selected because globally, the building sector accounts for around 35% of GHG emissions, 40% of total energy use, 12% of land use, 25% of water use, and 25% of solid waste generation (Bribi´anet al., 2009; UNEP, 2009a,b). The building sector therefore has great scope to reduce environmental pressures. Due to growing environmental awareness, environmental pressure of the built environment is an important issue for municipalities, developers, and construction companies. The City of Stockholm originally developed a complex tool, environmental load profile (ELP), to monitor the built environment in the city (Forsberg, 2003; Brick, 2008). Following the failure of this highly complex tool to shed light on the large-

22 scale built environment, a simplified version was redeveloped focusing on materials, with the aim of educating stakeholders about the design phase of building construction.

23 24 4 Modeling

4.1 Modeling in Paper I Plan-do-check-act (PDCA) is a widely used method for continuous improvement in management. However, a rudimentary limitation of modern management philosophy is a lack of consideration of broader systems, which leads to unintended consequences of an improvement action or neglect of known drawbacks that have no economic consequences. With the target of integrating relevant systems to analyze and manage a challenge towards sustainable development, a framework for planning was devised. Figure 4 shows the planning framework for integrating environmental, economic and social systems associated with the given issue for identifying drivers of a problem to manage physical resources in a complex system. First, the framework helps to understand and frame the problem by asking What is the challenge and Why is it a challenge? Second, it encompasses critical thinking on environmental, economic, and social aspects, and internalizes negative externalities to develop a conceptual model. Third, it encourages expansion of the system boundaries from the conceptual model illustrating stocks, flows, drivers, and their inter-linkages. This step involves applying systems thinking and life cycle thinking to identify relevant system expansions. It also involves analyzing causes and effects with a causal loop diagram (CLD) to realize and visualize the reinforcing and balancing feedbacks (Laurenti et al., 2015a). Fourth, it involves making the system operational (i.e., workable) by shrinking the system boundary. Fifth, it encourages setting of goals and indicators to assess the defined problems. Finally, the framework assists in identifying the drivers, in order to devise management strategies to take care of the problem.

The first iteration following the six steps in the framework can be conducted by a qualitative screening. In the second iteration, quantitative modeling, simulation and accounting can be carried out to enhance the management strategies to tackle the challenge. In Paper I, the framework was applied qualitatively to a mobile phone product system to identify a management strategy to close the material flow loop, with the focus on resource depletion and efficient use of resources. The product system was further analyzed by quantitative modeling and simulation in Paper II.

4.2 Modeling in Paper II Application of the framework proposed in Paper I to complex systems like the global mobile phone product system encountered a challenge to modeling the system. It was also difficult to identify the indicators for taking decisions on physical resource management and to monitor the performance of business strategies. Considering the modeling of complex systems and the identification of indicators to take action and

25 Figure 4: Planning framework to integrate relevant important systems for identifying strategies in physical resource management. A first iteration of the steps can be conducted by qualitative screening and a second iteration can be carried out by quantitative modeling, simulation and accounting. monitor the performance, Paper II applied element flow analysis to model the global mobile phone product system. A system dynamics modeling approach was employed to investigate the possibilities for closing the material flow loop by identifying potential drivers in the global mobile phone product system. The elements examined in the study were copper, gold, silver, and palladium (Pd). In the modeling, the element flows in the global mobile phone product system were based on the behavior of the gold flow and gold recovery rate, because in terms of economic value of materials in an end-of-life (EoL) phone, gold represents 80% palladium 10% and silver 7% (Navazo et al., 2013). In addition, gold recovery from e-waste is the main driving force for recycling due to its high economic value (Basel Convention, 2008; Tischner, 2012).

Two indicators were employed for examining the global mobile phone product system: loop leakage and loop efficiency. The loop leakage indicator determines the pressure, i.e., to what extent the elements leave the product system. The loop efficiency indicator indicates how efficiently the resources are utilized in the system. A sensitivity analysis was performed to identify the drivers for closing the element flow loop in the mobile phone product system. The sensitivity was tested by increasing (or decreasing) one parameter by up to 100%, while at the same time keeping other parameters at their default values. Based on the drivers, an optimization was conducted with the help of OptQuest optimizer (AnyLogic, 2014; OptTek Systems, 2015) to devise an eco-cycle scenario focusing on minimizing loop leakage and maximizing loop efficiency. The system dynamic (SD) model was implemented in AnyLogic 6 (AnyLogic, 2014; OptTek Systems, 2015).

26 The overarching model of the system was divided into two main domains having similar stocks and flows: industrialized countries (IC) and developing countries (DC). Since IC and DC consider similar subsystems, Figure 5 only presents a generalized view of the subsystems considered in the modeling. In the diagram, the mobile phone flows subsystem is shown with generalized stocks and flows of mobile phones (see Paper II for detailed stocks, flows, and causally linked parameters). Other subsystems (consumer demands, metal flows, economics investments) are linked to the mobile phone flows subsystem and process the necessary information for the subsystem.

Figure 5: Schematic diagram of subsystems considered in both industrialized countries and developing countries to portray the global mobile phone product system (modified from Paper II). The two-headed straight arrows represent the information flows between the subsystems.

In the system dynamics model, consumer demand is the driving force for production

27 and retail stocks. In the mobile phone flows subsystem, manufacturers sell new phones to consumers through retailers. Consumers use phones for a certain period of time. Then the consumers put the phones into storage without any use (i.e., hibernation time of phones), or pass them on to their friends or family members (Reuse 1 in Figure 5), or sell them to a second-hand customer or a retailer (Reuse 2), who might sell new, used, or refurbished phones. In the systems dynamics model, Reuse 1 and Reuse 2 are implemented through the refurbished phone stock, where collected phones are tested, refurbished, and sold to refurbished (i.e., used/old) phone customers.

In the system dynamics model, the collection process is defined by the accessibility of collection pathways, consumer awareness, and incentives to the consumers. In addition, the stock of collection is the decision-making point for recycle/discard or sell to potential refurbished phone retailers within the domain, or import/export to the other domain based on the demand and economic benefits from the EoL phones. Economics is assumed to be the driving force for phone flows in the global product system from manufacturer to recycler. Furthermore, investments are based on the economic profits to each sector, e.g., production (manufacture of new phones), collection, refurbishment. Furthermore, sales growth of new and refurbished phones depend on total consumer growth. In addition, shifting of consumer demand from new phones to refurbished phones depends on the utility (i.e., functionality compared with a new phone) and the price of refurbished phones.

4.3 Modeling in Paper III In Paper III, a source-transport-storage model was developed to identify the drivers of diffuse copper emissions through sediment monitoring. Figure 6 shows this source-transport-storage model, which was created by coupling a source model and a fate model in a lake. The fate model was designed considering the internal dynamics in a lake, using dynamic process-based element flow analysis (EFA). The fate model was then coupled to the source model using static element flow analysis. The system under study was defined as a lake and the sources of emissions included atmospheric deposition within the catchment area of the lake. The source accounting was conducted in an Excel-based model. The fate model was implemented in Similie to perform the dynamic analysis.

The source-transport-storage model was applied to five lakes in the Stockholm region. The geographical system boundary was the lakes and their catchment area. The source model covered the copper emissions from the urban metabolism within the catchment areas of the lakes. Copper is a widely used element in the catchment areas of the lakes, e.g., copper roofs, water pipes, and brake linings of vehicles. The pathways were defined by the weathering/erosion of copper from the products or

28 Figure 6: Conceptual source-transport-storage model (modified from Paper III) to identify the drivers of copper pollution through monitoring a lake. The arrows represent the copper flows, and the arrow heads show the direction of the flow. EFA = element flow analysis. goods, and transported to the lakes. The main pathway of the emitted copper to the lakes was through urban stormwater. The fate model was taken and modified from the mass-balance model for heavy metals in a lake by Lindstr¨omand H˚akanson (2001). The lake was divided into three parts: water column, sediment in the erosion transport area and sediment in the bottom or active area. Copper in the lake system ends up in the sediment through different complex processes in the aquatic system (e.g., internal loading). In the modeling, six processes were considered: inflow, outflow, sedimentation, resuspension, diffusion, and burial (Paper III).

The source and the fate models were compared with a second type of modeling approach to test the models (Sinha, 2009). Furthermore, the model outcomes were validated with the monitoring field data. Finally, the model outcomes in a time series were compared with the monitored copper concentrations in sediments and water columns in the lakes, to test if state monitoring could identify the drivers of the diffuse copper emissions.

4.4 Modeling in Paper IV The overarching concept of the household metabolism was to link economic activities in a household to both direct and indirect environmental pressures, in order to provide feedback to consumers for making an environmentally benign decision. In the overarch- ing conceptual model, products and services play the central role. Life cycle assessment (LCA) of a specific product can give feedback on environmental pressures/impacts.

29 However, it cannot tell the story of the overall household metabolism. On the other hand, environmentally wide input-output analysis (IOA) can address the overall house- hold metabolism by linking the household economy to the environmental pressures. Thus we designed an overarching hybrid model based on bottom-up (e.g., life cycle inventory based accounting) and top-down (e.g., IOA) approaches as shown in Figure 7.

Figure 7: The overarching conceptual model of household metabolism (taken and modified from Paper IV). EcoRunner is an internet-based tool based on the overarching conceptual model.

Based on the conceptual model, we developed an internet-based feedback tool, EcoRunner , aimed at increasing consumer awareness and focusing on Swedish households. The tool runs two models on the internet: EcoRunner Total and EcoRunner Purchase. EcoRunner Total is based on a top-down model using an environmentally wide IOT (cf. Figure 7) and tailoring the goods item/service (Figure 7) to the classification of individual consumption by purpose (COICOP) developed by the United Nations, which allows users to recognize all their expenditures relating to the overall household metabolism. Thus it allows users to explore their purchase decisions in terms of environmental pressure. The IOT was exported from SCB (2015) and processed the data for EcoRunnerTotal. The tool also offers the users a set of compensatory measures to offset the environmental pressures by their purchase decision and to invest money in different projects (Paper IV). The environmental pressures were estimated in terms of energy use (giga-joules, GJ), greenhouse gas emissions (carbon dioxide equivalents, CO2eqv), and total nitrogen (in NOxeqv).

30 EcoRunner Purchase, on the other hand, is based on bottom-up modeling focusing on the life cycle inventories of specific products from specific countries. It allows the users to compare different types of products from different countries in terms of environmental pressures. It provides feedback on specific purchase decisions, e.g., the environmental pressures of purchasing rice, chicken, beef, etc. from different countries. Environmental pressures are accounted based on life cycle inventories and expressed in terms of environmental footprints (see section 4.5 for an explanation of environmental footprint).

