A Life Cycle Assessment of Fibre Optic Submarine Cable Systems Craig

Home , Apollo (cable system), Atlantic Crossing 1, Telephone call, Telephone network

Twenty thousand leagues under the sea: A life cycle assessment of fibre optic submarine cable systems

Craig Donovan

Stockholm 2009

KTH, Department of Urban Planning and Environment Division of Environmental Strategies Research – fms

Kungliga Tekniska högskolan

Degree Project SoM EX 2009 -40 www.infra.kth.se/fms

Twenty thousand leagues under the sea: A life cycle assessment o f fibre optic submarine cable systems

Abstract

Submarine cables carry the vast majority of transcontinental voice and data traffic. The high capacity and bandwidth of these cables make it possible to transfer large amounts of data around the globe almost instantaneously. Yet, little is known about the potential environmental impacts of a submarine cable from a life cycle perspective. This study applies Life Cycle Assessment (LCA) methodology to collect and analyse the potential environmental impacts of a submarine cable system within a single consistent framework. The system boundary is drawn at the limits of the terminal station where the signal is transferred to, or from, the terrestrial network. All significant components and processes within the system boundary have been modelled to account for the flow of resources, energy, wastes and emissions. Data quality analysis is performed on certain variables to evaluate the effect of data uncertainties, data gaps and methodological choices. The results highlight those activities in the life cycle of a submarine cable that have the largest potential environmental impact; namely, electricity use at the terminal station and cable maintenance by purpose-built ship. For example, the results show that 7 grams of carbon dioxide equivalents (CO 2 eq.) are potentially released for every ten thousand gigabit kilometres (10,000Gb·km), given current estimations of used capacity. The potential environmental impacts are directly linked to capacity and system usage, thus, increasing data traffic improves the environmental performance of the submarine cable system per unit of data. A focus area for further improvements is the emissions from ships, where the greatest gains in environmental performance are likely to be made through reduced emissions. This study is perhaps the first tentative step in linking together research into the environmental impact of terrestrial ICT networks.

Keywords: Life cycle assessment, LCA, submarine cables, fibre optics.

Twenty thousand leagues under the sea: A life cycle assessment o f fibre optic submarine cable systems

Executive Summary

Submarine cables carry over 97 percent of our transcontinental voice and data traffic. The world map of submarine cable networks shows that Europe, North America and Asia are well connected with many cable systems. Yet, little is known about the potential impacts of submarine cable systems from a life cycle perspective. This study applies Life Cycle Assessment (LCA) methodology to collect and analyse the potential environmental impacts of a submarine cable system within a single consistent framework. A “cradle-to-grave” approach is considered, which begins with the extraction of raw materials from the natural environment and ends with the return of wastes back to the environment.

The goal of this study is to undertake an LCA of a fibre optic submarine cable system in order to assess the potential environmental impact of sending data over the cable network. To evaluate these impacts, the modelled flows within the system must be related to a quantifiable function of the system, described as the functional unit . The environmental impacts are described as “potential impacts” as they are not fixed in time and space and are often related to an arbitrary functional unit. In this study the functional unit is given as Ten thousand gigabit kilometres (10,000Gb.km), which is a scalable unit and can be interpreted as, for example, 1.25Gb of data sent over 8,000km of submarine cable. The technological system boundary is defined as the limits of the land terminal station where the signal is received from, or transmitted to, the terrestrial network and includes the submarine cable, submarine repeaters and all significant components within the terminal station. The temporal boundary is based on a commercial service lifetime of 13 years and the geographical boundary based on a generic system in a global perspective.

Detailed data of the flows within, and crossing, the system boundary has been collected during the inventory stage of this study. Key processes are the production of electricity and the production and combustion of marine fuel. A total of 127 gigawatt hours (GWh) of electricity are used, given the lifetime of 13 years, with 90 percent of this being consumed during the use & maintenance phase. Ship operations represent the other key activity requiring a total of 179 ship days per 1000 kilometres of cable, resulting in the combustion of a total of 1515 tonnes of fuel. A total of 54 percent of the fuel is consumed during the use & maintenance phase, with 19 percent consumed during the installation and end-of-life recovery phases. The end-of-life decommissioning scenario considers that the cable is recovered by purpose-built ship and recycled for the mechanical materials, such as plastic, steel and copper. Recycling of these particular mechanical materials is highly efficient and a “closed-loop” recycling process is modelled, which assumes that 90 percent of the virgin material input is offset by the recycled materials.

Impact assessment is undertaken on the modelled flows based on characterisation databases. This process assigns each flow to ten baseline impact categories based on an impact factor in relation to a single indicator, for example, carbon dioxide equivalents (CO 2 eq.) as an indicator of climate change. The ten impact categories used in this study are; abiotic resource depletion potential, acidification potential, ecotoxicity potential to freshwater, seawater and land, global warming potential, photochemical ozone creation potential, ozone depletion potential, eutrophication potential, human toxicity potential.

CO 2 equivalents are equally shared between electricity use at the terminal (47 percent) and maintenance of the cable by purpose-built ship consuming marine fuel (53 percent). However, further analysis shows that iii

Twenty thousand leagues under the sea: A life cycle assessment o f fibre optic submarine cable systems

the impact, per unit of primary energy input, from the combustion of marine fuel oil has a far greater impact on climate change than the impact from electricity use. This reflects the disparity in the environmental impacts of electricity and fossil fuel consumption. Focusing on climate change, the results show that a total of 7 grams of carbon dioxide equivalents (CO 2 eq.) are released for every 10,000Gb.km. This result can be applied to, for example, a telepresence conferencing system. Consider a conference between Stockholm and New York with a distance of 8000km and a bandwidth usage of 18Mbps, then

0.1 grams of CO 2 equivalents would potentially be released every second, which results in a potential release of 355 grams of CO 2 equivalents per hour. By comparison, this equates to only 3 kilometres of air travel for a single person or 2.2 kilometres road travel by the average EU passenger car. Expanding this example, a 2 day meeting could utilise the telepresence system for 16 hours resulting in a potential release of 5.7kg of CO 2 equivalents. By comparison, this same 2 day meeting in a face-to-face setting would require 16,000km of air travel, resulting in a release of 1920kg of CO 2. It should be noted that this example considers only the impact of sending data via the submarine cable system and not the telepresence system as a whole.

The function of the system is based on usage, or the actual used bandwidth, as opposed to the lit capacity, or the present technological limitations (at the terminal) of any system. Research shows that bandwidth usage is approximately 25 percent of current lit capacity. If this gap between usage and lit capacity was reduced, notwithstanding technical and commercial limitations, then a subsequent gain in environmental performance per data unit would be achieved. However, it should be noted that the overall environmental impact over the system lifetime remains unchanged. Similarly, increased system usage, in this case increased total data traffic, reduces the resulting potential environmental impacts per unit of data. The sensitivity analysis (described below) supports this conclusion and shows that increasing system usage over the 25 year technical lifetime of a submarine cable system reduces the potential environmental impact per unit of data. From a life cycle perspective, the longer a cable remains in service, the superior the environmental performance per unit of data. Used capacity and service life therefore have a significant effect on determining the results.

The limitations of the study affect the final result, therefore, as recommended by the ISO 14040 series guidelines, a sensitivity analysis has been undertaken to estimate the effect of data gaps, assumptions and methodological choices. The submarine repeaters and terminal components are two sub-models affected significantly by data gaps and assumptions. However, by changing parameters within these sub-models, the sensitivity analysis shows that they have little effect on the final result. This indicates that the LCA model is relatively unaffected by the greatest uncertainties and is thus, robust. Methodological choices include the use of database models for the production of electricity and heavy fuel oil (HFO) and for the combustion of HFO. The sensitivity analysis shows that methodological choices affect the final result by, on average, approximately 20 percent. It is also important to remember, when interpreting the results, that an LCA model is a simplification of reality.

The results of the normalisation calculation show that the relative environmental burden per capita is relatively small. Assuming that the total annual digital media consumption (1440GB) of the average US citizen is sent via submarine cable, then the normalised result shows that this represents only a fraction of the total annual climate change impact, at 1.2 percent. If the aim is to reduce the environmental impact of cable systems further, then the use & maintenance phase is the area where the greatest gains could be made, particularly, electricity use at the terminal and the emissions from the cable ships. The greatest gain is likely to be achieved with the reduction of ship emissions as these appear to have the most significant impact per unit of primary energy. Service lifetime and used bandwidth are also key parameters. An increase in either results in a corresponding decrease in the potential impact per unit of data. These are particular areas where cable owners could direct their focus. Finally: “Without sub-sea cable systems, global telecommunications at the level we know today would be impossible” (CPNI, 2006, p18). iv

Twenty thousand leagues under the sea: A life cycle assessment o f fibre optic submarine cable systems

Acknowledgements

This study has been completed to fulfil the requirements for a Master of Science at The Royal Institute of Technology (KTH) in Stockholm, Sweden. The study was undertaken at Ericsson Research in Kista, Stockholm and completed in the fall of 2009.

Given the broad range of data collected and analysed, many people hav e contributed to this work. Firstly , I would like to thank Ericsson Research, as a whole, for providing the opportunity and support to undertake my thesis project. I would like to thank my supervisor at Ericsson Research, Fredrik Jonsson, for his guidance and knowledge, particularly with the software program GaBi. I would like to thank my supervisor at KTH, Åsa Moberg for her guidance on LCA methodology and for many ideas and fruitful discussions . I would also like to thank others at Ericsson Research, name ly, Peter Håkansson, who was the catalyst for this study and Jens Malmodin , who has a great depth of LCA knowledge and has given additional support. Further, I would like to thank: Maria L öfgren at Ericsson Network Technologies for organising the site visi t to the cable manufacturing plant; Dean Veverka of Southern Cross Cable s Limited for organising the site visit to the cable terminal station in Auckland, New Zealand and for providing many leads within the cable industry; and others working within the ind ustry including ; Paul Betts, Andrew Louw and Kevin Todd.

Finally, I would like to thank my partner Sofie Dahlgren for her scientific objectivity and support during my studies.

Stockholm, October 2009

Craig Donovan

Twenty thousand leagues under the sea: A life cycle assessment o f fibre optic submarine cable systems

Table of Contents Abstract ...... ii Executive Summary ...... iii Acknowledgements ...... v 1. Introduction ...... 1 1.1. Background ...... 1 1.2. Purpose and Objectives ...... 1 1.3. Problem Area and Research Questions ...... 1 1.4. Delimitations ...... 2 1.5. Report Structure ...... 3 2. Theoretical Framework ...... 4 2.1. Methodology of Life Cycle Assessment (LCA) ...... 4 2.1.1. LCA Phases ...... 4 2.1.2. Limitations and Criticisms of LCA ...... 6 2.2. Submarine Cable Systems ...... 7 2.2.1. Historical Development ...... 7 2.2.2. Modern Systems ...... 8 2.2.3. System Architecture ...... 9 2.2.4. System Components ...... 11 2.2.5. System Design and Installation ...... 13 2.2.6. System Operation and Maintenance ...... 13 2.2.7. End-of-Life Decommissioning ...... 14 2.2.8. Submarine verses Satellite Transmission ...... 15 3. Goal and Scope ...... 16 3.1. Goal ...... 16 3.1.1. Target Audience ...... 16 3.1.2. Applicability of this Study...... 16 3.2. Scope ...... 16 3.2.1. System Description ...... 16 3.2.2. Functional Unit ...... 17 3.2.3. System Boundaries and Delimitations ...... 17 3.2.4. Data Requirements and Data Quality ...... 19 3.2.5. Methods for Inventory Analysis ...... 20 3.2.6. Methods for Impact Assessment ...... 20 3.2.7. Software ...... 21 3.2.8. Study-Wide Assumptions, Simplifications and Limitations ...... 22

Twenty thousand leagues under the sea: A life cycle assessment o f fibre optic submarine cable systems

3.2.9. Critical Review Procedure ...... 24 4. Life Cycle Inventory (LCI) ...... 25 4.1. Description of System ...... 25 4.2. Data Calculation ...... 26 4.3. Data Collection Process ...... 26 4.4. Description of Core Unit Operations and LCI Sub-Models ...... 27 4.4.1. Energy ...... 28 4.4.2. Transportation ...... 29 4.4.3. Raw Material Extraction Phase ...... 32 4.4.4. Design & Manufacturing Phase ...... 37 4.4.5. Installation Phase ...... 41 4.4.6. Use & Maintenance Phase ...... 42 4.4.7. End-of-Life Decommissioning Phase ...... 44 4.5. Allocation ...... 47 4.6. Inventory Results and Discussion ...... 48 4.6.1. Inventory Results ...... 48 4.6.2. LCI Discussion ...... 49 5. Life Cycle Impact Assessment (LCIA) ...... 51 5.1. General Allocation Procedure ...... 51 5.2. Definition of Impact Categories and Characterisation Factors ...... 51 5.2.1. Abiotic Resource Depletion ...... 52 5.2.2. Acidification Potential ...... 52 5.2.3. Ecotoxicity Potential to Freshwater, Land and Seawater ...... 52 5.2.4. Global Warming Potential ...... 52 5.2.5. Photochemical Ozone Creation Potential ...... 52 5.2.6. Ozone Depletion Potential...... 52 5.2.7. Eutrophication Potential ...... 53 5.2.8. Human Toxicity Potential ...... 53 5.3. Classification and Characterisation Summary...... 53 5.4. Definition of Normalisation Factors ...... 53 6. Calculation Procedure...... 55 7. Results of Life Cycle Interpretation ...... 57 7.1. Summary of Results ...... 58 7.2. Application of Results ...... 60 7.3. Results by Environmental Impact Category ...... 61 7.3.1. Energy Resources ...... 61

vii

Twenty thousand leagues under the sea: A life cycle assessment o f fibre optic submarine cable systems