4.5 Modeling in Paper V The Environmental Load Profile (ELP) developed by the City of Stockholm is an Excel-based tool to account for environmental pressure from the built environment. The tool has been facing implementation problems. To eradicate the implementation barrier, the City of Stockholm made a simplification of the tool, to focus on building materials. It was then necessary to check the credibility and consistency of the tool in facilitating the implementation phase and effecting a transition towards a more environmentally benign building construction. It was also important to illustrate the significance of the simplification of the ELP (i.e., material selection). Thus, Paper V made a comparative study of the ELP6 and two commercially available, world-leading softwares, SimaPro and GaBi. To illustrate the effect of the simplification of the ELP, two types of buildings were selected: one concrete-frame building and one wooden-frame building. Figure 8 illustrates the structure of the simplified ELP in Microsoft Excel.

Figure 8: Presentation of the simplified ELP of building structures in Microsoft Excel (taken from Paper V). The rectangles are a schematic illustration of Excel pages in the tool, and the numbers (1-3) indicate the page order.

In our research group, we have come to realize that there is an urgent need to collect information for accounting and reporting practices in terms of environmental footprint, in order to make LCA studies more operational (Frostell, 2013). Environmental footprint is an indicator representing the pressures from the human system. Several

6In this thesis, ELP represents the simplified ELP

31 methodological approaches have been discussed, e.g., ecological footprint (Wackernagel and Rees, 1998), carbon footprint (Finkbeiner, 2009), and water footprint (Hoekstra et al., 2009; Aldaya et al., 2012). However, the scientific stringency of these approaches is debatable and sometimes doubtful. Recently, the European Commission published a Product Environmental Footprint to harmonize methodology for calculation of the environmental footprint of products, under the initiative of a single market for green products (EU, 2013; Finkbeiner, 2014). To be consistent with the EU semantic, Paper V used environmental footprinting to estimate indicators, but instead of the product environmental footprint, the focus was on individual footprints, e.g., carbon footprint, energy footprint, water footprint, and others. In addition, the methodological approach according to the principles for life cycle inventories (LCI) of material and energy flows and stocks was followed (ISO 14044, 2006). Figure 9 presents the framework for environmental footprinting. This thesis studied environmental footprints focusing on pressures and examined the importance of element flow account- ing (EFA) using the pressure-based, driver-oriented response, as discussed in Chapter 5.

Figure 9: Framework for accounting environmental footprints (adopted and modified from ISO 14044 (2006))

Paper V estimated environmental footprints for a concrete building and a wooden building in a life cycle perspective and compared the footprints obtained with ELP, GaBi 6, and SimaPro 8. A cradle-to-gate approach was used in the accounting, i.e., material production in a life cycle perspective and transport to the construction gate. The modeling in SimaPro and GaBi was performed based on the identical processes/inventories in the ELP or the processes that were closest to the Swedish case

32 in the available databases. EcoInvent 2.2 and GaBi 6 professional databases were used in GaBi 6. The EcoInvent 2.2, ETH-ESU 96, and US LCI 2013 databases were used in SimaPro8. The study focused on energy footprint (kWh) and carbon footprint (kg CO2eqv) when making comparisons.

33 34 5 Results and discussion

5.1 Results of Papers I-V 5.1.1 Paper I In the past, efforts within physical resource management have primarily aimed at improving energy and material efficiency and promoting sustainability. However, the efficiency improvements were often offset by increased consumption (i.e., rebound effects). Laurenti et al. (2015a) and Paper I demonstrated how an improvement action can lead to unintended environmental consequences, e.g., consumption rebound effects, increased waste, pollution, and negative externalities. Together with our previous research (Sinha, 2014; Laurenti et al., 2015a,b), Paper I confirmed that the activities in the human system are interlinked, and one problem can trigger others through cause-effect chains. Physical resource management should therefore take a broader systems approach prior to implementing improvement actions. The proposed framework in Paper I can integrate broader systems concerning three principles of sustainability (environment, economics, society), and the framework offers a workable system by shrinking the system boundary to include only the major aspects of the defined challenge.

In the mobile phone product system, material disposal from the product system was perceived as a pressure due to the EoL phone containing a large number of precious elements (e.g., gold) as well as hazardous elements (e.g., cadmium, Cd). The dumping of phones and a low rate of material recovery were identified as a challenge to managing the system. To cope with this challenge, material flows need to be understood in the context of socio-economic drivers. Conventional environmental analysis (e.g., LCA, eco-design) includes only the material flows. In contrast, the proposed framework involves including not only the material flows, but also (i) the socio-economic drivers that can influence the stocks and flows, and (ii) cause-effect relations. With this con- ceptual and holistic understanding of the system, the framework helped to shrink the boundary to a manageable size by prioritizing (i) the collection system, (ii) hibernation of mobile phones, (iii) extraction costs, and (iv) recycling costs. Based on the indicators (material extraction, material disposal from the product system, and efficient use of resources), the framework suggested the following management strategies: (i) design phones for longer use, (ii) design for recycling and improving collection infrastruc- ture, (iii) design for limiting phone hibernation time, and (iv) internalize external costs.

The framework offered three unique characteristics to physical resource management in the planning stage of a product, a service, or an activity. First, the essence of the framework was to expand the problem to a broader system, in order to give a holistic picture of the problem. It then allowed the system boundaries to be shrunk to a workable size. Second, it required explicit accounting of causal relations and

35 feedback loops. Third, the framework identified responsibilities between stakeholders, e.g., policy makers, manufacturers, retailers, consumers, collectors, and recyclers. Application of the framework in policy analysis and policy making and in industry can enhance understanding of a holistic picture, internal feedbacks, and relations between stakeholders, which can help identify drivers of unintended consequences in physical resource management.

5.1.2 Paper II Following the qualitative study in Paper I, a quantitative study was conducted in Paper II to identify ways of closing the element flow loops in the global mobile phone product system. Element flow analysis was adopted to model this complex system. Paper II focused on elements that are economically recoverable from EoL mobile phones. The model was developed primarily based on gold flows in the global mobile phone product system, and it was assumed that silver, copper and palladium followed the flow behavior of gold in the model.

The sensitivity analysis in Paper II to identify drivers to close the element flow loops efficiently in the global mobile phone product system found the following potential drivers: (i) accessibility of collection pathways in industrialized (IC) and developing countries (DC), (ii) mobile phone use time in IC and DC, (iii) gold recovery in DC, and (iv) mobile phone hibernation in both IC and DC. Surprisingly, the costs of metal recovery in the analysis, which was envisaged as a potential driver in Paper I, made a negligible contribution to both loop leakage and loop efficiency. According to the model, a reason could be that formal recycling is not profitable due to the much bigger costs for collection compared to the costs of metal recovery and the benefits from the recovered metals. In addition, the collection infrastructure for phones is not sufficiently well developed to collect EoL phones. At the same time, consumers are not aware or not motivated to give their phone to recyclers. In the model, phone collection was driven by the profit perspective for both consumers and collectors. As a result, the model showed that 63% of phones were kept in hibernation (e.g., in drawers) due to the inaccessibility of the collection system. Ultimately, there were not many phones in the recycling facilities. Thus, the costs of recycling did not play a big role in the model. In addition to the recycling costs, other parameters (e.g., utility of a refurbished phone, manufacturing costs, export costs) made a negligible contribution to loop leakage and the loop efficiency.

The global mobile phone product system in the modeling analysis showed that the business as usual (BAU) scenario was unsustainable. For 2015, the model showed loop leakage of around 50% and loop efficiency of around 30%. The model also showed that loop leakage was continuously increasing, whereas loop efficiency was decreasing.

36 With the help of the OptQuest optimizer and based on the potential drivers, Paper II proposed an eco-cycle scenario that was able to significantly reduce 7 from the product system to the environment. The eco-cycle scenario secured 0% disposal by consumers in the model outcome and decreased the overall loop leakage to 5% at steady state. At the same time, it showed increased resource efficiency, to 95%, at steady state. It also showed a very low demand for raw materials for producing new phones. Figure 10 shows a comparison between the BAU and the eco-cycle scenarios exemplifying gold use by phone manufacturers and gold recovery at EoL as a function of time during 1994-2050. According to the diagram, the eco-cycle scenario requires three times less gold than BAU and also recovers more than three times as much gold. Interestingly, a very low amount of virgin gold is required if the recovered gold is used in manufacturing phones (see the thick and thin black solid lines in Figure 10).

Figure 10: Differences in gold use by phone manufacturers and gold recovery at end-of-life (EoL) between the business as usual (BAU) scenario and the eco-cycle scenario as a function of time, 1994- 2050. In the eco-cycle scenario, the model was run with the BAU parameter settings during 1994- 2013. The model was then paused and the eco-cycle scenario was introduced in 2014 by altering the input parameter settings to the eco-cycle. Thus, in the eco-cycle scenario, the period 1994-2013 shows the BAU and the period 2014-2050 the eco-cycle scenario. The social and political settings for accomplishing the dynamic transition from the BAU to the eco-cycle were not considered. Therefore, the diagram shows a rapid dynamic transition from the BAU to the eco-cycle.

37 The quantitative analysis in Paper II suggested to extend the life-span of mobile phones, improve EoL collection and recycling. The study also suggested that modular design of phones, product service system based extended producer responsibility, internalizing external costs, and improving informal recycling could lead the BAU to the eco-cycle model. Improved political strategy to support product service systems could significantly enhance the upcoming circular economy.

5.1.3 Paper III Papers I-II basically responded to research question R1 by modeling and analyzing complex systems to identify the drivers of a problem in order to make a proactive response. For accounting and monitoring the performance of management actions posed by research question R2, Paper III focused on how mainstream state monitoring can be utilized to identify drivers through confirming pressures.

Paper III developed the source-transport-storage model to evaluate whether sediment copper content can be a tracer to identify the drivers in the human system. The model was applied in five lakes in the Stockholm region. In all five cases, the source analysis showed that traffic, especially vehicle brake-linings, was the dominant source of copper emissions in the catchment area of the lakes. Copper roofs were also found to be an important source of copper emissions in the catchment areas. Figure 11a shows the copper loading from the catchment area to Lake R˚acksta Tr¨ask.

Figure 11: a) Copper load from the catchment area to Lake R˚acksta Tr¨askand b). copper concentration in the sediment of Lake R˚acksta Tr¨askas function of time, where the solid line shows the model simulated sediment copper concentrations in Lake R˚acksta Tr¨ask.The squares represent the monitored copper concentrations in sediment of Lake R˚acksta Tr¨ask(Sinha, 2009). The dashed lines represent the maximum (black) and the minimum (grey) model simulation results when the parametric uncertainty was considered.