7.3.2. Resource Depletion ...... 63 7.3.3. Acidification ...... 64 7.3.4. Ecosystem Toxicity ...... 65 7.3.5. Climate Change ...... 66 7.3.6. Photochemical Ozone Creation ...... 67 7.3.7. Stratospheric Ozone Depletion ...... 68 7.3.8. Eutrophication...... 69 7.3.9. Human Toxicity ...... 70 7.4. Results by Life Cycle Phases ...... 71 7.4.1. Raw Materials...... 72 7.4.2. Design & Manufacturing ...... 75 7.4.3. Installation ...... 78 7.4.4. Use & Maintenance ...... 80 7.4.5. End-of-Life Decommissioning ...... 84 7.5. Data Quality Analysis ...... 87 7.5.1. Sensitivity Analysis – Data gaps and uncertainties ...... 87 7.5.2. Sensitivity Analysis – Methodological Choices ...... 89 7.6. Normalisation ...... 92 8. Discussion ...... 93 9. Conclusions ...... 96 10. Recommendations and Future Improvements and Use of the Model ...... 97 11. Terminology ...... 98 12. List of Tables ...... 99 13. List of Figures ...... 100 14. References ...... 101 14.1. Public References ...... 101 14.2. Internal Ericsson References ...... 104 15. Appendices ...... 105 15.1. Appendix A – Calculation of the Generic Cable System ...... 105 15.2. Appendix B – Data Sources ...... 106 15.3. Appendix C – Stakeholder Analysis ...... 108 15.4. Appendix D – Detailed System Flowchart of GaBi software sub-model ...... 109 15.5. Appendix E – Questionnaire to Suppliers ...... 110 15.6. Appendix F – Sensitivity Analysis Results ...... 112

viii

1. Introduction A life cycle assessment o f fibre optic submarine cable syst ems

1. Introduction

1.1. Background Submarine cables carry the vast majority of transcontinental voice and data traffic (NEC, 2008). The world map of submarine cable networks (Figure 3) shows that Europe, North America and the Asia are well connected with many cable systems spanning the oceans. The high capacity and bandwidth of these cables makes it possible to transfer large amounts of data around the globe almost instantaneously (Jonsson, 2009a). “Without sub-sea cable systems, global telecommunications at the level we know today would be impossible.” (CPNI, 2006, p.18). Yet, little is known about the potential environmental impacts of a submarine cable from a life cycle perspective (Jonsson, 2009a).

Life Cycle Assessment is a method to model the inputs and outputs of a system from “cradle-to-grave”, in order to identify the potential environmental impacts. A “cradle-to-grave” approach begins with the extraction of raw materials from nature and ends with the return of wastes back to nature as emissions. (ISO 14040:2006). It is a holistic approach that collects and frames the environmental impacts into a single consistent framework (Guinée et al , 2004). Since its early beginnings in the 1960’s, LCA has developed into a systematic and phased methodology with guidelines defined by the International Organisation for Standardisation (ISO 14040:2006).

This LCA study has been completed in conjunction with the EMF Safety and Sustainability division of Ericsson Research, Stockholm, Sweden and undertaken to fulfil the requirements for the degree of Master of Science at the Royal Institute of Technology (KTH), Stockholm, Sweden. Ericsson has been working with life cycle assessment of their products for over 14 years. This study is seen as adding to their research into the total environmental impact of the global network of information and communication technologies (ICTs).

1.2. Purpose and Objectives The purpose of this study is to make a contribution to knowledge in the field of both information and communication technologies (ICTs) and environmental studies by undertaking a Life Cycle Assessment (LCA) on fibre optic submarine cables systems in order to assess their potential environmental impact.

The objectives of this study are to:

• Collect as complete and up-to-date data as possible for the life cycle inventory. • Construct an LCA model of the system using LCA software. • Analyse the system model to establish the potential environmental impacts. • Identify the phase or process in the life cycle of a submarine cable system that has the greatest potential environmental impact.

1.3. Problem Area and Research Questions The general problem field for this study is sustainable information and communication technologies (ICTs) and the environmental impact of the global ICT network.

The specific problem area is the environmental impact of fibre optic submarine cable systems . During preliminary research of the scientific databases and through consultation within the industry, it became apparent that no previous LCA in the area of submarine cables had been undertaken. Therefore, it appears that a

1. Introduction A life cycle assessment o f fibre optic submarine cable syst ems

knowledge gap has been identified and the potential environmental impact of a submarine cable, given a life cycle perspective, is unknown. Based on this knowledge gap, a number of research questions arise;

• What are the potential environmental impacts of submarine cables given an LCA perspective? • Which activities in the life cycle of a submarine cable have the largest potential environmental impact? • How could the potential environmental impact of those activities be reduced, or, in which activity could the greatest reduction be achieved?

1.4. Delimitations Research problems are often complex and inter-related (Viking and Österberg, 2004). Given the time limitation of 20 weeks and the resource constraints of this study, some problems in the study area have not been addressed. Following the recommended structure of an LCA report, the delimitations of this study are discussed in relation to the system boundary presented in Section 3.2.3.

1. Introduction A life cycle assessment o f fibre optic submarine cable syst ems

1.5. Report Structure The format for this report is based on Ericsson’s standard LCA report template which follows the recommendations set out in the ISO guidelines (ISO 14040:2006).

The report structure is presented in Figure 1 and is summarised as follows: firstly the theory of LCA methodology and of submarine cable systems is established; then the goal and scope of the study are presented; next the data collection process is explained in the life cycle inventory (LCI); followed by the life cycle impact assessment (LCIA); where after the calculation procedure and the results are presented; followed by the discussion, conclusion and recommendations; finally the references, list of figures and tables are detailed and the appendices are presented.

2. Theoretical Th e details of the theoretical framework for this study are provided in this section . Framework Firstly the methodology of LCA is explained, secondly, the key elements of a fibre optic submarine cable system are described.

3. Goal and Scope This section describes the goal and scope of the study and details the context within which the assessment of the environmental impacts of the system has been determined.

4. Life Cycle This section presents the system and the process of data collection and calculation. Inventory (LCI) Furthermore, a detailed description is given of the structure of the core sub-models for the life cycle inventory of the studied system.

5. Life Cycle Impact The life cycle impact assessment (LCIA ) is presented in this section , which “aims at Assessment (LCIA) describing the environmental consequences of the environmental loads quantified in the inventory analysis” (Baumann and Tillman, 2004, p.129).

6. Calculation This section describes the calculation procedure for taking the results of the LCIA and Procedure presenting them in relation to the functional unit of the study.

7. Results of Life In this section , the results of the life cycle interpretation a re presented in relation to Cycle Interpretation the functional unit using a variety of bar charts.

8. Discussion A discuss ion of the important issues based on the results of th is study is presented here. The limitations of the study are discussed, as are the findings in relation to usage and system capacity.

9. Conclusion The section concludes the study an d sums up the important issues and results.

10 . Here recommendations are made based on the findings of the study. Future Recommendations improvements of the model are suggested and the use of the model is explained. and future use

11 -14 . Terminology, In th ese sections the terminology used in this report is listed, as are the tables and List of tables and figures used. Finally, the reference listing is presented for both public and internal figures, References Ericsson reports.

This section provides additional supporting material to the main body of the report. 15. Appendices

Figure 1: Report structure

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

2. Theoretical Framework

This section details the theoretical framework for this study. Firstly the methodology of LCA is explained, secondly, the key elements of a fibre optic submarine cable system are described.

2.1. Methodology of Life Cycle Assessment (LCA) Life Cycle Assessment is a method to model the inputs and outputs of a system from “cradle-to-grave” in order to identify the potential environmental impacts. A “cradle-to-grave” approach begins with the extraction of raw materials from nature and ends with the return of wastes as emissions back to nature. (ISO 14040:2006). It is a holistic approach that collects and frames the environmental impacts into a single consistent framework (Guinée et al , 2004). An LCA attempts to map all significant resource and energy inputs with their subsequent product (or service) and waste outputs, then, interpret the calculated potential environmental burdens of the studied system (USEPA, 2006). LCAs can also be undertaken as a cradle-to-gate study, where only a portion of the product or system life cycle is studied (Baumann and Tillman, 2004). Since its early beginnings in the 1960’s, LCA has developed into a systematic and phased methodology with guidelines defined by the International Organisation for Standardisation (ISO 14040:2006).

2.1.1. LCA Phases The ISO guidelines identify four phases of an LCA study, as shown in Figure 2. Interpretation is undertaken throughout the LCA and the double-ended arrows indicate the iterative nature of an LCA study, or, the need to continually assess if the goal and scope are being fulfilled (Baumann and Tillman, 2004; EEA, 1997). The four stages of an LCA are described in the following sections.

Figure 2: LCA stages (ISO 14040:2006, p8)

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

2.1.1.1. Goal and Scope Definition

When beginning an LCA study, clear identification of the goal and scope is important to help determine the methodology and data requirements (Baumann and Tillman, 2004). In formulating the goal, the product, process or activity shall be described and in what context, in other words, why the study is being carried out (USEPA, 2006). The goal shall also state the intended application and audience, or, who will use the results (ISO 14040:2006). The scope of the study defines the main characteristics of the LCA study in terms of the system boundaries relating to the “temporal, geographical and technological coverage”, the impact assessment methodology and the level of detail for the study (Guinée et al , 2004, p35). Temporal coverage includes the definition of the period for the production, use and waste treatment for the system (Baumann and Tillman, 2004), which in turn defines the age of data, the data collection timeframe, the reference period for normalisation. Geographical coverage is important for assessing local or global impact characteristics. Technological coverage may consider the least efficient case or current average technology in the given geographical boundary (Guinée et al , 2004). Defining the system boundary is related to the goal of the study and can be somewhat subjective, therefore, all assumptions must be clearly stated (EEA, 1997). The limitations of the study must also be stated. These may be defined by decisions made regarding the system boundary or through lack of data availability for certain processes. The modelled flows within the system must be related to a quantifiable function of the system in order to evaluate the environmental impacts. This is described as the functional unit (Baumann and Tillman, 2004). The functional unit must be defined as clearly as possible to allow for the comparison of different systems and the evaluation of whether an equivalent function is performed (Guinée et al , 2004).

2.1.1.2. Life Cycle Inventory Analysis

The second phase in an LCA study is the life cycle inventory analysis (LCI). This is the data collection phase where data on relevant mass and energy flows are entered into a flow model of the system (Baumann and Tillman, 2004). It is the “process of quantifying energy and raw material requirements, atmospheric emissions, waterborne emissions, solid wastes, and other releases for the entire life cycle of a product, process, or activity” (USEPA, 2006, p19). The first step is to construct a detailed flowchart of the system to identify the data requirements. The second step is data collection, which is the most time consuming process in the LCA. Data pertaining to the raw material and energy inputs, the emissions and the product itself, should be sought. All calculations, assumptions and data gaps should be documented. If allocation between co-products is necessary then, data in support of the allocation method should also be collected (Baumann and Tillman, 2004). Otherwise, the system boundaries should be expanded to include the co- products (EEA, 1997). Validation of the data should be performed by comparison with other data or by examining mass or energy balances (Guinée et al , 2004). The final step is calculation of the environmental loads of the system in relation to the functional unit. This is performed by normalising the input and output data to the defined function of the system by linking the upstream and downstream processes. The LCI is an iterative process and the flow model may be revised as more detail is learned about the system. (Baumann and Tillman, 2004; EEA, 1997).

2.1.1.3. Life Cycle Impact Assessment

The life cycle impact assessment (LCIA) phase takes the results of the inventory analysis and aims to translate the environmental loads into potential environmental impacts or consequences, such as acidification or ozone depletion. The purpose of this is to formulate the results in a format that is easier to interpret by the intended audience (Baumann and Tillman, 2004; EEA, 1997). The LCIA is broken down into a number of sub-phases. Impact category definition ; the selection of the relevant impact categories, such as resource depletion or global warming potential. Classification; assigning the LCI results to the relevant impact categories. Characterisation ; modelling the LCI impacts in terms of scientifically established

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

indicators for each impact category. Grouping ; the sorting and ranking of indicators, for example, by global / regional / local impacts or by emissions to air or emissions to water . Normalisation ; applying a scaling factor to the characterisation results to relate them to a reference value, for example, background emissions in a certain region. Weighting; assigning a relative value to the impact categories based on their perceived importance. Data quality analysis ; assessing the uncertainty and sensitivity of the data, usually involving a sensitivity analysis on key inventory data (Baumann and Tillman, 2004; EEA, 1997; Guinée et al , 2004; USEPA, 2006). Grouping, normalisation and weighting can be considered optional elements of the LCIA (ISO 14044:2006).

2.1.1.4. Life Cycle Interpretation and Results

Interpretation is the process of assessing the results of the LCI and LCIA and presenting them in accordance with the goal and scope of the study. As shown in Figure 2, interpretation is undertaken throughout the LCA to continually assess if the goals and scope of the study are being fulfilled in order to facilitate the decision making process (EEA, 1997). The elements of interpretation are; identification of the significant environmental issues in accordance with the goal and scope of the study; evaluation of the robustness of the model, including a discussion of choice of data, assumptions made and checks for sensitivity and consistency of the data; making conclusions and recommendations and undertaking a critical review of the LCA study. This final element is important to ensure the transparency of the study, that conclusions are drawn based on facts and that the uncertainties and limitations are understood and communicated (EEA, 1997; USEPA, 2006). A particularly good method for communicating the most important impacts is in a bar diagram (Baumann and Tillman, 2004).

2.1.2. Limitations and Criticisms of LCA LCA is becoming an increasingly accepted tool for evaluating environmental impacts. While software programs and databases of generic processes are making LCA easier to perform, some limitations remain (Hunkeler and Rebitzer, 2005). The US Environmental Protection Agency note in their LCA guidelines that the “use of commercial software risks losing transparency in the data. Often there is no record of assumptions or computational methods that were used. This may not be appropriate if the results are to be used in the public domain.” (USEPA, 2006, p28).

LCA addresses only those environmental issues specified in the goal and scope. As such, LCA should not be considered a full environmental assessment of a product system (ISO 14040:2006). Social and economic issues are not addressed in an LCA and the environmental impacts are described as “potential impacts” as they are not fixed in time and space and are often related to an arbitrary functional unit (Sleeswijk et al, 2008). Furthermore, variation in the spatial and temporal dimensions of the LCI introduces uncertainty in the results (ISO 14040:2006). Sleeswijk et al ( 2008) note that the results cannot often be scaled down to the local level and that typically, an LCA focuses on the steady-state rather than a dynamic system, in other words, not taking into account future technological advances.

The use of databases of generic processes has been criticised as these may not correspond to actual system processes. Furthermore, that unpublished proprietary data is unverifiable therefore, it is often impossible to determine the reliability of data. The disregard for elementary mass balance is a further criticism. The mass of the inputs should equal the mass of the outputs (Ayres, 1995).

Methodologies for assessing the consistency and accuracy of inventory data are generally lacking, which can lead to uncertainty and inconsistency in the results (ISO 14040:2006). It is argued that within the LCIA phase there has been a lack of consistency regarding which impact categories to include. Efforts are underway to provide guidelines on recommended LCIA practice however, given the current lack of agreement, this is no small undertaking (Bare and Gloria, 2006; ISO 14040:2006).

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

Lack of validation in LCA is another criticism. Validation is the process of assessing the LCA model to see if it, in fact, mimics and behaves equally to the real system (Ciroth and Becker, 2006). If LCA studies can be validated, then Ciroth and Becker (2006) conclude it would provide improved models and thus, an improvement in the quality of decision making based on those models.

Finally, LCA should be considered an analytical tool used to support decision making. It should not be relied upon to replace the decision making process rather, be used in conjunction with other evidence to assist in decision making (Sleeswijk et al, 2008).

2.2. Submarine Cable Systems Communication by submarine cable has a long history, dating back over 150 years to the first telegraph cable between England and France. The first transoceanic cable was completed shortly after connecting England with the US. Since these early origins technology has developed from analogue telegraph and coaxial transmission, to high speed/high capacity digital transmission over fibre optic networks. A map of the global fibre optic submarine cable systems is shown in Figure 3.

Figure 3: World map of submarine cables (Alcatel, 2009)

2.2.1. Historical Development Communication by submarine cable has been described as having followed an “evolutionary process… punctuated by a series of epochal events marking technological advances” and, regulatory change (Beaufils, 2000, p15). This era began when the very first telegraph cable was laid in 1850 across the English Channel, from Dover to Calais. The first attempt at a transoceanic cable was undertaken in 1857 (Letellier, 2004), though it was not until 1866 that the first successful transoceanic cable was installed between the UK and the US. The second age of submarine communication was the development of the

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

first coaxial cables for submarine use in the 1950s. The first transoceanic cable to carry voice communication was Trans-Atlantic Telephone 1 (TAT-1), laid in 1956. This cable had a capacity of 36 circuits however, 15 of those circuits were only ever sold.

Figure 4: Development of submarine cables (Hilt, 2009)

The development of digital technology using optical fibre in the 1980s was the catalyst to facilitate rapid expansion of modern telecommunications. Fibre optic technology dramatically increased cable capacity relative to construction and maintenance costs, thereby increasing the returns (Beaufils, 2000). The first fibre optic submarine cable was laid in the Canary Islands 1985 and the second between the UK and Belgium in 1986 (Amano and Iwamoto, 1990). The first transoceanic fibre optic cable was Trans-Atlantic Telephone 8 (TAT-8), installed in 1988. At the time, this cable provided more capacity than all previous cables combined with 280Mbps or 35,000 voice circuits over 2 fibre pairs (Letellier, 2004). A timeline of submarine cable development is shown in Figure 4.

2.2.2. Modern Systems Modern fibre optic systems are composed of a number of key components; the submarine cable, submarine repeaters, branching units, submarine line terminal equipment (SLTE), and the power feed equipment (PFE) (Letellier, 2004).

As the light signal travels along the fibre it becomes degraded and consequently, longer systems need to be amplified at regular intervals by submarine repeaters. The repeaters are powered by approximately one ampere (1A) of direct current supplied from the terminal station via a copper conductor built into the cable. The early fibre optic cable systems used electronic amplifiers within the repeater, where the signal was regenerated and retransmitted by a light source within the repeater. As technology progressed, purely optical repeaters, which work by “exciting” the existing light signal, were developed. This created an end- to-end optical channel. The first fully optical system was installed in 1995 (ibid ). This was a major development, as optically amplified systems can generally be upgraded without replacement of the cable or subsea components. This is achieved by upgrading the transmission and network management systems at the terminal station to provide faster transmission and more signals per fibre. Wave division multiplexing (WDM) is used to transmit more than one active signal or light wavelength on a fibre-pair 1. By upgrading the WDM equipment, more optical channels can be added to a fibre-pair without upgrading the submarine cable (Beaufils, 2000). Greater flexibility, security and resilience is also provided with optical

1 Fibres are designed to function in pairs to support transmission in both directions (Letellier, 2004).

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

add/drop multiplexing (OADM) and optical cross-connect (OXC) networking, where signals can be converted or exchanged between wavelengths (Akiba and Yamamoto, 1998). Current technology provides 10Gbps transmission rates, with hundreds of wavelengths possible on a single fibre-pair (André and Brochier, 2007). However, there are technical limitations to the maximum transmission rates and number of wavelengths depending on age of the system (Beaufils, 2000). Research and development continues. Recently, it was demonstrated that 40Gbps transmission on a single wavelength was possible over the Atlantic using the existing TAT-14 cable, built in 2001 (Städje, 2009) and single-channel bit rates have attainted 100Gbps in other experiments (Gnauck and Chraplyvy, 2008).

2.2.3. System Architecture Systems are designed in a variety of configurations depending on the requirements of the network. There are three principle configurations; unrepeated, branched or ringed systems, described in more detail in the following sections.

2.2.3.1. Unrepeated systems

Unrepeated systems are typically 200-300 kilometres long and use point-to-point architecture. They do not have built in submarine repeaters and therefore, do not need to be powered along the length of the cable. However, due to the limiting factors of signal loss and corruption, unrepeated systems are unlikely to exceed 500 kilometres in length. The cable is design so that amplification takes place at both the transmitting and receiving terminal stations. Unrepeated systems have been used extensively in interconnecting the British Isles and in connecting the UK to Europe. Festoon systems are a series of unrepeated point-to-point cables connected into a network (Alcatel, 2009; CPNI, 2006). Figure 5 shows a typical unrepeated festoon system.

ADM: Add/Drop Multiplexer

TM: Terminal Multiplexer

Figure 5: Unrepeated "festoon" system (adapted from Alcatel, 2009)

2.2.3.2. Branched systems

Branched systems are designed for long routes that service several countries from a single trunk cable. Due to their length, electrically powered repeaters must be installed along the length of the cable to counteract the signal losses in the fibre (CPNI, 2006). Branching units, also acting as repeaters in some cases, are placed at the junction between the landfall cable and the trunk cable to direct the signals along the appropriate paths. These systems are used extensively between Europe and the Far East. They are

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

particularly beneficial where neighbouring countries are hostile toward one another as cables are generally routed in international waters (Alcatel 2009; CPNI, 2006). Figure 6 shows a typical branched system.

TM: Terminal Multiplexer

Figure 6: Branched system architecture (adapted from Alcatel, 2009)

2.2.3.3. Ringed systems

Ringed systems were designed to provide redundancy in the event of a cable fault in a point-to-point system. As technology developed, new cables were installed that had more capacity than all former cables combined and hence, there was a need to provide redundant capacity and back-up in the event of a cable fault. This was achieved by constructing the cable as a ringed network with two separate oceanic legs. Due to their length, ringed systems require repeaters and a powered cable, similar to branched systems. This type of system architecture is utilised particularly across the Atlantic and Pacific Oceans to connect the US with, Europe, Asia and Australasia. (Alcatel 2009; CPNI, 2006; Ruddy, 2006). Figure 7 shows a typical ringed system.

ADM: Add/Drop Multiplexer

Figure 7: Ring system architecture (adapted from Alcatel, 2009)

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

2.2.4. System Components The main components of a submarine cable system are the submarine cable itself, the repeaters and branching units and the terminal station equipment, as shown in Figure 8.

Figure 8: Components of a submarine cable system (Letellier, 2004)

2.2.4.1. Submarine Cable

Submarine cable is designed to protect the optical fibres 2 from a hostile marine environment. The cable must be able to withstand the tensions induced during installation and recovery, variable seabed conditions (such as rock or steep slopes) and the high pressures of deep sea installation down to 8,000m (Beaufils, 2000). Cables are built around the fibre unit structure ; a metal tube designed to house the fibres in a stress-free environment and contains the fibres and a water blocking gel used (Fullenbaum, 2004). For deepwater 3 installation, the fibre unit structure is typically surrounded with strengthening wire, the copper conductor and then, high-density polyethylene for insulation and abrasion resistance (see lightweight cable, Figure 9). Various levels of armouring using carbon steel wires and bitumen sealant are built around this basic cable structure to protect the cable in shallow water, where the majority of faults occur (Letellier, 2004). Cables are designed to withstand the marine environment for 25 years (Beaufils, 2000).

2 Optical fibre strands are very clear, flexible filaments of glass, slightly thicker than a human hair (Corning, 2009). 3 The division between shallow water and deepwater is normally considered to be the 1000 metre contour (Trischitta et al , 1997).

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

Figure 9: Types of submarine cable: Double Armoured (DA), Single Armoured (SA), Lightweight Protected (LWP) and Lightweight (LW) (Beaufils, 2000).

2.2.4.2. Repeaters and Branching Units

Repeaters and branching units are contained in pressure resistant structural housings made from specially blended alloys, such as, beryllium copper of nickel-chrome-molybdenum (Amano and Iwamoto, 1990). Similar to the cable, they must be able to withstand the tensions induced during installation and recovery and the high pressures of deep sea installation down to 8,000m. As a transoceanic system may require hundreds of repeaters to be installed along the route, they have stringent standards for reliability (Letellier, 2004). Recovery and replacement of a submerged plant involves considerable time, cost and disruption to service. The repeater manufacturer Fujitsu, for example, has never had a ship-based repair and has delivered roughly 2400 repeaters (Harasawa et al , 2008). The distance between repeaters is typically 40 to 110 kilometres and is dependent on the total system length (Suyama et al , 1999; Letellier, 2004).

Section 2.2.2 provides additional details on repeaters.

2.2.4.3. Terminal Equipment

The link between the submerged plant and the terrestrial networks is the cable terminal station. The main components of the terminal are the submarine line terminal equipment (SLTE) and the power feed equipment (PFE) (Trischitta et al , 1997), with additional support equipment such as, batteries and back-up generators to supply continuous power in the event of failure in the main supply.

The SLTE controls the transmission and receive of the light wave signals through the following components: High Precision Optical Equipment (HPOE) for transmitting and receiving each signal; Initial Loading Equipment (ILE) to load the unused spectrum of the channel wavelengths; Line Monitoring Equipment (LME) to detect errors in the subsea plant; Wavelength Terminating Equipment (WTE) to combine or separate the individual light wave signals; Terminal Line Amplifier (TLA) to strengthen the signal before being sent or received (Breverman et al , 2007). All of these components are housed within standard floor standing racks within the terminal station (Oikawa et al , 2006). The components of the SLTE are shown in Figure 10.

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

HPO E: High Precision Optical Equipment

ILE: Initial Loading Equipment

LME: Line Monitoring Equipment

WTE: Wavelength Terminating Equipment

TLA: Terminal Line Amplifier

STM: Synchronous Transport Mode

Figure 10: Architecture of the SLTE (Breverman et al , 2007)

The PFE supplies a direct current of approximately one ampere (1A) to the repeaters via the copper conductor in the cable. Two systems are used in parallel, each capable of supporting the entire load in the event of failure in one system. Depending on the configuration, a PFE can generate up to 10 kilovolts (10kV) in order to power the hundreds of repeaters in a transoceanic system (Letellier, 2004). The PFE consists of direct current converters and monitoring equipment housed within standardised floor mounted racks within the terminal station (Alcatel, 1998).

2.2.5. System Design and Installation Once the decision has been taken to construct a new system, the planning and implementation phase begins. This phase is critical for the long-term reliability of the network. The cable must be designed and installed to be protected from external aggression, such as, commercial fishing, anchoring and hostile seabed. All of this information is collected during the feasibility study known as the desktop study (Beaufils, 2000). This identifies the best route and estimates cable types and burial requirements. Following the desktop study, the electronic route survey (ERS) and the electronic burial assessment survey (EBAS) begin. The two surveys are completed by a research ship that collects the geophysical profile of the seabed along the route (Letellier, 2004) to map the topography, identify hazards and evaluate sediments (Beaufils, 2000). This data then facilitates the final system design and allows for the assembly of the sections of the submerged plant. For protection against external aggression, the cable is typically ploughed into the seabed out to 1,000 of metres water and surface laid in deeper water (Letellier, 2004). Prior to laying, route clearance is performed for areas that are to be buried, to remove old cables and other hazards. This operation is known as a pre-lay grapnel run . The installation of the cable is performed by purpose-built cable ships. Surface laying of the cable achieves between 150 to 250 kilometres per day, while burial reduces the speed to 10 to 40 kilometres per day (Beaufils, 2000).