The coupling between the source and the fate model was tested with the field

38 monitored concentrations in the water and the sediment. In the model testing for copper concentration in the water column, the model results for all five lakes showed higher concentrations than the field monitoring. The discrepancy can be attributed to the timing of collection of field samples, e.g., the samples were collected when the copper had settled down to the sediments. However, model simulation was not used to verify this hypothesis. In the model testing for sediment copper concentrations, the simulation results were in agreement with the field data for the three most heavily polluted lakes (R˚acksta Tr¨askin Figure 11b, Trekanten, and L˚angsj¨on). However, the model prediction for the remaining two moderately polluted lakes (Judarn and Laduviken) showed higher sediment concentrations than the field monitoring. This overprediction for the moderately polluted lakes was ascribed to the distance of the lake from urban activities and stormwater management in the lake catchments. Paper III concluded that the proposed model was applicable for predicting the upstream sources of emissions by monitoring the sediments of heavily copper-polluted lakes.

Paper III also demonstrated to policy makers how the model could be used in pollution abatement by exemplifying the case of Lake Trekanten. According to the environmental quality standards and based on the lake sediment analysis, the model recommended a 70% decrease in copper emissions by urban activity. The model also recommended a decrease in traffic activity and avoidance of copper roofs in order for the lake sediment to meet the environmental quality standards.

5.1.4 Paper IV After conducting the study in Paper III, we came to realize the importance of accounting and monitoring pressures as a sensor to make proactive responses. We also realized that modeling should be fit for purpose (Frostell, 2013; Keirstead, 2014). Thus, Paper IV developed models aimed at educating the absolute driver in the human systems − consumers. During the work in Paper III, we came to understand that most consumption activity occurred in households. Therefore, households (rather than individual consumers) determined the major proportion of resource consumption. Thus, Paper IV developed an environmental feedback tool, EcoRunner, based on the overarching hybrid model (cf. Figure 7), aimed at increasing consumer awareness of pro-environmental decisions. In addition, the tool was designed with an easily updated database and a way of expanding the model.

Paper IV examined EcoRunner Total by analyzing the expenditure of an average single Swedish household. In addition, the study explored options for the reduction of environmental pressures. Figure 12 shows the expenditures of an average single Swedish household and the associated energy footprint, carbon footprint, and nitrogen footprint. According to Figure 12, the most important energy- and emissions-intensive

39 spendings in a household is on food and non-alcoholic beverages, fuel and maintenance and air travel. The analysis found that fuel contributed more than 90% of both total energy use and total emissions in the category fuel and maintenance (Paper IV).

EcoRunner Purchase allows the user to compare different products from different countries in terms of energy footprint and carbon footprint. Figure 13 shows energy footprint and carbon footprint of some selected food items from EcoRunner Purchase. According to Figure 13, meat and fish have larger environmental footprints, mostly in the production phase, compared with other products. In Figure 13, tomatoes produced in Sweden and Holland show a significant energy footprint in both cases, but a lower carbon footprint compared with South American countries due to less carbon-intense energy in Sweden and Holland.

EcoRunner can address unintended consequences of purchase decisions. For example, the consumer can make response to rebound effects with the help of EcoRunner. It is assumed that consumers generally tend to maintain constant expenditure on con- sumption (Takase et al., 2005). Consumers can sometimes save money by purchasing cheaper products due to energy/material efficiency. However, they may spend the money saved on purchasing products associated with a large environmental footprint. Consequently, consumption rebound occurs (Hertwich, 2005). Concerning the rebound effect, EcoRunner allows consumers to spend the saved money on purchasing prod- ucts/services bearing a lower environmental footprint (Paper IV).

Since consumer awareness regarding environmental sustainability is increasing, EcoRunner provides consumers with a unique package to track their detailed ex- penditure records with environmental footprinting. With EcoRunner, consumers could develop insights into their consumption, which could lead to a change in their future consumption to an environmentally benign consumption pattern. Swedish young people generally consume without bothering to track their expenditure. EcoRunner offers them a unique opportunity to start tracking their expenditure, with environmental feedback. Thus, EcoRunner is suggested for use in the educa- tion system, to provide environmental feedback on consumption to pupils and students.

5.1.5 Paper V Paper V studied models based on building construction, for accounting environmental pressure to monitor, get feedback, and create a common platform among stakeholders. Following the implementation failure of a highly complex tool on the built environ- ment, ELP was simplified to building level, focusing on construction materials. The simplification was designed to make the tool easy to use and educate stakeholders in the design phase of a building. In order to achieve credibility and acceptance of the

40 Figure 12: a) Household expenditures of an average single Swedish household and b) associated energy footprint, c) carbon footprint, and d) nitrogen footprint. Modified from Paper IV. Keys: i = mainte- nance; ii = gases, liquid & solid fuels; iii = tobaco & alcoholic beverages; iv = footwear & clothing; v = household goods & furniture; vi = health; vii = telecommunication & post; viii = miscellaneous goods & services; ix = road & railway; x = air travel; and xi = sea & inland travel.

41 Figure 13: a) Energy footprint and b) carbon footprint of some selected products according to EcoRun- ner Purchase. The footprints represent 1 kg of a product from the cradle to the entry port of Gothen- burg, Sweden. tool by stakeholders, Paper V evaluated the simplified ELP7 by comparing it with two commercially leading two LCA softwares, SimaPro and GaBi. In addition, the study compared two reference buildings with similar total floor area (BTA): (i) a concrete and (ii) a wooden building, to explore the significance of the choice of construction materials in the design stage of a building.

Table 3 shows the energy footprint and carbon footprint obtained with ELP, GaBi, and SimaPro for the concrete building and the wooden building. For the concrete building, SimaPro and GaBi showed an energy footprint that was closer to the ELP value, whereas the carbon footprint was around 50% greater than with ELP. The higher carbon footprint in SimaPro can be explained by the emissions intensity per unit energy use in Sweden compared with the average of the world or even the EU.

7Simplified ELP is synonymous with ELP in Paper V and in the present thesis. All analyses in Paper V were based on simplified ELP.

42 Table 3: Energy footprint and carbon footprint obtained with ELP, GaBi, and SimaPro for a concrete building and a wooden building (taken from Paper V)

Environmental Concrete building Wooden building footprints Total /m2 Total /m2 BTA† BTA† ELP 3 700 0.820 2 300 0.47 Energy SimaPro 4 000 0.89 2 500 0.50 (MWh) GaBi 4 100 0.91 4 800 1 ELP 720 000 160 200 000 40 Carbon SimaPro 1 000 000 220 330 000 67 kg CO2eqv GaBi 1 100 000 240 380 000 76 Energy SimaPro 7% 7% 7% 7% (% varies) GaBi 10% 10% 109%‡ 109%‡ Carbon SimaPro 41% 41% 68% 68% (% varies) GaBi 57% 57% 92% 92% †BTA = total floor area. ‡Erroneous conversion in GaBi (cf. Paper V).

For the wooden building, SimaPro was similar in comparison with ELP as described for the concrete building, but GaBi gave a much larger footprint than SimaPro due to an erroneous conversion for glulam (Paper V). Since energy footprint in SimaPro and GaBi was close to that in ELP, Paper V concluded that the ELP database provided carbon accounting closer to the Swedish case than those two commercially available databases.

According the ELP results in Table 3, the concrete building showed around 43% larger energy footprint and 75% larger carbon footprint than the wooden building. This confirms claims made in many previous studies that wood-based construction materials reduce overall carbon and energy footprint and are sometimes more cost-effective (B¨orjessonand Gustavsson, 2000; Sathre and Gustavsson, 2009; N¨ass´enet al., 2012). In addition, wood has been gaining attention for modern constructions due to its light weight and mechanically strong, renewable material, ability to absorb humidity and release it in dry conditions, ability to absorb unwanted noise, good thermal conductivity, warmth to the touch, and so on (Pajchrowski et al., 2014).

Paper V pointed out important differences between GaBi and SimaPro. In other studies, the results obtained with GaBi and SimaPro for an identical process using data taken from an identical database also show significant differences (Herrmann and Moltesen, 2015). Herrmann and Moltesen (2015) found that the differences were sometimes so large that the conclusions could be doubted. Within the software and database discussion, ELP provides a very easy-to-use tool with consistent results,

43 although the tool has limitations. The City of Stockholm intends to use it to educate/engage designers to include environmental footprint in the design phase of buildings. Paper V concluded that ELP could be used for standard calculations of the environmental footprint of building structures after minor updating, and that it could create a common platform for discussions among stakeholders in the design phase of building construction.

5.2 Synthesis and key findings Table 4 summarizes the contributions of Paper I-V to the key findings, focusing on the approaches to respond to problems (i.e., the pressure-based, driver-oriented approach) and the research questions (R1 and R2). Although Papers III-V were primarily designed to account and monitor environmental performance (R2) (Table 2), they also sought to understand the complex systems involved (R1). The static accounting of systems provides the basis for dynamic modeling to explore complex systems. The contributions of Paper I-V are further synthesized, analyzed, and discussed in the following sections.

5.2.1 Pressure-based, driver-oriented approach The mainstream approach (research and policy decisions) generally attempts to solve a problem by identifying drivers of the problem and taking necessary measures to remove them (Kristensen, 2004; Song and Frostell, 2012; Song, 2012; Sinha, 2014; Zhou et al., 2015). The European Environmental Agency (EEA) uses the DPSIR framework to organize information, establishing a causal relation between the drivers, the pressures, the state and the impacts perceived as a problem. Thus, the DPSIR framework allows policy makers to make necessary responses not only to the drivers but also to the other DPSIR phases. This mainstream approach is currently the main way used to manage environmental problems. However, this approach is a reactive response, since society takes necessary measures when the impacts are perceived. This approach does not allow potential unknown impacts to be investigated and only encourages a response to the drivers of a known/identified impact.

The mainstream approach identifies problems in the impacts on the environment or the human system. However, the impacts are not actually the problem, but rather symptoms of the problem, while the main problem lies inside the drivers in the human system (Meadows et al., 1972, 2004; Meadows and Wright, 2008). The present thesis, therefore, advocated solving these problems by connecting the drivers in the human systems to the closer/immediate basis of the problems, i.e., the pressure according to the DPSIR framework. Monitoring the pressure can even identify potential unknown

44 Table 4: Key findings from Papers I-V in respect to the pressure-based, driver-oriented approach and the research questions.

impacts. For example, understanding the metabolism of elements (of the periodic table) in human systems can identify the pressures from the activities in terms of flows of elements from/to the environment. These flows could reveal unknown potential impacts on the environment and the human system. The significance of element flow analysis in a pressure-based approach is further discussed in the next section, where the pressure-based approach is viewed as a proactive approach, and the pressure is taken as a basis to identify the drivers. In addition, the pressure is the indicator to monitor drivers.