2.2.6. System Operation and Maintenance Due to the time, cost and disruption to service involved in replacing submerged plant (Harasawa et al , 2008), cable systems are designed for high resilience against failure. The most common type of fault results from physical damage to the cable from external aggression (CPNI, 2006). External aggression relates to fishing activity, anchoring, dredging, crushing and geological activity and accounts for over 70 percent of all system faults (Kordahi et al , 2007). Components are extremely reliable and component failure rare. Over 80 percent of external aggression faults can be attributed to human activity, with fishing the major cause at 60 percent. The majority of external aggression faults, at 40 percent, also occur in less than 100 metres of water. Statistical data from 2001 to 2006 shows an annual fault rate, normalised to 1,000 kilometres of cable, of 0.8 to 0.05 for depths less than 1,000 metres and 0.02-0.13 for depths greater

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

than 1,000 metres (Kordahi et al , 2007). Fault protection against service disruption for transoceanic systems can be further provided by pairing cables in a ring structure with adequate capacity and automatic protection switching in the event of failure (CPNI, 2006). Should a fault occur in the cable, a purpose- built cable ship must be dispatched to the localised fault area. The repair operation involves cutting and recovery of each end of the cable, identifying and removal of the faulty section of cable and splicing in a new section, before lowering it to the seabed again (Alcatel, 2009). In the past, cable owners generally maintained their systems with in-house maintenance programs. However, in recent years, it is more common to out-source maintenance operations to specialist Marine Maintenance Providers (Herron et al , 2007). Willey et al (2007) estimated that in 2007 there were 45 ships capable of undertaking submarine cable operations.

2.2.7. End-of-Life Decommissioning Submarine cable systems are built with a technical lifetime of 25 years, however in reality, most systems are retired much earlier as they no longer remain commercially viable (Willey et al, 2007). Once a cable system becomes too expensive to maintain due to age or commercial reasons, they are decommissioned. Here, two options exist:

Firstly, no recovery is attempted and the cable remains on the seabed with no material recovery. Many countries have requirements to remove cables from their shallow water coastal regions when retired, though, the obligation to do so differs from country to country. When recovery is not mandatory, then the cable owners can decide to leave the cable on the seabed or recover the cable. In most cases the cable must be removed from within the 12 nautical mile limit (Ridder, 2007). Willey et al (2007) make an analysis of the number of decommissioned submarine cables and noted that there is a large amount of fibre optic cable lying on the seabed, predominantly in the northern hemisphere.

Secondly, the cable is recovered and either re-laid (e.g. for scientific use) or recycled for the materials (Ridder, 2007). Recently, a company has considered recovery and recycling of submarine cables commercially viable and have recovered 350km of the SAT-1 cable 4 in order to recycle the steel, copper and polyethylene (Louw, 2009). Recovery and relaying of the system for further use, however, needs greater care to ensure that the cable is not damaged. Successful operations have been undertaken where decommissioned systems provide donor cable for new systems giving both time and cost benefits (Merret and Laude, 2007). Cable sections are also being re-laid to bring power and telecommunications to underwater observatories used for scientific purposes (Lecroart et al , 2007).

4 SAT-1 is an older coaxial cable decommissioned in 1993. Recovery and recycling is assumed to be similar for fibre optic cables. (Louw, 2009)

2. Theoretical Framework A life cycle assessment of fibre optic submarine cable systems

2.2.8. Submarine verses Satellite Transmission Submarine cables are often challenged by satellite transmission. While satellite systems are appropriate for television and remote access, submarine networks are unsurpassed for transmitting high capacity data traffic between countries. Submarine networks are superior in capacity, transmission quality, confidentiality, capacity to upgrade, lifetime and maintenance, among other factors (Beaufils, 2000). A comparison between satellite and fibre optic submarine cable communication is presented Table 1.

Table 1: Comparison between satellite and submarine cable communication (Adapted from Barattino and Koopalethes, 2007; NEC, 2008).

Comparison Factor Satellite Optical Subsea

Latency 250 milliseconds 50 milliseconds

Design life 10-15 years 25 years

Capacity 48,000 channels 80,000,000 channels 5

Unit cost per Mbps capacity $737,316 US $14,327 US

Share of traffic: 1995 50% 50%

Share of traffic: 2008 3% 97%

Thus, “owing to their high capacity, high reliability, and high signal quality, submarine cable systems are well suited to trunk transport and backbone network infrastructure” (Beaufils, 2000, p.17).

5 Based on a system specification utilising: 10Gbps at 128WDM and 4 fibre pairs.

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

3. Goal and Scope

This section describes the goal and scope of the study and details the context within which the assessment of the environmental impacts of the studied system has been determined. The functional unit and system boundaries are defined, along with the methodology, assumptions and limitations of the study.

3.1. Goal The goal of this LCA is to study a fibre optic submarine cable system from a life cycle perspective to examine the potential environmental impacts of sending data over the cable networks that span the oceans. This study will attempt to collect as complete and up-to-date data in order to construct an LCA model. This model will be analysed in order to identify the significant resource and energy inputs and the subsequent emissions over the life cycle of the system. Furthermore, it will attempt to identify those activities in the life cycle of a submarine cable that have the most significant potential environmental impacts. This information may then be used to highlight those activities where the environmental performance of submarine cable systems may be improved and may also be used as a link in the process of mapping the impacts from the global ICT network.

3.1.1. Target Audience The results of this study will be used primarily by Ericsson in their business activities, thereby making a positive contribution in the area of environmental sustainability. However, as no previous LCA study of submarine cables appears to have been undertaken, it is envisaged that the results will also be of interest to other researchers in the field of environmental assessment and ICT and the submarine cable industry as a whole.

3.1.2. Applicability of this Study Submarine cables carry the vast majority of transcontinental voice and data traffic. The world map of submarine cable networks (Figure 3) shows that Europe, North America and the Far East are well connected with many cable systems spanning the oceans. The high capacity and bandwidth of these cables makes it possible to transfer large amounts of data around the world almost instantaneously (Jonsson, 2009a). “Without sub-sea cable systems, global telecommunications at the level we know today would be impossible.” (CPNI, 2006, p.18). Yet, little is known about the environmental impacts of a submarine cable from a life cycle perspective (Jonsson, 2009a). Research revealed no known LCA studies in this specific area and consequently, it appears that a knowledge gap has been identified. This study is seen as the first step toward bridging that gap.

3.2. Scope

3.2.1. System Description The system studied in this LCA is a generic submarine cable system utilising fibre optic cable, submarine repeaters and land terminal stations. The system boundary is drawn at the limits of the two end terminal stations and includes all components within the terminal to the point where the signal is transferred over to the terrestrial network. The function of the system is to transfer large volumes of data at high-speed around the globe. Using a cradle-to-grave approach, five phases have been identified over the 13 year life cycle of the system; raw material extraction , design & manufacturing , installation , use & maintenance and end-of-life decommissioning .

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

3.2.2. Functional Unit The function studied in this LCA is that of sending a specific amount of data over a specific length of fibre optic submarine cable. This function has been chosen in order to evaluate the potential environmental impact of sending data across the seabed using submarine cables given a lifecycle perspective. The functional unit is defined as;

ten thousand gigabit kilometres

and, can be written as 10,000Gb·km . This can, for example, can be interpreted as sending 1Gb of data over 10,000km of cable, which is approximate the average distance of sending data one way across the Atlantic and Pacific Oceans (see Yellow/Atlantic Crossing 2 and Pacific Crossing 1 in Appendix A.

In choosing this functional unit, consideration has been given to developing a model that is scalable from point to point. For example, should the end user of this study wish to estimate the impact of sending telepresence data at 18Mbps (Jonsson, 2009b) between say Stockholm and New York, then the model can be scaled by the amount of data traffic and the length of the cable between terminal stations. In this case, the functional unit would equate to 1.25Gb of data sent over 8,000km of cable and this amount of data would be sent every 69 seconds. To give another example; Beaufils (2000) estimates that a telephone call consumes 85kbps, assuming the same distance, Stockholm to New York, then the functional unit would relate to approximately 4 hours of call time.

3.2.3. System Boundaries and Delimitations This study considers the life cycle of a fibre optic submarine cable system from cradle-to-grave . The boundaries begin with nature and the extraction of raw materials and end again with nature and the emissions to soil, air and water. All significant phases, including raw material extraction, design & manufacturing, installation, use & maintenance and end-of-life treatment have been considered for inclusion within the system boundary, as shown in Figure 11.

Figure 11: Life Cycle stages of a submarine cable (Adapted from USEPA, 2006).

Defining system boundaries is rather subjective and needs to consider several dimensions; temporal, geographical and technological coverage (EEA, 1997). System boundaries in this study have been selected based principally on the theoretical framework of submarine cables systems set out in Section 2.2 and professional experience of over 10 years working with cable systems.

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

3.2.3.1. Temporal Boundary

The lifetime of a submarine cable system can be divided into three distinct temporal phases: the planning, design, manufacture and installation phases of the system, today, taking approximately 18 months from the signing of the contract (Letellier, 2004); the use & maintenance phase, where the commercial lifetime of the cable system is approximately 10-15 years, though the cables are conservatively designed to operate for 25 years (Beaufils, 2000); and the time horizon of a few years for decommissioning where emissions continue to have an effect until the material recycling process is complete. Analysis of 29 retired systems shows an average lifetime of 13 years, which is due to technological advances and capacity upgrading making older systems expensive to maintain per unit capacity (Ridder, 2007). Thus, the low capacity systems become uneconomic due to the high administration and maintenance costs (CPNI, 2006). In this study, the commercial lifetime of 13 years has been considered for the use & maintenance phase, with a sensitivity analysis undertaken for a 25 year technical lifetime scenario.

Data from the year 2000 to present day is considered to relate to the modern submarine cable system. A list of all significant data sources and the given age is found in Appendix B. Based on the CML methodology, no consideration is given in this report to the “temporal distribution of activities, emissions and effects” (Guinée et al , 2004, p.42). Furthermore, while finite time horizons for future emissions from landfills can be modelled through sensitivity analysis, this has not been explored and this problem is not addressed in this study. Nor, has this been considered for an end-of-life scenario where the cable is abandoned on the seabed.

3.2.3.2. Geographical Boundary

Submarine cable systems span the Earth’s oceans and, as such, have global environmental impacts. Subject to availability, data from a number of cable systems, representing different geographical regions, has been aggregated. It is assumed that aggregated figures will represent a reasonable estimate of the average global impact. The vast majority of cables connect the US, Europe, Japan and China; see Figure 3. Cable and component manufacturing plants are also located in these regions; see Appendix C – Stakeholder Analysis. Therefore, electricity has been calculated from an equal mixture of production from the US, EU, Japan and China as detailed in Section 4.4.1.1.

3.2.3.3. Technological Boundary

The technical boundary of this study is defined as the limits of the land terminal station; where the signal is received from, or transmitted to, the terrestrial link. This includes the submarine cable and repeaters connecting the terminals and all significant internal components of the terminal station such as the power feed equipment (PFE) and the submarine line terminal equipment (SLTE), as shown in Figure 8 (Letellier, 2004). Optically amplified systems are generally designed to be upgraded without replacement of the cable itself, achieved by upgrading to higher capacity transmission components at the terminals (Beaufils, 2000). Current technology uses 10Gbps transmission rates with up to 64 signals on a single fibre pair (Ridder, 2007). It is assumed that cable systems installed after the millennium represent the current technical capacity. No future scenario utilising 40Gbps technology has been considered, as the model can be scaled linearly to account for this increase in capacity.

The inclusion of capital goods in the LCA model is a debated topic. In accounting LCAs, such as this study, it is recommended that the study be as complete as possible and thus include capital goods. In reality the limitations of the study, particularly time and data availability, may make this often infeasible. (Baumann and Tillman, 2004). In this study, only the use phase of capital goods, such as buildings, cable

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

ships and other vehicles have been included, as described in Section 4 – Life cycle inventory. All other phases of the capital goods life cycle have been excluded.

End-of-life decommissioning includes the recovery of the cable from the seabed and the recycling of materials. A portion of these recovered materials, such as bitumen soaked polypropylene yarn, cannot be recycled and end up in landfill (Louw, 2009). The system boundary is drawn at the landfill. Any emissions from the landfill are accounted for in the standard GaBi software database processes. Materials entering the landfill are described in Section 4 - Life cycle inventory.

3.2.3.4. Other Delimitations

Submarine cable system architecture is shown to consist of transoceanic repeated systems and shorter non-repeated systems (less than 500 kilometres). Non-repeated systems do not require power to be sent down the cable and hence do not have the copper conductor present in cable designed for repeated systems. In order to reduce the complexity of the LCA model, non-repeated systems have not been included in this study.

This LCA is performed to account for the significant inputs and outputs of a generic cable system and, as such, no comparison with other methods of data transfer, for example satellite transmission, has been attempted. The calculation of the “generic” cable system is defined in Section 6 - Calculation Procedure.

Branching units, equalisers and joints are necessary parts of the cable system (Ridder, 2007), however their frequency is minor in relation to the overall cable length and number of repeaters. For this reason, these components have been delimited from this study, as their potential impacts are considered to be small in relation to the total system.

System spares are housed in depots throughout the world. These spares include different types of cable, repeaters, jointing boxes and consumable kits for making repairs to the cable should a fault occur (Ridder, 2007). Spares have not been considered in this study, as it is assumed that their impact on the final result is minor in relation to the total system.

3.2.4. Data Requirements and Data Quality Throughout the study, one objective has been to collect as complete, accurate and up-to-date data as possible on all processes within the system boundary and in the given timeframe. Another objective has been to provide a general picture of a generic cable system given current industry practice and technology. Cable systems are highly variable in length and level of cable protection depending on regional characteristics. However, system components are generally similar. Where possible, data has been collected from multiple sources to account for this variation and as far back in the processing chain as was considered necessary. Much of the data relating to intermediate material production has been collected from actual suppliers, though much of this is propriety information and single source, making verification difficult. Propriety data will also reduce the transparency and reproducibility of the study, nevertheless, this is an unavoidable limitation. Raw material extraction and refining has been estimated from processes contained in commercial inventory databases. Where no data was available from suppliers, cable owners or service providers, previous LCA studies, commercial databases and other literature were used to provide a best approximation. In using data from these sources, temporal, geographical and technological considerations were made to evaluate suitability and uncertainty. In some cases, qualified assumptions have been made to fill data gaps and these are clearly stated in the life cycle inventory, Section 4. Data validation was attempted on all data by comparing multiple sources (where available) and by considering the age, location and technological relevance of the data. A list of all data sources, year and region is presented in Appendix B.

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

Where uncertainty is considered high or a data gap exists, sensitivity analysis has been undertaken. The results of the data quality analysis are presented in Section 7.5.

3.2.5. Methods for Inventory Analysis The methodology for inventory analysis follows the steps of system modelling, data collection and calculation of the environmental loads, as detailed in Section 2.1.1.2.

Firstly, the general flow chart (Figure 11) was expanded to provide more detail of the system processes from a cradle-to-grave perspective. The flow chart sets out the hierarchical levels of the system. Two main sub-models were identified: the cable (including repeaters) and the terminal station. Under each sub-model the individual components were then categorised. The life cycle phases of the system were identified by the system boundary considering a cradle-to-grave approach and modelled for each sub-model. System modelling for a submarine cable system was based on the literature and professional experience. The detailed system flowchart is presented in Appendix D.

Completion of the general flow chart facilitated the data collection process. Initially a broad range of data was collected for all process within the system boundary. Then, as the collection process advanced, focus was placed on data gaps and modelling in more detail those processes assumed to have the greatest influence on the result. Materials having substantial weight or processes assumed to affect the result have been included in the model. Where data was not available, inventory cut-off was applied to processes with minimal weight and considered to have minimal environmental impact in relation to the total result. Commercial databases from the GaBi software were used to represent raw material extraction and processing. Where these processes were deemed insufficient to model actual material processing, an attempt was made to contact suppliers further back in the product chain, for example, optical fibre manufacturers. Processes that could not be found in public databases or in the literature, have been represented by a comparable process or treated as a data gap.

A closed loop recycling process is assumed for all recycled materials at end-of-life. Closed-loop recycling assumes that the inherent properties of the material are maintained and that the production of virgin material is offset by the recycled material, thus avoiding the need for allocation (EAA, 2007; Giurco et al , 2006).

Calculation of the flows within, and the flows crossing, the system boundary has been facilitated by the LCA software GaBi, explained further in Section 3.2.7. All cable related data has been normalised to 1,000 kilometres of cable within the model, in order to allow for straightforward up-scaling to the functional unit. All terminal data has been scaled to a single terminal station. The data collection process is further explained in Section 4.3.

3.2.6. Methods for Impact Assessment The methodology used for impact assessment in this study follows the framework of the ISO guidelines (ISO 14044:2006). The characterisation models are those developed by the Institute of Environmental Sciences (CML) at the University of Leiden, in the Netherlands. This CML methodology uses a problem- oriented approach that focuses on environmental problems or the so-called midpoint of the cause-effect chain (Guinée et al, 2004). In this study ten baseline impact categories (group A categories) have been selected using the CML 2001 characterisation database, and are detailed in section 5.2.

A submarine cable system uses significant amounts of electricity in order to power the repeaters and terminal equipment over its lifetime. Emissions from the combustion of marine fuel are considered to present the other major impact, in particular, having significant acidification potential (Eyring et al , 2007).

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

On this basis, three additional categories were defined in order to analyse the energy consumption over the life cycle of the system.

Table 2: Additional categories used for the life cycle impact analysis.

Impact Category Unit

Primary Energy MJ

Electricity kWh

Heavy Fuel Oil kg

Primary energy has been modelled using the mass balance utility in the GaBi software. Parameter counters were made within the software to track the electricity and heavy fuel oil inputs for each sub-model.

Normalisation has been undertaken to relate the characterisation results to a background, or reference value, in order to frame the magnitude of the potential impact. This identifies if the impact is significant in relation to the total impacts of the studied area, which may have global or regional consequences (Baumann and Tillman, 2004). Normalisation factors used in this study are taken from Sleeswijk et al (2008), where the annual world reference values for the emissions and consumption of the significant substances under each impact category are collated for the year 2000.

3.2.7. Software There are many programs available to assist the LCA practitioner with life cycle modelling (Spatari et al , 2001). The software system GaBi 46 was used for this study. GaBi is a software tool developed by PE International and the University of Stuttgart to assist in modelling life cycle balances and analysing and interpreting the results (PE & LBP, 2008). It has been described as one of “the top ten fully integrated and comprehensive software tools” (Spatari et al , 2001, p82). It allows the user to assess the technical, social- economic and environmental impacts of a product, service or system to produce a comprehensive balance of inputs and outputs. The software has been developed to conform to the ISO 14040 series and is built on a modular system of plans, processes and flows. This modular system is shown in Figure 12. In addition, standard databases are supplied with life cycle balance data for common industrial processes (PE & LBP, 2008).

6 GaBi 4 version numbers; Compilation: 4.3.78.1, DB version: 4.126

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

Figure 12. Graphical representation in GaBi, built on a modular system (PE & LBP,2009).

LCIA can be performed using standard classification and characterisation databases, such as CML2001 (Guinée et al, 2004) and Eco-indicator 99 (Pre, 2000). However, it is also possible to enter environmental indicators manually (Bergelin, 2008).

3.2.8. Study-Wide Assumptions, Simplifications and Limitations A life cycle assessment study requires a significant effort in collecting and analysing large amounts of environmental and product data. As the whole life cycle is studied, the complexity of a system makes it impossible to model every detail (Baumann and Tillman, 2004). As such, the LCA model is a simplification of reality and this is important to remember when interpreting the results (3GLCA, 2002). This section describes the central assumptions, simplifications and limitations of this study.

This LCA has been undertaken to assess the potential impact of a generic submarine cable system. As such, data has been taken from many sources covering different regions, yet the results are intended to represent an average cable for any geographic region. Where only single source data was available, it is assumed that this is representative of the generic process. Single source data raises the uncertainty of the result as verification is not possible for this data. An unavoidable limitation is the use of propriety data. This reduces the transparency and reproducibility of the study as the data remains confidential (Baumann and Tillman, 2004). Whenever propriety data has been used, it is clearly stated in the LCI, Section 4. Detailed confidential data has not been presented in the LCI section, however, this has not restricted the presentation of the final result.

Energy use appears to be the largest contributor to the environmental impacts. Electricity production has been simplified to four differing geographical regions and these are assumed to represent a realistic electricity mix for the purposes of this study. Heavy fuel oil (HFO) production has been modelled from an

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

equal mix of US and EU production. As no other data for HFO production could be sourced in the given timeframe, it is assumed that this simplified model is representative of world production. However, this data will be biased toward production methods in the EU and the US.

All transportation of raw materials is assumed to be undertaken overland at an average distance of 1,000km. All transportation of wastes and recycled materials are assumed to be undertaken overland at an average distance of 100km. Given the doubt in these assumptions, a sensitivity analysis has been undertaken for each process in order to evaluate the impact on the result. Further details of these assumptions are presented in Section 4.4.2 - Transportation. The results of the sensitivity analysis are presented in Section 7.5.1.

This study uses data supplied by Ericsson for the production of submarine cable. It is assumed that the manufacturing process is representative of all cable manufacturers. However, Ericsson does not produce cable for the repeated systems that use additional materials, in particular the copper conductor. Therefore, the mass of each material per unit length of cable has been scaled from the cross-sectional area of cable designed for a repeated system. This induces a level of uncertainty, though the resulting total mass per unit length was in good agreement with the documentation and therefore, it is concluded that the result is not affected.

Data for the raw materials and manufacturing processes of the terminal equipment and submarine repeaters could not be sourced given the time limitations. This presents the largest data gap in the study. Assumptions have been made based on total weight and what are assumed to be equivalent processes from available data, as detailed in Section 4. This, naturally, induces a high level of uncertainty for these two sub models and was considered to be a significant limitation of this study. However, the sensitivity analysis (detailed in Section 7.5.1) shows that the results are not significantly affected by the uncertainties in these two sub-models.

The end-of-life scenario assumes a recovery rate of 100 percent for the mechanical materials (Louw, 2009). It is assumed that these materials are recycled in a closed-loop process and offset 90 percent of the virgin material input. The remaining 10 percent is assumed to be lost from the system during the recovery and recycling process. No attempt has been made to account for this material as emissions. A sensitivity analysis has been undertaken to explore the effects of these recycling assumptions on the result, see Section 7.5.2.

The capacity and usage calculation for the generic cable system (i.e. how many gigabits are sent annually) is based on 11 systems built between 2000 and 2006 (detailed in Section 6). This places a limitation on the applicability of the results to today’s 10Gbps technology. However, terminal station technology is rapidly developing and commercialisation of 40Gbps transmission is fast approaching (Ishida et al , 2007). In theory, this will reduce the environmental impacts per unit of data further as a greater amount of data can be sent down the same cable. Future scenarios based on greater capacity can be modelled by scaling of the results relative to the capacity calculation presented in Section 6.

Validation of the result against an actual system, is not possible in this case, due to the complexity and geographical dimension of a submarine cable system.

When interpreting the results, it is important to take into consideration that the analysis has been undertaken given a 13 year commercial lifetime of the cable (Ridder, 2007). It is shown that the technical lifetime of the cable is significantly longer, at 25 years (Beaufils, 2000). This has been addressed in the sensitivity analysis, where a scenario based on the average technical lifetime is explored.

3. Goal and Scope A life cycle assessment o f fibre optic submarine cable systems

3.2.9. Critical Review Procedure Critical review of the study is necessary in order to verify that the “LCA has met the requirements for methodology, data, interpretation and reporting and whether it is consistent with the principles” of LCA (ISO 14040:2006, p17). As defined in the ISO standard ( ibid ), both internal and external experts with appropriate scientific and technical expertise have supervised this study and provided review and guidance respectively, Fredrik Jonsson of Ericsson and Åsa Moberg of The Royal Institute of Technology (KTH). Further, a review of the report structure has been undertaken by a panel of interested parties (ISO 14040:2006) during the presentation and defence of this study at KTH.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

4. Life Cycle Inventory (LCI)

This section describes the system and the process of data collection and calculation. Furthermore, a detailed description is given of the structure of the core sub-models for the life cycle inventory.

4.1. Description of System This study is undertaken on a generic fibre optic submarine cable network, where the system boundary is drawn at the limits of the two end terminal stations to the point where the signal is transferred over to the terrestrial network. This includes all cable and repeaters used for the subsea link between terminals and all transmission and receive components within the terminal station itself. Each component is further explained in the theoretical framework, Section 2.2.4.

The LCA model has been assembled within the GaBi software. The model has been constructed to account for all significant material and energy inputs and subsequent emissions from the system. Two main sub-models have been developed: one based on the submarine cable and associated repeaters; the other, on the land terminal station. This gives the model greater versatility in assessing the potential environmental impacts of a particular repeated submarine cable system as the model can be scaled by the length of cable and the number of terminal stations. For this study, a definition of the generic cable system has been established based on the literature and is presented in Section 6.

Using a cradle-to-grave approach, five phases have been identified over the life cycle of the system; raw material extraction , design & manufacturing , installation , use & maintenance and end-of-life decommissioning . The five phases are represented by LCI sub-models defining each progressive life cycle process and are further divided into cable and terminal station processes. The basic structure of the life cycle model is shown in Figure 13, with a detailed system flowchart presented in Appendix D.

RAW MATERIAL DESIGN & INSTALLATION USE & END -OF -LIFE EXTRACTION MANUFACTURING MAINTENANCE DECOMMISSIONING

CABLE CABLE CABLE CABLE CABLE LW cable Route Survey Cable ship modes Cable ship modes Recovery by ship LWP cable LW cable Transit Transit Material Recycling SA cable LWP cable Manoeuvring Manoeuvring Cable DA cable SA cable In port In port Repeater Subsea repeater DA cable Cable energy use Subsea repeater

TERMINAL TERMINAL TERMINAL TERMINAL TERMINAL PFE Desktop Study No process Energy use Material recycling SLTE PFE Lead Acid battery Lead acid battery SLTE Printed board Back-up generator Lead acid battery assembly (PBA) Back-up generator

Figure 13: Basic structure of the life cycle of a submarine cable system.

The raw material extraction phase defines the materials required for the cable, repeaters and terminal equipment. Further sub-sets model each type of cable, a submarine repeater and the terminal components in more detail. Transportation of the raw materials is accounted for in the sub-models.

The design & manufacturing phase defines the processing of raw materials into the final products ready for installation. Manufacturing of the cable, a submarine repeater and the terminal components are

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

modelled. Consumed energy and wastes from the manufacturing process are accounted for in the sub- models.

The installation phase accounts for the at-sea installation of the cable by purpose-built cable ship. Repeaters are installed with the cable and are accounted for in the cable sub-model. Terminal station installation is assumed to have a minor impact on the result and is not modelled.

The use & maintenance phase accounts for the energy consumed by the terminal station and the cable itself, and, the at-sea maintenance of the cable by cable ship. This phase represents the significant process of the life cycle with respect to time. Given the 13 year use & maintenance period, this phase is assumed to have the greatest impact on the results.

The end-of-life phase is modelled on a scenario of cable recovery and material recycling. Cable recovery is undertaken by cable ship and is assumed to be the reverse process of installation, therefore, utilising the same resources. Cable material recycling is based on the main mechanical materials in a closed-loop recycling process. The recycling processes for each material are simplified models based on the documented energy consumption during material reprocessing. Recycling of the terminal materials has been considered for the lead-acid batteries and the printed board assembly (PBA). No other end-of-life processes are modelled for the terminal station.

4.2. Data Calculation All collected data has been input into the GaBi software to enable the relative scaling of each process within the sub-models and in relation to the reference flow of the functional unit. “The calculation result is a set of linked and scaled processes” with resulting scaled environmental impacts (Guinée et al , 2004, p.60).