Many might argue that life cycle assessment (LCA) is a proactive approach for decision makers and the basis of the assessment is on the impact categories. For example, conducting an LCA of a product at the design stage to predict impacts beforehand and taking necessary measures to reduce the impacts at the design stage is a proactive approach to decision making and management strategies. The present thesis absolutely acknowledges that LCA takes a comprehensive approach to environmental problems and may therefore be regarded as proactive. However, the impact assessment in LCA makes it sensitive to time delays in the pressure-state-impact chain from

45 societal activity to environmental impact. It would therefore be valuable to further improve the proactive approach in life cycle thinking.

When the uncertainty analysis was conducted in Paper III and in Sinha (2009), it was realized that the uncertainty in identifying the drivers increased in the downstream causal chain. For example in Paper III, the accounting of diffuse copper emissions in the catchment areas (i.e., pressure) was less uncertain in identifying the drivers than the model outcomes in the lakes (i.e., state). Based on the analysis, it was predicted that the uncertainty propagated in the impact following an increasing trend. Figure 14 shows the increasing uncertainty in accounting pressure, state, and impact posed by the drivers in human systems. This propagation of uncertainty raised questions about the basis of the assessment. To lower the uncertainty in the assessment and in decision making, the present thesis advocates the pressure-based approach to identify the drivers and take necessary measures.

Figure 14: Uncertainty in accounting to identify drivers propagating from pressures to the impacts following an increasing trend. The legends: D = drivers, P = pressures, S = state of the environment, I = impacts, and R = responses. The thicker grey lines represent the uncertainty in accounting due to error propagation (propagation of uncertainty). The vertical thin grey lines represent the range of uncertainty, and the horizontal thin grey line represents the axis where there is no uncertainty. The question marks and the dotted lined arrows indicate the basis for making a response to the drivers. The solid arrow represents the response to the drivers. Environmental footprints involve pressure-based assessment, and life cycle assessments involve impact-based assessment.

Furthermore, the thesis suggests assessment of the pressure instead of the impact. The environmental footprint framework illustrated in Figure 9 in Paper V was therefore based on pressure assessment. In addition, the framework followed the methodological approach of life cycle inventory (LCI) analysis according to ISO 14044

46 (2006). The environmental footprint framework can assess pressure in a life cycle perspective. In addition, it can provide the foundation to expand life cycle pressure assessment to include/foresee a potential unknown impact. Thus, the present the- sis attempted to move the focus from the impact to the pressure in life cycle assessment.

Element flow analysis In physical resource management, material flow analysis (MFA) is one of the main tools used for communication and decision support for policy response within physical resource management. Material flow analysis examines and optimizes existing systems by the mass balance and identifying the hotspots. However, MFA, sometimes, is very aggregated to make actual mass balance, for example national input-output tables. On the other hand, substance flow analysis (SFA) is based on the same principle as MFA and deals with the actual mass balance of substances. In general, MFA focuses on resource extraction, production, and consumption for physical resource management, and SFA is used in accounting of pollutants from the source of emissions to a fate (e.g., copper fate in a lake) or a point of interest (e.g., cadmium in fish). Somewhat similarly in LCA, the emissions of pollutants are connected to impact categories in a life cycle perspective. Therefore, SFA and LCA are heavily applied in a pollution perspective. In many cases, they mainly tell the story regarding the sources of emissions of the pollutants to the downstream system. Upstream metabolism of the pollutants in a resource perspective is not often revealed until it comes to scarcity or criticality (Graedel et al., 2015). This thesis suggests element flow analysis (EFA) for both pollution control and resource management (for more information, see Appendix). Figure 15 shows the applicability of EFA in relation to MFA, SFA, and LCA.

As discussed above, MFA generally deals with the aggregated material flows in a resource management perspective, while SFA and LCA often account the emissions to link drivers and potential impacts. To understand societal metabolism compre- hensively, MFA needs to be disaggregated. In addition, the flows in SFA and EFA need to be linked comprehensively to the socioeconomic drivers, e.g., interactions between drivers in the human system should be explored dynamically to understand the pressures (emissions and resource extractions). In this regard, elements can be considered a higher resolution of information on the disaggregation and mass balance of materials or goods or physical resources, complementing MFA and SFA. In addition, elements generally do not transform/destroy through systems which could allow unlimited systems to be integrated. Thus, EFA can permit modeling of more complex systems.

Elements in the periodic table are now extensively used in modern products, for example smart phones contain more than 40 elements. We know very little about the metabolism of the periodic table in human systems. For improved management of the periodic table, it is highly important to understand the metabolism of all elements in

47 Figure 15: Applicability of element flow analysis (EFA) in relation to material flow analysis (MFA), substance flow analysis (SFA) and life cycle assessment (LCA) of element flows. human systems. To understand the societal metabolism of the periodic table, element flow analysis is a key tool. In addition, it can reveal some unknown potential impacts in the future. It can also complement MFA (in terms of disaggregation and proper mass balance), and SFA/LCA (by connecting the elements to the policy interests). To sum up, MFA, SFA and LCA mainly focus on analysis of perceived societal problems, while EFA can reveal potential problems which have not yet been perceived. Thus, EFA can foster a pressure-based, driver-oriented response complementing LCA to develop a pressure assessment that can foresee potential unknown impacts. Therefore, EFA can be a proactive tool in both resource management and pollution control.

5.2.2 Understanding the system to be managed: static vs dynamic Systems are inter-related, and improving the performance of one system can shift the burden to other systems, causing unintended consequences to the environment (Meadows and Wright, 2008; Liang et al., 2012; Laurenti et al., 2015a). Understanding complex systems is a vital step in tackling unintended consequences, as well as managing physical resources in the human system (Liang et al., 2012; Yang et al., 2012; Liang et al., 2013; Sinha, 2014; Laurenti et al., 2015a). The proposed framework

48 in Paper I can help users to grasp broader systems, in order to tackle unintended consequences. When broader systems are difficult to comprehend, the framework proposed shrinks the system to make it manageable. System expansion and shrinking (Paper I) to capture influential parameters is extremely important for uncovering complexity in systems. With the help of the framework, qualitative analysis can provide strategies to manage complex systems. The causal loop diagram (CLD) modeling approach is a key tool to analyze complex systems qualitatively. However, quantification or quantitative analysis is essential to establish management strategies. For example, recycling costs were considered as a potential driver to close the material flow loops in the global mobile phone product system in the qualitative analysis in Paper I. However, the quantitative system dynamic modeling in Paper II found a negligible contribution of the recycling costs. The dynamic quantitative modeling also provided a deeper understanding of the global mobile phone product system. Most importantly, it encouraged use of systems thinking to identify the drivers and thus to improve the closed loop system. Therefore, dynamic modeling is an essential approach to understanding and exploring complex systems.

Table 5 compares the strengths and weaknesses of different systems modeling approaches to understand complex systems in physical resource management. A plus sign (+) indicates a strength and a minus sign (-) a weakness. The characteristics of complex systems are characteristics taken from the list discussed in Chapter 2 that were quantitatively implemented in Papers II-V.

The systems dynamics modeling and simulation was conducted with time series analysis (dynamism) in Paper II. In the modeling, inter-dependencies between different subsystems were established and analyzed dynamically (cf., Figure 5). Non-linearities were also addressed in the modeling by implementing both macro aspects (e.g., feed- back loops) and micro aspects (e.g., if business is profitable then invest). Hibernation of mobile phones in the system dynamics model addressed the delay characteristic of complex systems. Continuous feedback dynamics were implemented between consumption and production systems in the model.

In Paper III, the process-based modeling and simulation was conducted to model lake dynamics to predict the fate of copper in Swedish lakes using time series analysis. Other criteria of complex systems in Table 5 were not explicitly addressed in the modeling (i.e., the model can only partly explain other criteria, e.g., inter-dependency, non-linearity, and delay). The implementation of process-based modeling was quite similar to systems dynamics modeling. Although it was possible to implement other characteristics of complex systems (cf. Table 5) in that model, the intention of the modeling was to predict the lake dynamics and validate the model with the lake (i.e., state) monitoring data. Thus the modeling approach is capable of addressing other criteria, but the intention of the modeling approach restricts their implementation.

49 Table 5: Strengths (+) and weaknesses (-) of the systems modeling approaches applied in this thesis to address characteristics of complex systems in the context of physical resource management.

Process-based modeling is generally used to model ecological systems and understand the dynamics of these systems and how they function (Raftery et al., 1995; Buck-Sorlin, 2013). This approach is also used in predicting futures and illustrating if-then scenarios (Cuddington et al., 2013; Wong et al., 2015). In contrast, a system dynamics modeling approach is used particularly in studying inter-dependencies, feedback dynamics, non-linearities, and delays, which is more appropriate in order to understand societal metabolism and policy support (Meadows et al., 1972; Sterman, 2000; Meadows et al., 2004; Meadows and Wright, 2008).

Substance flow analysis (Paper III), environmental footprintings (Papers IV and V), and input-output analysis (Paper IV) did not address dynamism, dynamic inter-dependencies, non-linearities, delays, and feedback dynamics. However, these tools showed strengths in mapping physical resources, environmental management, accounting, and monitoring. Most importantly, they were easy to model and re- quired less data than process-based modeling and system dynamics modeling. These static approaches can create multiple scenarios to assess, but the dynamic modeling approaches provide more resolution in terms of dynamism, inter-dependencies and non-linearities. The dynamic approaches can account physical resources, but these

50 approaches are not suitable for monitoring. These dynamic tools are appropriate to study complexity to support physical resource management. On the other hand, the static tools are suitable for monitoring past activities.

Moreover, the static modeling approaches are fundamental tools for studying resource flows and metabolism in societal systems. In addition, these tools are appropriate to map stocks and flows and to monitor them. The dynamic version of these approaches is available in the literature. In fact, Papers II and III conducted dynamic substance flow analyses. In this thesis, these approaches were considered to be hybrid modeling or multi-method modeling. Through the systems modeling studies, it emerged that static approaches are appropriate to map, account, and monitor physical resource metabolism in production and consumption systems. Furthermore, dynamic modeling approaches on top of static modeling provides an understanding of the systemic complexity of the metabolism when testing policy instruments, as shown in Figure 16. The diagram also shows that the static modeling approaches are the fundamental tools to understand systems, and these tools provide the basis for dynamic modeling. In addition, the pyramid in the diagram presents the steps to policy implications and getting feedback for systems understanding and refining the accounting models.