4.3. Data Collection Process Data has been collected from a variety of sources within the scientific community and companies linked to the cable industry.

The theoretical framework has been established from the ISO 14040 series guidelines for LCA, published journal articles, books and reports. Articles from scientific journals were searched through the online KTH library portal. Databases such as, Highwire Press, Science Direct, Scopus, SpringerLink, Web of Science and Wiley InterScience , were searched firsthand, with a more general search of the internet through Google and Google Scholar . Articles and reports collected through the general internet search have been validated by checking the author’s credentials and the source of the data. The literature is presented in two categories; the methodology of LCA and the description of a submarine cable system. For the LCA methodology, preference was given to the ISO standard (ISO 14040:2006), peer reviewed journal articles and LCA handbooks such as Baumann and Tillman (2004) and Guinée et al ( 2004). For submarine cable systems, preference was given to journal articles, however these were fewer and industry reports and company websites provided additional valuable information.

The goal and scope definition is based on the ISO 14040 series guidelines and LCA handbooks.

System modelling and identification of data requirements was based on the theoretical framework and 12 years of professional experience working with submarine cable systems. While professional experience can be considered subjective, it provided a starting point for the literature search and many leads through contacts within the industry. A stakeholder analysis was undertaken to identify possible contacts within the industry, attached in Appendix C.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

Data for the raw materials and manufacture of the cable was collected from Ericsson Network Technologies (Ericsson Cables) and their suppliers. A site visit was made to Ericsson’s Hudiksvall cable manufacturing plant in order to gain a clearer understanding of the cable production process. A meeting was held with four Ericsson representatives there, leading to email contact with their suppliers. A questionnaire was sent out to these suppliers with the aim to collect data relating to the material and energy consumption, transportation and wastes from their products. The questionnaire did not cover specific site emissions. All suppliers were given the same questionnaire containing semi-structured questions, which resulted in open answers and statistical data. The questionnaire is attached in Appendix E.

Data collection for terminal equipment and submarine repeaters has proven to be more problematic and has resulted in a data gap. The four known manufactures of these components were contacted and invited to participate however, no company was willing to disclose commercially sensitive proprietary information. As such, estimates of these processes, based on similar processes from the literature, have been necessary to fill the data gap. Data from previous LCA studies at Ericsson, published articles and other industry reports have been used to fill the data gaps and approximate the terminal and repeater processes. The effect of this substitution is explored in the sensitivity analysis in Section 7.5.1. Standard database processes supplied with the GaBi software have been used to model the raw material processing, energy production, transportation and waste handling. Additional information on system processes has been gained by email and phone conferences with professional contacts within the cable industry. Confidential proprietary operation reports have been used to estimate activities for the cable design, installation and use & maintenance phases. The details of these reports are not presented in this study, however this does not affect the presentation of the final result. Published articles and reports have provided further data, particularly in the area of emissions from ships and material recycling.

A site visit was made to a cable terminal station in Auckland, New Zealand. A walk around and open discussion with two Telecom New Zealand representatives provided a greater understanding of the components that make up a terminal station. Notes were taken during this session which were later verified against the literature and used to develop the terminal station sub-model.

This study was also partially completed at sea during a 4800 kilometre cable route survey. Data collected during this period relates to the design sub-model, and includes data on Research vessel operations, fuel use and cable routing. This data is also considered confidential and full details are not presented in this report. Again, this does not affect the presentation of the final result.

Characterisation factors and impact categories based on the CML methodology for LCIA (base year 2001) have been used directly from the GaBi software database.

For details of the data used in this study, refer to Section 4.4, Description of Core Unit Operations and LCI Sub- Models and Appendix B.

The data collection period was approximately 5 months.

4.4. Description of Core Unit Operations and LCI Sub-Models This section defines of the structure of the core sub-models for the life cycle inventory (LCI) of the submarine cable system. Modelling for energy production, transportation, raw material extraction, design & manufacturing, installation, use & maintenance and end-of-life decommissioning are described in detail.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

4.4.1. Energy

4.4.1.1. Electricity Generation

The vast majority of cables connect the US, Europe, Japan and China, see Figure 3 – World map of submarine cables . Based on the stakeholder analysis, cable and component manufacturing plants are also shown to be predominantly located in these regions. The environmental impact of electricity generation has therefore been calculated from an equal mixture of production from the US, EU, Japan and China (25% per country). It is assumed that this provides an average mix relevant to cable manufacture and utilization. The impacts of electricity generation for each country have been adopted from two previous studies at Ericsson. Chinese production has been model by Bergelin (2008) from the base years of 2003 to

2006. Figures for CO 2, SO 2, NO X and PM 10 (dust) were calculated from published data relating to the environmental load cost in US dollars per kilowatt hour (kWh). This was converted to actual emissions per kWh based on total environmental cost and total electricity production. All other emissions for electricity production in China were estimated from published US results, having the second highest electricity from coal production at 51 percent compared to China at 80 percent. Some uncertainty exists in this data. Further details of the method of calculation, assumptions and simplifications, can be found in Bergelin (2008). The impacts of production in the US, EU and Japan were modelled during an extensive LCA into the impacts of a Third Generation (3G) mobile telephone system undertaken at Ericsson. Included in the electricity model was generation from fossil fuels, nuclear fuel and hydro plants. Furthermore, the whole supply and distribution chain was considered, including, distribution to industrial consumers, production and transport of fuels, materials consumption and construction of generation and distribution facilities. Generic production models were adapted to local conditions by applying each region’s statistical fuel base for percentage production from coal, gas, oil, nuclear and hydro, using base years between 1994 and 1999, as shown in Table 3 (3GLCA, 2002). The effect on the results of the uncertainty in the electricity data as a whole is explored in the sensitivity analysis in Section 7.5.2.

Table 3: Distribution of electricity generation processes (Adapted from 3GLCA, 2002; Bergelin, 2008)

Region Total Generation TWh/year Coal % Gas % Oil % Nuclear % Hydro % Other

China (2003) 1580 80 U7 U8 U8 U8 U8 Europe (1994) 1605 28 10 10 37 14 1 Japan (1994) 1027 15 26 10 36 11 2 US average (1999) 3728 51 16 3 20 8 2

Other sources of generation, such as, geothermal, biomass, wind and solar were not modelled. Further details of the method of calculation, assumptions and simplifications, can be found in (3GLCA, 2002). No attempt has been made in this study to update these figures.

In determining the energy count, the GaBi software does not distinguish between primary energy and secondary energy, such as electricity. Based on the previous work of Bergelin (2008), a parameter was used to count all electricity (secondary energy) used within each sub-model. Consequently, for each unit of secondary energy, the primary energy needed to produce it, can then be analysed. Based on the electricity mix used for this study, a total of 13.7megajoules (MJ) primary energy is needed to produce 3.6MJ of electricity, giving a secondary to primary energy factor of 3.81. Electricity used for the standard database processes within the GaBi software cannot be extracted from primary energy.

The reference flow for the electricity sub-model is 1 kWh.

7 U = unknown, though it is assumed that the majority of the remaining 20% comes from hydro. (Bergelin, 2008)

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

4.4.1.1. Heavy Fuel Oil

The environmental impact from the production of heavy fuel oil (HFO) has been modelled using standard database processes available in the GaBi software. A 50/50 mix of United States (US) production and European Community (15 countries - EU-15) was constructed to provide an approximate world mix of HFO at the refinery. Though, it must be kept in mind that this sub-model is biased toward US and EU-15 production. Data from other countries was not available. Transportation from the refinery to the port where bunkering may take place has not been considered. It is assumed that the HFO sub-model represents the production of residual oil (RO) fuels used for marine transportation, described further in Section 4.4.2.1.

The reference flow for the HFO sub-model is 1kg.

4.4.2. Transportation The transportation sector is a major contributor to pollution and a significant fact in anthropogenic climate impacts. Fossil fuel combustion still dominates the freight transportation sector. Heavy-duty truck, rail and water transport combined account for over 25 percent of CO 2 emissions for all mobile emission sources in the US and over 30 percent in the EU (Corbett and Winebrake, 2007). In this study, it is assumed that transportation of the raw materials or products used for cable manufacture and the wastes and materials for recycling is undertaken overland by long-distance 34-40t truck and trailer. Transportation of the manufactured cable is always undertaken by sea route due to the sheer weight and bulk of a completed cable. Personnel transfers requiring international air travel to and from the ships, has been modelled to evaluate the effects from aviation in relation to the total result. No rail transportation has been modelled in this study.

4.4.2.1. Marine Transportation

Ships, naturally, play a key role in the design, installation and maintenance of submarine cable systems during their life cycle and, as such, much focus has been placed on the methodology for estimating emissions from ships. Commercial shipping makes a significant contribution to global air pollution (Lack et al , 2009). Emissions from ships engines are among the highest polluting sources per ton of fuel (Corbett and Koehler, 2003). It is estimated that 2.7% of anthropogenic CO 2 emissions in 2000 were from shipping, contributing significantly to total emissions from the transport sector (Eyring et al , 2007). Studies show that air pollutants from ships are related to three factors; engine type, engine loads and fuel type (CARB, 2008; Cooper and Gustafsson, 2004; Corbett and Koehler, 2003, European Commission, 2002). Installed engine power is the important factor for estimating emissions, rather than the number of engines (Corbett and Koehler, 2003). Generally, ships have a main engine (ME) used for propulsion and auxiliary engines (AE) used for electricity generation. MEs and AEs can be further classified based on their speed measured at the crankshaft, another factor affecting emissions. Emissions from boilers, emergency engines and other equipment are small in comparison and can generally be excluded. Fuel types, classified by their viscosity, range from light marine distillates (MDs) to heavier residual oils (ROs). Some emissions are directly related to the content within the fuel, particularly sulphur. (Cooper and Gustafsson, 2004). While other emissions such as nitrogen oxides (NOx) are related to the type of combustion system (Corbett and Fischbeck, 1997). MDs are considered to have fewer impacts both from an environmental and economic standpoint (Fet, 2009). Three operational modes are identified, each having a differing effect on the engine load and the combustion efficiency of the fuel; transit , manoeuvring and in port . ME emissions occur during transit and, to a lesser extent, manoeuvring and AE emissions during all modes. At cruise speed, the ME load is estimated at 82.5 percent. Vessels generally do not cruise beyond this load rating as fuel consumption and maintenance increase significantly (CARB, 2008; Cooper and Gustafsson, 2004). Manoeuvring load factors are estimated at lower and varying loads. Auxiliary engine load factors are

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

estimated at 50 percent. Manoeuvring and in port emissions are relatively small in relation to transit emissions, though equally important as they occur close to populated areas. Emissions have been based on the Cooper and Gustafsson (2004) study, as a larger number of pollutants are accounted for when compared to similar studies. They conclude that, at the time (base year 2002), the emission factors presented were an “up to date and “best possible” estimate” for Swedish sold fuel (Cooper and Gustafsson, 2004, p32). Many of the emissions factors have high uncertainty ratings based on the assumptions and availability of data. Emission factors are tabulated in Cooper and Gustafsson (2004). It is assumed that emissions from Swedish fuel are representative of emissions from other developed nations. This assumption has been verified against two other studies into shipping emissions by the European Commission (2002) and the California Air Resources Board (CARB, 2006), with good correlation. It should be noted that each study uses the same methodology for estimating emissions with similar results given the temporal and regional differences in the inventory datasets; California - base year 2006 (CARB, 2006) and Europe – base year 2000 (European Commission, 2002). The latter was regarded “as the most comprehensive published test data for commercial marine vessels.” (Corbett and Koehler, 2003, p9-2). Further verification was undertaken by Corbett (2004), concluding that good correlation exists between estimated values and stack test results, with in-plume observations having greater uncertainty due to complex chemical processes. Furthermore, the emission factors calculated by Cooper and Gustafsson (2004) show good correlation to actual emission rates (single vessel observation) and the Lloyd’s Marine Emissions Research Programme test data (Corbett, 2004).

As nitrogen oxide (NO X) reduction techniques are only common in newer vessels (Cooper and Gustafsson, 2004), reduced emission factors have not been considered. Table 4 presents emission factors for the most significant pollutants from marine diesel engines, full emission factors are tabulated in Cooper and Gustafsson (2004). Emission factors for residual oil (RO) and slow speed diesel engines have been used in this study as they represent the majority of fuel used and engine type in commercial shipping (Corbett and Fischbeck, 1997). It should be noted that RO has exceptionally high sulphur and particle matter (PM) emissions (Corbett and Winebrake, 2007). The process for the production of heavy fuel oil (HFO) was used from the GaBi software, which is assumed to represent the production process of RO fuel.

Table 4: Selected engine emission factors (g/kWh) for Residual Oils (Cooper and Gustafsson, 2004).

Engine Operation Engine Type Fuel NOx CO SOx PM PM CO CH Speed Mode 10 2.5 2 4 Main Engine Slow At Sea RO 18.1 0.5 9 1.3 1.3 620 0.006

Main Engine Slow Manoeuvring RO 14.5 1.0 9.9 2.6 2.6 682 0.012

Aux. Engine Medium In Port RO 14.5 0.9 10.4 0.5 0.5 722 0.004

Manoeuvring loads are assumed to compare with the operational phase of cable route survey and installation. Main engine manoeuvring loads are assumed by Cooper and Gustafsson (2004) at 20 percent. This is likely to be an under estimation, as the ship’s engines are likely to be under greater load. However, review of the emission factors shows that the emissions are greater for manoeuvring, in relation to “at sea” emissions, in all categories except NO X emissions with are reduced by 20 percent. Therefore, it is assumed that this does not affect the results significantly. Table 5 shows the load characteristics during each operation mode.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

Table 5: Average Cable Ship engine load characteristics by operational mode. (Cooper and Gustafsson, 2004)

Operation Mode Transit Manoeuvring In Port

Main Engines 80% 20% 0%

Auxiliary Engines 50% 50% 50%

The construction of ships is not included in this study. It was intended to expand the system boundaries to include the construction of the purpose-built ships used to install and maintain the cables. While there have been many studies relating to the construction of ships, specific data could not be sourced. However, previous studies in the area indicate that the use phase totally dominates the environmental impacts due to the emissions created from fuel combustion (Ellingsen et al , 2002; Fet, 2003; Fet et al , 1996; Lingg and Villiger, 2002). The construction phase of a ship’s life cycle has been estimated to have one hundredth (100 th ) of the impact of the use phase; therefore, limiting the study to the operating phase is considered sufficient (Fet et al , 2000). Steel scrapping and recycling at end-of-life has been shown to compensate for over 80 percent of virgin material input and reduce other emissions (Fet, 2003). Although, Ms. Fet did advise in personal communication that this was a rudimentary LCA study with inherent uncertainty.

Local and regional impacts result from the maintenance of ships, where hulls are typically sand blasted and repainted with antifouling. Antifouling coatings are the paint used on the hull of the ship to inhibit marine growth and hence, release a number of pollutants to seawater. An estimation of the impacts due to hull maintenance activities has been calculated during one LCA study completed for a Platform Supply Vessel servicing the oil fields between Norway and the United Kingdom (Fet et al , 1996). It is assumed that this vessel type is comparable to the typical cable ship. Though, uncertainty is high given that this is a single case study, some years old and based on regional practice (Norway).

4.4.2.2. Road Transportation

Transportation of the cable materials is undertaken in two stages, firstly, raw material transportation to the intermediate processing factories, secondly, transportation of the intermediate products to the cable manufacturing plant. Cable manufacture has been modelled based on the processes at Ericsson cables in Sweden. Data relating to transportation distances for materials and intermediate products have been collected from Ericsson cables and their suppliers. Based on this data and that the intermediate materials used in cable manufacture are generally specialised items made by few suppliers, a distance of 1,000km is assumed for all materials road transportation. It is assumed that this distance is relevant for all cable manufacturers irrespective of geographic location. For waste treatment and the recycling scenario, road transportation has been estimated at 100km. Louw (2009) advises that materials for recycling are transported over a radius of 30 kilometres. It is there assumed that a 100 kilometre radius for transportation of waste and recycling materials is a reasonable estimate. Road transportation processes, modelling production (diesel at refinery) and combustion of fuel (truck by gross tonnage), have been used from the GaBi4 database. Other elements, such as, road infrastructure, vehicle manufacture, vehicle maintenance and fuel handling, have not been included. A sensitivity analysis has been undertaken to determine the effect on the final result of both road transportation assumptions and is presented in Section 7.5.1.

4.4.2.3. Air Travel

“Personnel-related environmental impact is usually not included in LCA” (Baumann and Tillman, 2004, p82). Exclusion may be justified if the impacts are considered small, however, transportation of personnel to and from the workplace can be significant (USEPA, 2006). Air travel is considered likely to have a

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

noticeable environmental impact , as personnel transportation involves substantial internatio nal air travel to and from the ships. Therefore , a ir travel is calculated based on the input and emissions for one person kilometre travelled (1pkm) determined from previous studies at Ericsson . Th e Ericsson study shows that, 1pkm uses 1.6MJ of primary ene rgy and, among other emissions, releases 120 grams of carbon dioxide into the atmosphere (3GLCA, 2002 : LCA database).

4.4.3. Raw Material Extraction Phase The sub-model for the extraction of raw materials begins with the extraction of the ma terials from the environment, includes the proce ssing of these materials into their respective products (such as aluminium ingots) and ends with the transportation of the materials to the assembly site. The sub -model is further divided into the three main components of a s ubmarine cable system; the cable, submarine repeaters and the terminal station. Database processes from the GaBi software have been used to model the extraction and processing of raw materials.

4.4.3.1. Cable Raw Materials

Data for the cable raw materials has been collected from Ericsson Network Technologies (Ericsson Cables) and calculated from other confidential cable specifications . A full list of data sources can be found in Appendix B.

The four primary cable types were chosen to represent the generic system; Lightweight (LW), Lightweight Protected (LWP), Single Armour (SA) and Double Armour (DA), as shown in Figure 9. Cable systems use a variety of these cable types with a corresponding variance in material use directly tied to the amount of steel wire armour protection provided. The link between steel armouring and total cable weight is shown in Figure 14.

9000 8000 7000 6000 5000 4000 Steel 3000 kg per1000m kg Total 2000 1000 0 LW LWP SA DA Cable Type

Figure 14: Total weight verses weight of steel per 1,000 metres of cable .

In order to normalise the use of each cable type to 1 ,000 kilometres of cable, four submarine systems representing a total of over 40,00 0km of cable, were analysed and the ratio of each cable type determined (shown in Table 6). This ratio was then applied to the raw materials sub-model for each cable type and combined to give the resulting raw material usage for 1,000 kilometres.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

Table 6: Ratio of cable types in the generic system.

LW LWP SA DA

70% 14% 13% 3%

All significant materials used in the cable have been accounted for and are represented by proc esses within the GaBi software. Figure 15 shows the distribution of all other m aterials (excluding steel) by cable type, which illustrates how the cable armouring is built around the core of the fibre unit structure (stainless steel tube filled with gel and fibre), the surrounding copper conductor and the high -density polyethylene insulating plastic. It also highlights the dominance of the main mechanical materials.

500,00

450,00

400,00

350,00

300,00

250,00 LW LWP

kg per1000m kg 200,00 SA 150,00 DA 100,00

50,00

0,00

LW: Lightweight, LWP: Lightweight Protected, SA : Single Armour, DA: Double Armour Figure 15: Distribution of all other raw materials (excluding steel) by cable type.

No data could be sourced for the production of high quality optical fibre. Purifying and drawing the fibre uses significant amounts of energy ( Corning, 2008). The GaBi database process for glass fibres (for composites materials) was used to provide an approximation of the process. It is assumed that this does not affect the results greatly as fibre represents 0.3% of the total weight of the lightweight cable. All other processes and materials are well defined in the GaBi software database.

The reference flow for the sub-model is 1,000km of cable.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

4.4.3.2. Submarine Repeater: Raw materials

Specific data for the submarine repeaters could not be sourced, therefore assumptions are necessary. Data for the repeater raw materials has been defined by the weight of the housing with the remaining materials assumed from similar processes based on previous studies at Ericsson Research. The total weight of the repeater itself is given as 255 kilograms, including cable terminations, the total weight is 459 kilograms and the weight of the housing alone is given at 170 kilograms (confidential source). It is assumed that the repeater housing is made from beryllium copper (BeCu) alloy. No process for the extraction of beryllium was available in the standard databases, however, BeCu alloys used for submarine repeaters contain typically 98% copper, 1.7% beryllium an 0.3% cobalt (Davis, 2001). Therefore for this study, it is assumed that the housing is made of 100 percent copper.

The remaining internal components and cable terminal materials have been estimated based on an apportioned weight distribution. Previous studies at Ericsson have been used to model these components. The Ericsson 3G mechanical enclosure study (MECH, 2001) was used to provide the reference for the mechanical materials. The printed board assembly (PBA) model constructed during the Ericsson LCA study into a 3G mobile network has been used to estimate the electronic components within the repeater (3GLCA, 2002; PBA,2001). The total weight of the internal components is estimated to be 85 kilograms. Based on experience at Ericsson, the PBA weight to mechanical weight ratio of the sub-racks has been estimated at 10 percent (Malmodin, 2009), thereby resulting in a PBA weight of 8.5 kilograms. The mechanical enclosure model has been used to estimate the processes for the remaining 280.5 kilograms. Naturally, these assumptions introduce significant uncertainty for this sub-model.

The reference flow for the sub-model is 1 repeater unit. It is assumed that repeaters are placed every 50km, requiring 20 units per 1,000km of cable.

4.4.3.3. Terminal Equipment: Raw materials

Specific data for terminal station equipment could not be sourced, therefore assumptions are necessary. Data for the terminal raw materials has been defined by confidential terminal station specifications (year 2000) and assumed from similar processes based on previous studies at Ericsson Research. Four sub- models were created to account for the back-up batteries, the power feed equipment (PFE), the submarine line terminal equipment (SLTE) and the back-up generators. A full list of data sources can be found in Appendix B.

Figure 16 shows the distribution of components by weight in the terminal station.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

9000 8000 7000 6000 5000 kg 4000 3000 2000 1000 0 Lead -Acid PFE SLTE Generator

PFE: Power Feed Equipment, SLTE: Submarine Line Terminal Equipment Figure 16: Distribution of raw materials in the terminal station.

The lead acid battery materials were provided by a previous Ericsson cradle-to -gate study for the manufacturing of a two volt (2V) cell used in a radio base station (Pb-Battery, 2001). Terminal stations use lead acid batteries to supply approximately 50V DC power in the event of a power failure. Four battery banks were observed during a visit to one terminal station. It is assumed that this is commo n for terminal stations. The reference flow for the sub -model is a 2V cell, thereby utilising 100 2V cells.

The power feed equipment ( PFE ) specification was determined from both confidential sources and published sources (Alcatel, 1998) , and is assembled from LCA studies of similar products undertak en by Ericsson and ABB (ABB, 2009a; ABB, 2009b; ABB , 2009c) . The PFE comprises direct current (DC) converters, monitoring equipment and mechanical enclosures (cabinets). Terminals have two parallel systems to provide protection against failure in any one system (Alcatel, 1998). Based on the literature, a total of 20 kilowatt DC converters and 4 monitoring modules were estimated. This allows for two 10kV PFE systems operating at approximately one ampere (1A) . Material specification for the cabinets has been taken from a previous Ericsson cradle -to-gate study for the mechanical enclosure for a third generation (3G) mobile system. It was determined that 97 percent of the materials were accounted for in the study (MECH, 2001). It is assumed that the materials used in the 3G cabinets are similar to those used in the terminal station. A total of five cabinets each wit h a weight of 80kg have been estimated . The weight of the cabinet was comparable in Ericsson documentation at 75kg (Char acteristics Specs , 2009). The reference flows for the sub-model are; 1 kW equivalent per DC converter (total 20kW), one monitoring unit (total 4) and weight of mechanical enclosure (total 400 kg).

The submarine line terminal equipment ( SLTE) equipment inclu des the various components needed to generate and manage the light wave signals. Materials for the SLTE equipment are based on the specification of a single terminal station and the floor plan showing the layout of a second. A diagrammatic layout of the SLTE is shown in Figure 17.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

SDH : Synchronous Digital Hierarchy

HPOE: High Precision Optical Equipment

LME: Line Monitoring Equipment

WTE: Wavelength Terminating Equipment

TLA: Terminal Line Amplifier

Figure 17: Diagram of Submarine Line Terminal Equipment (SLTE) (Adapted from Markow, 2009).

Weights for each cabinet and sub-rack where provided in the terminal documentation, however no detailed data was available for the actual electronic boards themselves. Again, previous studies at Ericsson have been used to model these components. The Ericsson 3G mechanical enclosure study (MECH, 2001) was used to provide the reference for the mechanical materials. The printed board assembly (PBA) model constructed during the Ericsson LCA study into a 3G mobile network has been used to estimate the electronic components within the sub-racks (3GLCA, 2002; PBA,2001). Based on experience at Ericsson, the PBA weight to mechanical weight ratio of the sub-racks has been estimated at 10 percent (Malmodin, 2009). A total of four fibre-pairs with 16 light wave signals have been assumed for the generic system. This provides a total capacity of 64 wavelengths and equates, based on the above assumptions, to 30 cabinets with 79 sub-racks. Estimated weights are presented in Table 7. The abbreviations are explained in Section 2.2.4.3.

Table 7: Estimated weights of terminal components.

SDH HOPE WTE TLA LME Component Total (kg) (kg) (kg) (kg) (kg)

Cabinets (80kg unit) 320 1280 640 80 80 30 units

Sub-Rack Equipment 224 3328 448 74 56 79 units

Total Mechanical Sub-Racks (90%) 202 2995 403 67 50

Total PBA Equipment (10% of Sub-Rack) 22 333 45 7 6 413kg

Total Mechanical (Cabinets + Sub-Racks) 522 4275 1043 147 130 6117kg

The PBA components of the sub-racks calculate at 413kg and the mechanical materials at 6177kg, giving 6.8% PBA weight to total mechanical weight. These figures are used to construct the model for the terminal station SLTE. However, they come with high uncertainty and are likely an over-estimation, due to the uncertainty of the weight of the PBAs and the age of the terminal specification (taken from year 2000). System upgrading adds further uncertainty due to the refinement of electronics. Upgrading, in most cases, makes savings in both volume and mass. The results from an LCA of a personal computer (Williams and Sasaki, 2003) provide the estimate for the Network Management Equipment (NME). Upgrading must also be considered with regard to material inputs. Over the 13 year life cycle of cable system, it is estimated that 3 upgrades will undertaken (Betts, 2009; Veverka, 2009). In each case the capacity of the cable is increased by replacing only the electronic equipment of the terminal. Therefore,

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

the PBA and the NME inputs for have been scaled by a factor of three in the terminal model to provide a linear approximation of the upgrade process. This is a simplification for the purposes of this study and does not account for the gains in equipment efficiency or the temporal variation of each upgrade. The effect of this simplification on the results is explored in the data quality analysis is Section 7.5. The reference flows for the sub-model are; weight of the mechanical enclosure (total 6117kg), weight of the PBA (total 1239kg) and one PC unit for the NME (total 3 units).

Two back-up generators are provided to keep the cable powered during outages in the main electricity network. These have been estimated from an LCA study into an AC generator with a functional unit of 1kW (ABB, 2009d). During the site visit to one terminal station, it was advised that two 650kW generators were installed (McGrath, 2009). The reference flow for the sub-model is 1kW. The ABB study has been scaled up to 1300kW and it is assumed this is comparable to the installed generators of the common terminal station.