Figure 16: Assessed purpose of modeling approaches.

Figure 17 presents different quantitative modeling approaches in relation to the understanding of complex systems. First, mapping the stocks and flows is fundamental to analysis of complexity and system understanding. Static accounting plays a key role in mapping the stocks and flows in the preliminary discussion on the understating of complex systems. In a quantitative analysis, material flow analysis and attribution LCA can portray the mapping of stocks and flows. Approaching broader systems, element flow modeling allows comprehensive systems to be integrated in order to capture complexity. The drivers, linked with the stocks and flows, are then important to assess, in order to understand the complexity. In this regard, consequential LCA

51 could shed more light on complex system understanding and on additional drivers than attributional LCA (Thomassen et al., 2008; Zamagni et al., 2012; Ekvall et al., 2016).

Figure 17: Different quantitative modeling approaches in terms of static and dynamic modeling. The diamond shapes represents the logic of the modeling approach or the purpose of the modeling. The arrows show the directions to select a modeling approach. The grey scale gradient in the lower bar shows the degree of complex systems understanding from low (light grey) to high (black), achieved in the modeling approaches.

To introduce more complexity and inter-connections between the systems, the drivers are generally linked to the stocks and flows. When the causal links and the processes are well understood, input-output analysis can help identify drivers by assessing the linkage between the flows and the drivers. In a complex system, a leverage point (Meadows, 1999) can be achieved by investigating how changes in one sector affect the other sectors. Computable general equilibrium (CGE) modeling (Dixon and Jorgenson, 2012) has the potential to establish inter-relations between drivers, stocks, and flows to optimize the system towards achieving a certain goal (West, 1995; Bhattacharyya, 1996; Hertel et al., 2009; Dixon and Jorgenson, 2012).

When processes and links between the pressure and the drivers are well known; however, predictions/forecasts with time series are important and a dynamic process based (PB) approach is suitable to apply. If the project wants to apply systems thinking to achieve a deeper systems understanding, system dynamics (SD) and agent-based modeling (ABM) are appropriate approaches to choose. System dynamics is an approach to understanding the behavior of complex systems over time. It deals with internal feedback loops and systems evolution. However, the SD approach is not suitable for modeling social interactions and individual behavior or network analysis. In this case, ABM provides the advantage of individual interactions for social learning

52 (Gilbert, 2008).

Increasing inter-relations between systems makes modeling and analysis critical to developing insights into complex real world problems. Ideally, the system should be modeled in a way that allows the investigation of multiple future scenarios, in order to formulate plans that enable flexibility in future decision options. The modeling approach should have the ability to examine the system as a whole. Therefore, the modeling framework should be formulated in a way that allows changing sub-system modules to share information with the different stakeholders involved in the complex problem.

System dynamics, agent-based modeling or process-based modeling, individually, can create a dynamic modeling environment. However, the individual modeling approach (e.g., SD) is also restricted in several ways. For example, one modeling technique may be restricted in one way, while another modeling approach may have the flexibility to overcome the restriction. Therefore, coupling multiple modeling approaches (i.e. combining SD, ABM, or PB) by taking advantage of different techniques regarding the problem would be a way to overcome these restrictions and to create a flexible and dynamic modeling environment. The present thesis clearly shows that a flexible dynamic modeling environment can provide the platform to explore complex systems and enhance system understanding. Thus, a multi-method modeling approach could increase understanding of complex systems and help explore and improve physical resource management in production and consumption systems.

5.2.3 Environmental accounting to monitor the performance Life cycle assessment provides a comprehensive accounting approach to monitoring environmental performance. However, the assessment in mainstream LCA is based on impact categories (Tillman and Baumann, 2004). Based on the above discussion on the pressure-based, driver-oriented approach, however, the message to the LCA community should be to shift the focus to assessment of pressures instead of impacts.

According to the work in this thesis, a proactive monitoring system should have indicators based on pressure. In addition, pressure-based indicators should be based on both stocks and flows. Paper II showed how pressures can be linked to manage stocks, cf., loop efficiency. For flow management, the pressure-based indicators can be either the element flows themselves (e.g., carbon footprint), or closely related to drivers in human systems. In addition, indicators with strong scientific causality with the impact categories, CO2eqv, should be monitored. Finally, element flow accounting would provide a higher resolution of disaggregation in accounting, and it can be linked to policy interests and established impact categories in, e.g., LCA. For example,

53 carbon could be connected to the CO2, CH4, and CO2eqv; however, the main purpose of the accounting would be for the management of carbon in the human systems. Therefore, element flow accounting can offer a proactive approach, complementing the mainstream interests in physical resource management.

Our research conducted a content analysis to study the DPSIR approach (i.e., pressure-, state-, or impact -based) applied in in urban planning (Zhou et al., 2015). The results surprisingly showed that most planning reports from all over the world focus on a pressure-based approach. In addition, a pressure-based, driver-oriented approach receives some attention in the scientific community. For example, carbon footprint is a widely accepted LCA-based tool in the scientific community. In addition, many scientific scholars are working on the connection between pressures and drivers, e.g., Liang et al. (2016). However, practical applications lack the pressure-based approach. For example, in the study by Zhou et al. (2015), US cities showed very high scores in a pressure-based approach in planning, but showed lower performance in application than European cities. The study indicated that the pressure-based approach and LCA-based studies receive good attention in the scientific community and in the theoretical discussion. However, they are lacking in practical applications.

If LCA-based knowledge is not applied in the real case, there is no point in discussing it at the theoretical level. During my PhD research, I came to realize that accounting and monitoring models should be specific to the purpose. Thus, Paper IV conducted household metabolism analysis designed for the Swedish consumer and Paper V targeted educating stakeholders involved at the design stage of building construction. Thus, this thesis devised a pressure-based, driver-oriented approach to design tailor-made tools to monitor, get feedback, and create a common platform for discussing and improving physical resource management.

5.3 Further research This thesis demonstrated the strength of dynamic modeling approaches for exploring complex systems and indicated how these could help formulate policy instruments. A further comparative study between static and dynamic analysis of policy instru- ments could provide insights on adopting dynamic analysis before implementing policies. To analyze policy instruments, a holistic approach including hard and soft aspects is important to consider, and a flexible dynamic modeling environment (e.g., multi-method modeling as discussed above) is required to make a better analysis. Further research could provide instructions on how to make links between different tools, taking the advantages of each. A harmonized systems modeling approach connecting all stakeholders (and their interests) with a shared vision for fostering a circular economy in production and consumption systems would be interesting to study.

54 Application of element flow analysis (EFA) was demonstrated qualitatively in the present thesis, and partially quantitatively in Paper II. In future work, EFA should be developed further to establish a connection to life cycle pressure assessment. In addition, EFA should be studied in real cases to establish connections between different tools and policy interests for supporting pressure-based monitoring. Development of a pressure-based monitoring platform connecting all stakeholders in production and consumption systems is also recommended.

55 56 6 Conclusions

This thesis identified a need to identify the drivers causing environmental pressures and shifting burdens from one system to another. To address drivers in production and consumption systems, a pressure-based, driver-oriented approach was viewed as a proactive response to environmental management compared with the mainstream impact/state-based approach. This proactive approach was employed in this thesis to experiment with different systems modeling tools for understanding complex systems and environmental accounting to support monitoring system. The overarching contri- bution of the thesis was to synthesize the advantages of different systems modeling approaches to support improved physical resource management in production and con- sumption systems. The responses to research questions R1 and R2 are presented below.

R1. What systems modeling approaches can foster an understanding of complex systems, in order to explore and improve the physical resource management in production and consumption systems?

To address drivers in complex systems, environmental pressures linking socio- economic drivers were modeled to improve systems understanding and ultimately management strategies. The proposed planning framework in Paper I showed the potential to understand complexity by offering qualitative screening and systems integration, as well as including quantitative modeling to give greater resolution in management strategies. The important contribution of the framework is system expansion and shrinkage to capture the influential parameters for comprehending complexity in systems. The framework also showed how the causal loop diagram technique can be utilized to assess complex systems qualitatively. In addition, Paper II demonstrated the strength of system dynamics modeling in understanding complex systems. For dynamic quantitative modeling, the significance of element flow analysis in integrating comprehensive systems for assessing complexity was illustrated.

Based on the assessment of the quantitative systems modeling to understand complex systems, the thesis concluded that dynamic modeling approaches (e.g., systems dynamics and process-based) had the advantage over the static mod- eling approaches (e.g., material flow analysis, environmental footprinting, and input-output analysis). In the systems modeling exercises, the dynamic modeling approaches gave higher resolution in systems understanding in order to address drivers. However, static modeling or accounting provided a fundamental mapping of stocks and flows, which was the basis for dynamic modeling. In addition, static modeling proved suitable for monitoring environmental performance. These findings suggest that using a dynamic modeling approach on top of static modeling

57 can provide a better understanding of the systemic complexity of the metabolism, facilitating work to devise and test policy instruments. A multi-method modeling approach was recommended by taking advantage of different modeling approaches to create a flexible dynamic modeling environment to foster understanding of complex systems, in order to explore and improve physical resource management in production and consumption systems. Further research could provide instructions on how to make links between different tools, taking the advantages of each.

R2. What systems modeling approaches can foster environmental accounting to monitor the environmental performance of production and consumption systems for proactive physical resource management?

Three systems modeling approaches were demonstrated at different system scales to address drivers and to foster accounting for environmental monitoring of production and consumption systems for proactive physical resource management. Conventional accounting focuses either on accounting sources (e.g., diffuse sources of emission), or fate, e.g., in a lake. Paper III bridged the gap by coupling the sources of pressure and fate in the state of the environment to respond to drivers by monitoring state for improved management. The results indicated that the uncertainty in accounting to identify drivers propagated from the pressure assessment to the impact assessment in an increasing trend. It was therefore concluded that a pressure-based monitoring system should be used to address drivers. In addition, Papers IV and V illustrated how households and business sectors can monitor their activities to take better decisions. These studies thereby contributed to supporting pressure-based monitoring, by demonstrating systems modeling of households and business scales to educate the drivers (consumers). Future studies should seek to develop a pressure-based monitoring platform connecting all stakeholders in production and consumption systems.

The thesis confirmed the importance of life cycle-based accounting in monitoring environmental performance. To achieve proactive monitoring, pressure-based assessment was proposed by arguing the significance of the approach compared with impact-based LCA. Thus, the message to the LCA community was to shift the focus of their assessment to pressures instead of impacts.

To monitor environmental performance, accounting indicators based on both stocks and flows are recommended. Finally, the systems modeling should be fit for purpose, in order to promote life cycle assessment and thinking in practice.

In conclusion, a pressure-based, driver-oriented approach to systems modeling can

58 be recommended for proactive physical resource management in production and con- sumption systems, in order to understand complexity and monitor environmental per- formance.

59 60 References

Achachlouei, M. A., 2015. Exploring the effects of it on environmental assessment: from life cycle assessment o complex systems modeling. Ph.D. thesis, Department of Sustainable development, Environmental science and Engineering (SEED), KTH Royal Institute of Technology, Stockholm, Sweden.

Aldaya, M. M., Chapagain, A. K., Hoekstra, A. Y., Mekonnen, M. M., 2012. The water footprint assessment manual: Setting the global standard. Routledge.

Anderson, P., 1999. Perspective: Complexity theory and organization science. Organi- zation science 10 (3), 216–232.

AnyLogic, 2014. Multimethod simulation software. Available at: http://www. anylogic.com, [accessed March 21, 2014].

Astier, M., Garc´ıa-Barrios,L., Galv´an-Miyoshi, Y., Gonz´alez-Esquivel, C. E., Masera, O. R., 2012. Assessing the sustainability of small farmer natural resource management systems. a critical analysis of the mesmis program. Ecology and society 17 (3), 25.

Ayres, R. U., 1989. Industrial metabolism. Technology and environment 1989, 23–49.

Ayres, R. U., Ayres, L., 2002. A handbook of industrial ecology. Edward Elgar Pub- lishing.

Baccini, P., Brunner, P. H., 1991. Metabolism of the Anthroposphere. Springer-Verlag.

Baccini, P., Brunner, P. H., 2012. Metabolism of the anthroposphere: analysis, evalua- tion, design. MIT Press.

Balestrini Robinson, S., 2009. A modeling process to understand complex system ar- chitectures. Ph.D. thesis, School of Aerospace Engineering.

Bar-Yam, Y., 1997. Dynamics of complex systems. Vol. 213. Addison-Wesley Reading, MA.

Basel Convention, 2008. Guidance document on the environmentally sound manage- ment of used and end-of-life mobile phones. Prepared by Mobile Phone Working Group. [Online]. Available at: http://archive.basel.int/industry/mppi/MPPI% 20Guidance%20Document.pdf, [accessed August 07, 2013].

Baynes, T. M., 2009. Complexity in urban development and management. Journal of Industrial Ecology 13 (2), 214–227.

BBC, 2011. Track my trash. [Online]. Available at: http://www.bbc.co.uk/ programmes/b0116gw0, [accessed July 25, 2013].

61 Bhattacharyya, S. C., 1996. Applied general equilibrium models for energy studies: a survey. Energy Economics 18 (3), 145–164.

Binder, C. R., 2007a. From material flow analysis to material flow management part i: social sciences modeling approaches coupled to mfa. Journal of Cleaner Production 15 (17), 1596–1604.

Binder, C. R., 2007b. From material flow analysis to material flow management part ii: the role of structural agent analysis. Journal of Cleaner Production 15 (17), 1605– 1617.

Boccara, N., 2010. Modeling complex systems. Springer Science and Business Media.

Bollinger, L., 2010. Growing cradle-to-cradle metal flow systems: An application of agent-based modeling and system dynamics to the study of global flows of metals in mobile phones. Delft University of Technology, Delft, Netherlands.

Bollinger, L. A., Davis, C., Nikoli´c,I., Dijkema, G. P., 2012. Modeling metal flow systems. Journal of Industrial Ecology 16 (2), 176–190.

Boons, F., Howard-Grenville, J. A., 2009. The social embeddedness of industrial ecology. Edward Elgar Publishing.

B¨orjeson,L., H¨ojer, M., Dreborg, K.-H., Ekvall, T., Finnveden, G., 2006. Scenario types and techniques: towards a user’s guide. Futures 38 (7), 723–739.

B¨orjesson,P., Gustavsson, L., 2000. Greenhouse gas balances in building construction: wood versus concrete from life-cycle and forest land-use perspectives. Energy policy 28 (9), 575–588.

Bossel, H., 2007a. System Zoo 3 Simulation Models: Economy, Society, Development. Vol. 3. BoD–Books on Demand.

Bossel, H., 2007b. Systems and models: complexity, dynamics, evolution, sustainability. BoD–Books on Demand.

Brand˜ao,M., 2012. Food, feed, fuel, timber or carbon sink?: Towards sustainable land- use systems: A consequential life cycle approach. Ph.D. thesis, University of Surrey.

Bribi´an,I. Z., Us´on,A. A., Scarpellini, S., 2009. Life cycle assessment in buildings: State-of-the-art and simplified lca methodology as a complement for building certifi- cation. Building and Environment 44 (12), 2510–2520.

Brick, K., 2008. Barriers for implementation of the Environmental Load Profile and other LCA-based tools. Licentiate Thesis at Industrial Ecology. KTH Royal Institute of Technology.

62 Bringezu, S., 2002. Towards sustainable resource management in the european union. Tech. rep., Wuppertal papers.

Brunner, P. H., 2004. Materials flow analysis and the ultimate sink. Journal of Industrial Ecology 8 (3), 4–7.

Buck-Sorlin, G., 2013. Process-based model. In: Encyclopedia of Systems Biology. Springer, pp. 1755–1755.

Burstr¨om,F., 1999. Materials accounting and environmental management in municipal- ities. Journal of Environmental Assessment Policy and Management 1 (03), 297–327.

Cao, M., Woodward, F. I., 1998. Dynamic responses of terrestrial ecosystem carbon cycling to global climate change. Nature 393 (6682), 249–252.

Chappin, E. J., Dijkema, G. P., 2010. Agent-based modelling of energy infrastructure transitions. International journal of critical infrastructures 6 (2), 106–130.

Checkland, P., 1981. Systems thinking, systems practice. John Wiley and Sons, Chich- ester.

Checkland, P., 2000. Soft systems methodology: a thirty year retrospective. Systems Research and Behavioral Science 17, S11–S58.

Churchman, C. W., 1968. The syatems approach. Dell Publishing Co., Inc., New York.

Clift, R., Druckman, A. (Eds.), 2015. Taking Stock of Industrial Ecology. No. ISBN978- 3-319-20571-7. Springer.

Cotsaftis, M., 2009. What makes a system complex?-an approach to self organization and emergence. In: From System Complexity to Emergent Properties. Springer, pp. 49–99.

Cuddington, K., Fortin, M.-J., Gerber, L., Hastings, A., Liebhold, A., O’Connor, M., Ray, C., 2013. Process-based models are required to manage ecological systems in a changing world. Ecosphere 4 (2), 1–12.

Cui, Q., 2010. Tracing copper from society to the aquatic environment: Model devel- opment and case studies in stockholm. Licentiate Thesis at Industrial Ecology.

Cui, Q., Brandt, N., Sinha, R., Malmstr¨om,M. E., 2010. Copper content in lake sedi- ments as a tracer of urban emissions: Evaluation through a source–transport–storage model. Science of the Total Environment 408 (13), 2714–2725.

Daniels, P. L., 2002. Approaches for quantifying the metabolism of physical economies: A comparative survey: Part ii: Review of individual approaches. Journal of Industrial Ecology 6 (1), 65–88.

63 Daniels, P. L., Moore, S., 2001. Approaches for quantifying the metabolism of physical economies: Part i: Methodological overview. Journal of Industrial Ecology 5 (4), 69–93.

Davis, C., Nikolic, I., Dijkema, G. P., 2010. Industrial ecology 2.0. Journal of Industrial Ecology 14 (5), 707–726.

Dijkema, G. P., Basson, L., 2009. Complexity and industrial ecology. Journal of Indus- trial Ecology 13 (2), 157–164.

Dixon, P. B., Jorgenson, D. W., 2012. Handbook of computable general equilibrium modeling. Newnes.

Duchin, F., 2009. Input-output economics and material flows. In: Handbook of input- output economics in industrial ecology. Springer, pp. 23–41.

Duchin, F., et al., 2004. Input-output economics and material flows. Rensselaer Poly- technic Institute, Department of Economics (Rensselaer Working Papers in Eco- nomics 424.

Earles, J. M., Halog, A., 2011. Consequential life cycle assessment: a review. The International Journal of Life Cycle Assessment 16 (5), 445–453.

Eco-Cycle, 2014. Eco-cycle: building zero waste communities. [Online]. Available at: https://www.ecocycle.org/index.php.

Edmonds, B., 1995. What is complexity?-the philosophy of complexity per se with application to some examples in evolution. The evolution of complexity.

Ehrenfeld, J., 2004. Industrial ecology: a new field or only a metaphor? Journal of Cleaner Production 12 (8), 825–831.

Ehrenfeld, J., Hoffman, A., 2013. Flourishing: A Frank Conversation about Sustain- ability. Stanford University Press.

Ekvall, T., Azapagic, A., Finnveden, G., Rydberg, T., Weidema, B. P., Zamagni, A., 2016. Attributional and consequential LCA in the ILCD handbook. The International Journal of Life Cycle Assessment, 1–4. URL http://dx.doi.org/10.1007/s11367-015-1026-0

Ekvall, T., Ciroth, A., Hofstetter, P., Norris, G., 2004. Evaluation of attributional and consequential life cycle assessment. Chalmers University of Technology, G¨oteborg.

EU, 2013. Commission recommendation of 9 april 2013 on the use of common methods to measure and communicate the life cycle environmental performance of products and organisations. Official Journal of the European Union, Volume 56, 4 May 2013.

64 Finkbeiner, M., 2009. Carbon footprinting: opportunities and threats. The Interna- tional Journal of Life Cycle Assessment 14 (2), 91–94.

Finkbeiner, M., 2014. Product environmental footprint: breakthrough or breakdown for policy implementation of life cycle assessment? The International Journal of Life Cycle Assessment 19 (2), 266–271.

Finnveden, G., Moberg, A.,˚ 2005. Environmental systems analysis tools: an overview. Journal of cleaner production 13 (12), 1165–1173.

Fischer-Kowalski, M., Haberl, H., 1998. Sustainable development: socio-economic metabolism and colonization of nature. International Social Science Journal 50 (158), 573–587.

Fischer-Kowalski, M., H¨uttler,W., 1998. Society’s metabolism. Journal of industrial ecology 2 (4), 107–136.

Fischer-Kowalski, M., Krausmann, F., Giljum, S., Lutter, S., Mayer, A., Bringezu, S., Moriguchi, Y., Sch¨utz,H., Schandl, H., Weisz, H., 2011. Methodology and indicators of economy-wide material flow accounting. Journal of Industrial Ecology 15 (6), 855– 876.

Forrester, J. W., 1958. Industrial dynamics: a major breakthrough for decision makers. Harvard business review 36 (4), 37–66.

Forrester, J. W., 1997. Industrial dynamics. Journal of the Operational Research Society 48 (10), 1037–1041.

Forsberg, A., 2003. Environmental assessment of the urban environment development and first application of the environmental load profile for hammarby sj¨ostad.Licenti- ate Thesis. Stockholm: Industrial Ecology, Department of Chemical Engineering and Technology, Royal Institute of Technology.

Frosch, R. A., Gallopoulos, N. E., 1989. Strategies for manufacturing. Scientific Amer- ican 261 (3), 144–152.

Frostell, B., 2013. Life cycle thinking for improved resource management: LCA or? In: Handbook of Sustainable Engineering. Springer, pp. 837–857.

Geels, F. W., 2005. Technological transitions and system innovations: a co-evolutionary and socio-technical analysis. Edward Elgar Publishing.

Geels, F. W., 2012. A socio-technical analysis of low-carbon transitions: introducing the multi-level perspective into transport studies. Journal of Transport Geography 24, 471–482.

65 Geng, Y., Doberstein, B., 2008. Developing the circular economy in china: Challenges and opportunities for achieving’leapfrog development’. The International Journal of Sustainable Development & World Ecology 15 (3), 231–239.

Geyer, R., Blass, V. D., 2010. The economics of cell phone reuse and recycling. The International Journal of Advanced Manufacturing Technology 47 (5-8), 515–525.

Gilbert, G. N., 2008. Agent-based models. No. 153 in 07. Sage.

Giljum, S., Hubacek, K., 2009. Conceptual foundations and applications of physical input-output tables. In: Handbook of input-output economics in industrial ecology. Springer, pp. 61–75.

Goudriaan, J., Van Laar, H., 2012. Modelling potential crop growth processes: textbook with exercises. Vol. 2. Springer Science and Business Media.

Graedel, T., Harper, E., Nassar, N., Reck, B. K., 2015. On the materials basis of modern society. Proceedings of the National Academy of Sciences 112 (20), 6295–6300.

Graedel, T. E., Allenby, B. R., 2003. Industrial Ecology, 2nd Edition. Prentice Hall, Upper Saddle River, New Jersey.

H˚akanson, L., 2004. Internal loading: A new solution to an old problem in aquatic sciences. Lakes & Reservoirs: Research & Management 9 (1), 3–23.

Hanson, D. A., John, A., et al., 2009. Input-output equations embedded within climate and energy policy analysis models. In: Handbook of Input-Output Economics in Industrial Ecology. Springer, pp. 417–432.

Hashimoto, S., Moriguchi, Y., 2004. Proposal of six indicators of material cycles for describing society’s metabolism: from the viewpoint of material flow analysis. Re- sources, Conservation and Recycling 40 (3), 185–200.

Herat, S., Agamuthu, P., 2012. E-waste: a problem or an opportunity? Review of issues, challenges and solutions in Asian countries. Waste Management & Research 30 (11), 1113–1129.

Herrmann, I. T., Moltesen, A., 2015. Does it matter which Life Cycle Assessment (LCA) tool you choose?–a comparative assessment of SimaPro and GaBi. Journal of Cleaner Production 86, 163–169.

Hertel, T. W., Rose, S., Tol, R. S., 2009. 1 land use in computable general equilibrium models. Economic analysis of land use in global climate change policy 14, 1.

Hertwich, E. G., 2005. Consumption and the rebound effect: An industrial ecology perspective. Journal of Industrial Ecology 9 (1-2), 85–98.

66 Hoekstra, A. Y., Chapagain, A. K., Aldaya, M. M., Mekonnen, M. M., 2009. Water foot- print manual: State of the art 2009. Tech. rep., Water Footprint Network Enschede, The Netherlands. ISO 14044, 2006. Environmental management—life cycle assessment— requirements and guidelines. Geneva, Switzerland. Jackson, T., 2011. Prosperity without growth: Economics for a finite planet. Routledge. Jones, C., Kammen, D., 2011. Quantifying carbon footprint reduction opportunities for us households and communities. Environmental Science and Technology-Columbus 45 (9), 4088. Joshi, S., 1999. Product environmental life-cycle assessment using input-output tech- niques. Journal of Industrial Ecology 3 (2-3), 95–120. Kapur, A., Graedel, T. E., 2004. Industrial ecology. Encyclopedia of Energy 3, 373–382. Keirstead, J., 2014. Fit for Purpose? Rethinking Modeling in Industrial Ecology. Jour- nal of Industrial Ecology 18 (2), 161–163. Kelly, R. A., Jakeman, A. J., Barreteau, O., Borsuk, M. E., ElSawah, S., Hamilton, S. H., Henriksen, H. J., Kuikka, S., Maier, H. R., Rizzoli, A. E., et al., 2013. Selecting among five common modelling approaches for integrated environmental assessment and management. Environmental Modelling & Software 47, 159–181. Kleijn, R., Huele, R., Van Der Voet, E., 2000. Dynamic substance flow analysis: the delaying mechanism of stocks, with the case of PVC in Sweden. Ecological Economics 32 (2), 241–254. Kneese, A. V., Ayres, R. U., d’Arge, R. C., 2015. Economics and the environment: a materials balance approach. Routledge. Kristensen, P., 2004. The DPSIR framework. National Environmental Research Insti- tute, Denmark, 10. Kuehr, R., 2012. Global e-waste initiatives. In: Goodship, V., Stevels, A. (Eds.), Waste electrical and electronic equipment (WEEE) handbook. No. 30 in Woodhead Pub- lishing Series in Electronic and Optical Materials. Woodhead Publishing Ltd., Ch. 1, pp. 3–16. Kumar, A., Kim, H., Hancke, G. P., 2013. Environmental monitoring systems: a review. Sensors Journal, IEEE 13 (4), 1329–1339. Kytzia, S., 2009. An input-output framework to enhance consistency in hybrid model- ing. In: Handbook of Input-Output Economics in Industrial Ecology. Springer, pp. 99–121.

67 Laszlo, E., 1996. The Systems View of the World a Holistic Vision for Our Time. No. ISBN 1-57273-053-6 in 96-711. Hampton Press.

Laurenti, R., 2013. Applications of systems thinking within the sustainability domain: Product design, product systems and stakeholder perspectives. Licentiate Thesis at Industrial Ecology.

Laurenti, R., 2016. The karma of products: Exploring the causality of environmental pressure with causal loop diagram and environmental footprint. Ph.D. thesis, Division of Industrial Ecology, KTH Royal Institute of Technology, Stockholm, Sweden.

Laurenti, R., Singh, J., Sinha, R., Potting, J., Frostell, B., 2015a. Unintended envi- ronmental consequences of improvement actions: A qualitative analysis of systems’ structure and behavior. Systems research and behavioral science.

Laurenti, R., Sinha, R., Singh, J., Frostell, B., 2015b. Some pervasive challenges to sustainability by design of electronic products–a conceptual discussion. Journal of Cleaner Production 108, 281–288.

Leontief, W. W., 1986. Input-output economics. Oxford University Press on Demand.

Liang, S., Qu, S., Xu, M., 2016. Betweenness-based method to identify critical trans- mission sectors for supply chain environmental pressure mitigation. Environmental Science & Technology.

Liang, S., Xu, M., Suh, S., Tan, R. R., 2013. Unintended environmental consequences and co-benefits of economic restructuring. Environmental science & technology 47 (22), 12894–12902.

Liang, S., Xu, M., Zhang, T., 2012. Unintended consequences of bioethanol feedstock choice in china. Bioresource technology 125, 312–317.

Lindstr¨om,M., H˚akanson, L., 2001. A model to calculate heavy metal load to lakes dominated by urban runoff and diffuse inflow. Ecological Modelling 137 (1), 1–21.

M˚ansson,N., Bergb¨ack, B., Hjortenkrans, D., Jamtrot, A., S¨orme, L., 2009. Utility of substance stock and flow studies. Journal of Industrial Ecology 13 (5), 674–686.

Matthews, R. B., Gilbert, N. G., Roach, A., Polhill, J. G., Gotts, N. M., 2007. Agent- based land-use models: a review of applications. Landscape Ecology 22 (10), 1447– 1459.

McDonough, W., Braungart, M., 2010. Cradle to cradle: Remaking the way we make things. MacMillan.

68 McGuire, A., Sitch, S., Clein, J., Dargaville, R., Esser, G., Foley, J., Heimann, M., Joos, F., Kaplan, J., Kicklighter, D., et al., 2001. Carbon balance of the terrestrial biosphere in the twentieth century: Analyses of CO2, climate and land use effects with four process-based ecosystem models. Global Biogeochemical Cycles 15 (1), 183–206.

McKenzie-Mohr, D., 1999. Community based social marketing.

Meadows, D., 1999. Leverage points: Places to intervene in a system. The Sustainability Institute, Hartland, VT 05048, USA.

Meadows, D., Meadows, D., Randers, J., 2004. The limits to growth: the 30-year update. Chelsea Green Publishing.

Meadows, D., Wright, D., 2008. Thinking in systems. Chelsea Green Publishing.

Meadows, D. H., Goldsmith, E., Meadow, P., 1972. The limits to growth. Vol. 381. Universe books New York.

Moll, H., Noorman, K., Kok, R., Engstr¨om,R., Throne-Holst, H., Clark, C., 2005. Pur- suing more sustainable consumption by analyzing household metabolism in European countries and cities. Journal of Industrial Ecology 9 (1-2), 259–275.

Mongelli, I., Tassielli, G., Notarnicola, B., 2009. Carbon tax and its short-term effects in italy: an evaluation through the input-output model. In: Handbook of Input-Output Economics in Industrial Ecology. Springer, pp. 357–377.

NASA, 2009. Earth’s energy budget. National Aeronautics and Space Administration. [Online]. Available at: http://science-edu.larc.nasa.gov/energy_budget/pdf/ ERB_Litho_v2014.06.26.pdf.

N¨ass´en,J., Hedenus, F., Karlsson, S., Holmberg, J., 2012. Concrete vs. wood in build- ings: An energy system approach. Building and Environment 51, 361–369.

Navazo, J. M. V., M´endez,G. V., Peir´o,L. T., 2013. Material flow analysis and energy requirements of mobile phone material recovery processes. The International Journal of Life Cycle Assessment, 1–13.

Nelson, R., 1976. Structure of complex systems. In: PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association. Vol. 1976. JSTOR, pp. 523–542.

Ness, B., Anderberg, S., Olsson, L., 2010. Structuring problems in sustainability science: The multi-level dpsir framework. Geoforum 41 (3), 479–488.

Nordhaus, W. D., 2011. Integrated economic and climate modeling. Cowles Foundation for Research in Economics at Yale University.

69 Odum, E. P., 1989. Ecology and our endangered life-support systems. Sinauer Asso- ciates. Odum, E. P., Odum, H., 1959. Fundamentals of ecology, 546 pp. Philadelphia: WB Saunders Company.

OptTek Systems, 2015. Optquest. Available at: http://www.opttek.com/OptQuest, [accessed January 11, 2015]. Pajchrowski, G., Noskowiak, A., Lewandowska, A., Strykowski, W., 2014. Wood as a building material in the light of environmental assessment of full life cycle of four buildings. Construction and Building Materials 52, 428–436. URL http://dx.doi.org/10.1016/j.conbuildmat.2013.11.066 Panambunan-Ferse, M., Breiter, A., 2013. Assessing the side-effects of ICT develop- ment: E-waste production and management: A case study about cell phone end-of- life in Manado, Indonesia. Technology in Society. Peters, G. P., Hertwich, E. G., 2009. The application of multi-regional input-output analysis to industrial ecology. In: Handbook of input-output economics in industrial ecology. Springer, pp. 847–863. Preston, F., 2012. A Global Redesign? Shaping the Circular Economy. Energy, Envi- ronment and Resource Governance. London: Chatham House. Raftery, A. E., Givens, G. H., Zeh, J. E., 1995. Inference from a deterministic population dynamics model for bowhead whales. Journal of the American Statistical Association 90 (430), 402–416. Ranhagen, U., Frostell, B., 2014. Eco-cycle model 2.0. for stockholm royal seaport city district: Feasibility study-final report. Tech. rep., KTH, School of Architecture and the Built Environment (ABE), Stockholm, Sweden. Ravetz, J., 2000. Integrated assessment for sustainability appraisal in cities and regions. Environmental impact assessment review 20 (1), 31–64. Rose, A., 1995. Input-output economics and computable general equilibrium models. Structural change and economic dynamics 6 (3), 295–304. Ruth, M., 2009. Dynamic modeling of industrial ecosystems. Journal of Industrial Ecol- ogy 13 (6), 839–842. Ruth, M., Davidsdottir, B., 2008. Changing stocks, flows and behaviors in industrial ecosystems. Edward Elgar Publishing. Ruth, M., Davidsdottir, B., 2009. The dynamics of regions and networks in industrial ecosystems. Edward Elgar Publishing.

70 Sathre, R., Gustavsson, L., 2009. Using wood products to mitigate climate change: External costs and structural change. Applied Energy 86 (2), 251–257.

SCB, 2015. Statistics sweden. www.scb.se.

Schelling, T. C., 1971. Dynamic models of segregation. Journal of Mathematical Soci- ology 1, 143–186.

Schluep, M., Hagelueken, C., Kuehr, R., Magalini, F., Maurer, C., Meskers, C., et al., 2009. Sustainable innovation and technology transfer industrial sector stud- ies: Recycling–from e-waste to resources. United Nations Environment Programme & United Nations University, Bonn, Germany.

Shui Zhang, J., 2001. Iterative method for finding the balanced growth solution of the non-linear dynamic input–output model and the dynamic CGE model. Economic Modelling 18 (1), 117–132.

Simms, D., Thomas, J., 1982. Integrated monitoring: Principles and practice. Environ- mental monitoring and assessment 1 (4), 405–412.

Singh, J., 2016. Beyond waste management: Challenges to sustainable global physi- cal resource management. Ph.D. thesis, Division of Industrial Ecology, KTH Royal Institute of Technology.

Sinha, R., 2009. Modelling copper sources and fate in Lake R˚acksta Tr¨ask,Stockholm: Sediment copper content as indicator of urban metal emissions. Master of Science Thesis. Industrial Ecology, KTH Royal Institute of Technology, Stockholm, Sweden.

Sinha, R., 2014. Industrial ecology approaches to improve metal management: Three modeling experiments. Licentiate Thesis at Industrial Ecology. KTH Royal Institute of Technology.

Slocombe, D. S., 1992. Environmental monitoring for protected areas: review and prospect. Environmental monitoring and assessment 21 (1), 49–78.

Smeets, E., Weterings, R., 1999. Environmental indicators: Typology and overview. European Environment Agency Copenhagen.

Song, X., 2012. A pressure-oriented approach to water management. Ph.D. thesis, In- dustrial Ecology, KTH Royal Institute of Technology, Stockholm, Sweden.

Song, X., Frostell, B., 2012. The DPSIR framework and a pressure-oriented water quality monitoring approach to ecological river restoration. Water 4 (3), 670–682.

S¨orme,L., Lagerkvist, R., 2002. Sources of heavy metals in urban wastewater in stock- holm. Science of the Total Environment 298 (1), 131–145.

71 Sterman, J., 2000. Business Dynamics: Systems Thinking and Modeling for a Complex World. Irwin-McGraw-Hill.

Stockholm Stad, 2014. Koppar. [Online]. Available at: http://www.miljobarometern. stockholm.se/sub.asp?mo=6&dm=1, [accessed February 22, 2014].

Suh, S., 2009. Handbook of input-output economics in industrial ecology. Vol. 23. Springer Science & Business Media.

Suh, S., Huppes, G., 2002. Missing inventory estimation tool using extended input- output analysis. The International Journal of Life Cycle Assessment 7 (3), 134–140.

Suh, S., Kagawa, S., 2009. Industrial ecology and input-output economics: a brief history. In: Handbook of Input-Output Economics in Industrial Ecology. Springer, pp. 43–58.

Sundberg, M., 2005. Spatial computable general equilibrium modelling: static and dynamic approaches. Licentiate Thesis at KTH, School of Architecture and the Built Environment (ABE), Transport and Economics, Transport and Location Analysis.

Tabara, J. D., Roca, E., Madrid, C., Valkering, P., Wallman, P., Weaver, P., 2008. Integrated sustainability assessment of water systems: lessons from the ebro river basin. International journal of innovation and sustainable development 3 (1-2), 48– 69.

Takase, K., Kondo, Y., Washizu, A., 2005. An analysis of sustainable consumption by the waste input-output model. Journal of Industrial Ecology 9 (1-2), 201–219.

Thomassen, M. A., Dalgaard, R., Heijungs, R., De Boer, I., 2008. Attributional and consequential LCA of milk production. The International Journal of Life Cycle As- sessment 13 (4), 339–349.

Tillman, A., Baumann, H., 2004. The Hitchhikers Guide to LCA. Lund, Sweden: Stu- dentlitteratur AB.

Tischner, U., 2012. Sustainable electronic product design. In: Goodship, V., Stevels, A. (Eds.), Waste electrical and electronic equipment (WEEE) handbook. No. 30 in Woodhead Publishing Series in Electronic and Optical Materials. Woodhead Pub- lishing Ltd., Ch. 19, pp. 404–441.

Tukker, A., Charter, M., Vezzoli, C., Stø, E., Andersen, M. M., 2008. System innovation for sustainability 1. Vol. 1. Greenleaf Publishing.

Umair, S., Bj¨orklund,A., Petersen, E. E., 2013. Social Life Cycle Inventory and Impact Assessment of Informal recycling of Electronic ICT Waste in Pakistan. on Information and Communication Technologies, 52.

72 UNEP, 2009a. Buildings and climate change: summary for decision-makers. United Nations Environment Programme.

UNEP, 2009b. Common carbon metric for measuring energy use and reporting green- house gas emissions from building operations. United Nations Environment Pro- gramme.

Wackernagel, M., Rees, W., 1998. Our ecological footprint: reducing human impact on the earth. No. 9. New Society Publishers.

Wang, F., Huisman, J., Stevels, A., Bald´e,C. P., 2013. Enhancing e-waste estimates: Improving data quality by multivariate input–output analysis. Waste Management.

Weidema, B. P., 2003. Market information in life cycle assessment. Vol. 863. Miljøstyrelsen.

Wenheng, W., Shuwen, N., 2008. Impact study on human activity to the resource- environment based on the consumption level difference of china’s provinces or au- tonomous regions. China population, Resources and environment 18 (4), 121–127.

West, G. R., 1995. Comparison of input–output, input–output+ econometric and com- putable general equilibrium impact models at the regional level. Economic Systems Research 7 (2), 209–227.

Williams, E., Kahhat, R., Bengtsson, M., Hayashi, S., Hotta, Y., Totoki, Y., 2013. Link- ing informal and formal electronics recycling via an interface organization. Challenges 4 (2), 136–153.

Wing, I. S., 2004. Computable general equilibrium models and their use in economy- wide policy analysis. Technical Note, Joint Program on the Science and Policy of Global Change, MIT.

Wolman, A., 1965. The metabolism of cities. Scientific American 213 (3), 179–190.

Wong, C. P., Jiang, B., Kinzig, A. P., Lee, K. N., Ouyang, Z., 2015. Linking ecosystem characteristics to final ecosystem services for public policy. Ecology letters 18 (1), 108–118.

Wrisberg, N., de Haes, H. A. U., Triebswetter, U., Eder, P., Clift, R., 2002. Analyt- ical tools for environmental design and management in a systems perspective: the combined use of analytical tools. Kluwer Dordrecht, The Netherlands.

Xu, M., Allenby, B., Kim, J., Kahhat, R., 2009. A dynamic agent-based analysis for the environmental impacts of conventional and novel book retailing. Environmental science & technology 43 (8), 2851–2857.

73 Yang, Y., Bae, J., Kim, J., Suh, S., 2012. Replacing gasoline with corn ethanol results in significant environmental problem-shifting. Environmental Science & Technology 46 (7), 3671–3678.

Yang, Z., 1999. A coupling algorithm for computing large-scale dynamic computable general equilibrium models. Economic Modelling 16 (3), 455–473.

Zamagni, A., Guin´ee,J., Heijungs, R., Masoni, P., Raggi, A., 2012. Lights and shadows in consequential LCA. The International Journal of Life Cycle Assessment 17 (7), 904–918.

Zhou, G., Singh, J., Wu, J., Sinha, R., Laurenti, R., Frostell, B., 2015. Evaluating low- carbon city initiatives from the dpsir framework perspective. Habitat International 50, 289–299.

74