Construction materials for the terminal building have not been included in this study, see Section 4.4.4.4.

4.4.4. Design & Manufacturing Phase The design & manufacturing sub-model accounts for the energy used in processing the raw materials into the mechanical parts for the system. Similar to the raw materials sub-model, the manufacturing sub-model is further divided into the three main components of a submarine cable system; the cable, submarine repeaters and the terminal station. Surveying of the cable route (design) has been included under the cable manufacturing process as the specific cable lengths must be defined prior to accurate manufacturing to length. Energy recovery from incinerated waste is included in the sub-models. Database processes from the GaBi software have been used to model the processing of wastes. The system includes energy use (electrical) for the recycling of waste metals, which is then assumed to form a closed-loop recycling process to reduce the initial virgin raw materials. This closed-loop recycling process is a simplification based on input electrical energy and is further discussed in Section 4.4.7. Transportation of waste is estimated at 100km as described in Section 4.4.2.2.

4.4.4.1. Cable Design (Route Surveying)

Prior to manufacturing, the cable route must be defined and surveyed by a research ship. The survey determines the seabed topography, geology and any other obstacles or dangers to the cable. This in turn affects the armouring selection for the cable and thus the manufactured lengths (Beaufils, 2000; Letellier, 2004). The average engine power rating for a research ship is estimated from two vessels working specifically with cable route surveys (GMSL, 2009; confidential source). Fuel figures are taken from a single survey where the author was present and averaged over 71 days of operations. It was not possible to distinguish between transit and manoeuvring. A review of the emission factors shows that in most cases manoeuvring emissions are slightly greater than at sea or transit emissions, thereby allowing for a worst case scenario. The emission factors associated with the combustion of fuel onboard ships are further defined in Section 4.4.2.1. Average research ship engine power and daily fuel consumption are presented in Table 8 and Table 9.

Table 8: Engine Power: Average Research Ship (calculated from GMSL, 2009; confidential source).

Main Engine Power Auxiliary Engine Power Total Engine Power (kW) (kW) (kW)

2311 650 3261

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

Table 9: Daily Fuel Consumption: Average Research Ship (calculated from confidential source).

In Port Manoeuvring (tonne) (tonne)

1.02 5.03

The typical mission length to survey the cable route has been calculated from a survey (where the author was present) of 4800km of cable route, lasting 71 days. The figures are normalised to 1,000km of cable. The operations included both shallow water (<1,000m) and deepwater (>1,000m) surveys and are assumed to be representative of the average cable survey. Typical mission lengths are presented in Table 10 below.

Table 10: Average Research Ship: typical survey mission normalised to 1,000km of cable (calculated from confidential source).

In Port Manoeuvring Operation Mode (days) (days)

Route Survey 2.7 12.1

Based on the normalised survey mission, it is calculated that a total of 64 tonnes of fuel are used to survey 1,000km of cable.

Air travel to and from the ship has been estimated based on 35 personnel travelling an average of 2559km per 1,000 kilometres of cable, giving a total of 89574pkm normalised to 1,000km of cable route survey. Air travel is further described in Section 4.4.2.3.

The daily maintenance of the research ship has been calculated using the LCA study of Fet et al (1996) described in Section 4.4.2.1. The surface area of the hull was estimated based on the length, breadth and draft of the studied vessel in relation to the research vessel. A factor of 0.46 was calculated and applied to the results of the study to account for the difference in ship size. The results of the study represent 10 years of total maintenance and were reduced to a daily figure. The paint itself is estimated from a process for enamel paint, assumed to represent a similar process to the antifouling paint used for ships (Chalmers, 2009).

The reference flow for the sub-model is 1,000km of cable.

4.4.4.2. Cable Manufacturing

Data for cable manufacturing was collected from the Ericsson Network Technologies (formerly Ericsson Cables) production plant. The manufacturing process has been simplified to the effect rating of the plastic extrusion and armouring stations (Norlund, 2009). These are the two significant processes undertaken during cable manufacture, though it is likely to underestimate the manufacturing energy demand as it does not account for the plant operation as a whole. This simplification was made as the Hudiksvall plant manufactures many other types of fibre optic cables. No allocation was attempted. It is assumed that all input energy is in the form of electricity.

Waste from the production process is estimated at 1.5% ( ibid ) and is sent to scrap handlers for recycling (Berggren, 2009). It is assumed that the cable is separated and that steel is the only recycled material, with all other materials incinerated for energy recovery, providing between 0.8% to 1.6% return into the system. A simplified closed-loop recycling process is included in the sub-model for the recycling of steel,

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

which is assumed to reduce the input of virgin raw materials. This closed-loop recycling process is further discussed in Section 4.4.7.

Table 6 shows the ratio of the four primary cable types representing the generic system. The cable types of four submarine systems, representing a total of over 40,000km of cable, were analysed and the results normalised to 1,000km of cable. These values are applied to the manufacturing sub-model for each cable type to give the resulting energy consumption and waste output for 1,000km of cable.

The reference flow for the sub-model is 1,000km of cable.

4.4.4.3. Submarine Repeater Manufacturing

No data could be sourced for the manufacture of submarine repeaters. Four manufacturers were contacted, however, for confidentiality reasons, data could not be released. A repeater is made up of a heavy outer metallic casing with internal electronics that boost the light signal. Due to the lack of data, the assumption is made that the manufacture of a repeater is comparable by weight to that of the mechanical enclosure and PBA sub-models assumed for the raw materials phase (3GLCA, 2002). This requires approximately 50,000MJ of primary energy per repeater, where 75 percent of this energy is consumed for the PBA process representing less than 2 percent of the weight. This assumption adds significant uncertainty to this sub-model, as such a sensitivity analysis has been undertaken to determine the effect on the results. Refer to Section 7.5.1 for the results of the sensitivity analysis.

The reference flow for the sub-model is 1 repeater unit. It is assumed that repeaters are placed every 50km, requiring 20 units per 1,000km of cable.

4.4.4.4. Terminal Equipment Manufacturing

Specific data for terminal station equipment could not be sourced in the given time limitation, therefore, the manufacturing model is based on similar assumptions made for the raw materials and detailed in Section 4.4.3.3. Data for the terminal raw materials have been defined by confidential terminal station specifications and assumed from similar processes based on previous studies at Ericsson Research. Four sub-models were created to account for the manufacturing of the raw materials into back-up batteries, the power feed equipment (PFE), the submarine line terminal equipment (SLTE) and the back-up generators. A fifth sub-model was created to account for the desktop study, which involves personnel visits to each terminal site (Poole, 2009). A full list of data sources can be found in Appendix B. All electrical energy is assumed to come from the average mix as described in Section 4.4.1.

A previous Ericsson cradle-to-gate study defined the electrical energy input and the waste outputs for the manufacturing of a two volt (2V) cell, used in a radio base station (Pb-Battery, 2001). It is assumed that wastes are treated in a hazardous waste process. Waste lead is recycled back into the raw materials sub- system in an assumed closed-loop process. The recycling process has been included as a simplified energy input (electrical and heavy fuel oil energy) for the recycled lead based on the study by Salomone et al (2005). No emissions from the recycling process are included. The reference flow for the sub-model is a 2V cell. Four battery banks are assumed, utilising 100 2V cells.

The PFE comprises direct current (DC) converters, monitoring equipment and mechanical enclosures (cabinets). Manufacturing of the DC converters and monitoring equipment is defined in the ABB documentation which includes electricity inputs and waste outputs (ABB 2009a; ABB, 2009b). Manufacture for the cabinets has been taken from a previous Ericsson cradle-to-gate study for the mechanical enclosure for a third generation (3G) mobile system (MECH, 2001). It is assumed that these studies are representative of the common terminal station PFE. The PFE is described in more detail in

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

Section 4.4.3.3. It is assumed that all wastes are sent to landfill or treated if hazardous. These two processes are defined by standard database models. No specific time horizon is given for emissions from these processes, though it is assumed to be the 100 years surveyable period for landfill (Baumann and Tillman, 2004). The reference flows for the PFE sub-model are; 1 kW equivalent per DC converter (total 20kW), one monitoring unit (total 4) and weight of mechanical enclosure (total 400kg).

The manufacture of the SLTE equipment is based on previous LCA studies for similar products. The SLTE manufacture sub-model comprises the Ericsson 3G mechanical enclosure study (MECH, 2001), manufacturing of printed board assembly (PBA) model (3GLCA, 2002; PBA, 2001) and the results from an LCA of a personal computer (Williams and Sasaki, 2003) to model the network management equipment (NME) manufacture.

The mechanical enclosure sub-model accounts for the energy consumption and waste output for the manufacturing process. All incinerated wastes are assumed to generate recoverable energy, which provides approximately 0.6% electrical energy and 4% thermal energy return into the system.

The PBA components have been assembled from previous work at Ericsson (3GLCA, 2002). The study modelled the electrical components in a telecom node based on the construction of a standard board model, with separate sub-models for the important manufacturing processes related to integrated circuits (IC), printed circuit boards (PCB) and the PBA assembly process. IC manufacturing is important as it consumes large amounts of electricity. PCB manufacturing has potential for very large and hazardous emissions. A detailed description of these processes can be found in the Ericsson internal 3GLCA report (2002) and the public report by Bergelin (2008). With these four sub-models it was assumed that PBAs in all telecom nodes could be modelled (3GLCA, 2002). At the time of the study the component models were considered complete, however, uncertainty is introduced as it is most likely that there has been subsequent improvements in the manufacturing process (Bergelin, 2008).

Due to lack of data, the manufacturing process for the network management equipment (NME) is simplified to the energy required to produce a single PC, estimated by Williams and Sasaki (2003) to be 5600MJ. It is assumed that this energy is secondary energy in the form of electricity modelled on the average mix. No wastes or emissions are modelled for the NME, though it is assumed that their impact would not significantly affect the result as the NME primary energy expenditure is 1% of the total terminal primary energy use during manufacture. The PBA and NME sub-models have been scaled by a factor of three in the terminal model to account for upgrading of the system. The reference flows for the SLTE sub-model are; weight of the mechanical enclosure (total 6117kg), weight of the PBA (total 1239kg) and one PC unit for the NME (total 3 units).

The manufacturing of the back-up generators is estimated from the ABB study into an AC generator (ABB, 2009d). Energy consumption (electricity and thermal) and waste outputs are modelled for the manufacturing process. All wastes are assumed to be treated as hazardous or sent to landfill, again with an assumed 100 year time horizon. The reference flow for the sub-model is 1kW. Comparable to the raw materials sub-model, the ABB study has been scaled up to 1300kW (2 x 650kW) and it is assumed this is comparable to the installed generators of the common terminal station.

A desktop study (DTS) is undertaken prior to the manufacture of the cable in order to determine the preliminary cable route and location of the terminal stations. The DTS generally includes one site visit per cable landfall (Poole, 2009). The DTS is a simplified model based solely on air travel, which is assumed to have the most significant impact. It is assumed that 3 persons travel a 20,000pkm round trip per site visit, giving a total of 60,000pkm per terminal station. The air travel sub-model is further explained in Section 4.4.2.3. The reference flow for the DTS sub-model is 1 unit per terminal station.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

The construction of the terminal building has not been included in this study. LCA studies relating to the environmental impacts of buildings over their service life show that energy consumption during the use phase is equal to, or in the case of more recent studies, significantly dominates the environmental impacts of the building over the entire life-cycle (3GLCA, 2002; Guggemos and Horvath, 2005; Jönsson et al , 1998). Construction materials, energy and emissions can generally be excluded for capital equipment when the impact of construction is minor in relation to the total impact of the use phase (USEPA, 2006). Two cable terminal stations studied show a footprint size of approximately 270m 2 and 220m 2, with mixed construction of concrete and steel framing. Given the energy demand of the cable and the terminal equipment over its 13 year commercial lifetime, the impact from the construction of each terminal building is assumed to be relatively small.

4.4.5. Installation Phase

4.4.5.1. Cable Installation

The installation phase of the cable involves a special purpose cable ship designed to store and lay the cable from cable tanks within the ship’s hold. Due to the sheer weight and volume, 1,000km of lightweight cable weighs approximately 500 tonnes, the cable must be transported by sea. In most cases the cable is spooled onto the installation ship directly from the manufacturing plant. Review of confidential load and lay reports show that loading achieves an average of approximately 90km per day. The installation rate of the cable varies with the type of cable and the method of installation. The cable is generally ploughed into the seabed out to 1,000 metres of water and surface laid thereafter. Lay reports show that on average a distance of approximately 20km per day is achieved with ploughing and approximately 140km per day with surface lay. These figures are assumed to represent the typical cable installation.

The average engine power rating for a cable ship is estimated from the average installed power rating of seven vessels working specifically with cable installation and maintenance. Fuel figures are averaged over the fleet based on the vessel specifications (GMSL, 2009). The emission factors associated with the combustion of fuel onboard ships is further defined in Section 4.4.2.1. Average cable ship engine power and daily fuel consumption are presented in Table 11 and Table 12.

Table 11: Engine Power: Average Cable Ship (calculated from GMSL, 2009).

Main Engine Power Auxiliary Engine Power Total Engine Power (kW) (kW) (kW)

6896 2106 11537

Table 12: Daily Fuel Consumption: Average Cable Ship (calculated from GMSL, 2009).

In Port Manoeuvring At Sea (tonne) (tonne) (tonne)

3.0 13.2 16.0

The typical mission lengths to load, clear the route and install the cable are calculated from confidential installation reports. The figures are normalised to 1,000km of cable and are presented in Table 13 below. The lay operations includes the combined average for both shallow water (<1,000m) and deepwater (>1,000m) installation.

Table 13: Average Cable Ship: typical installation mission normalised to 1,000km of cable (calculated from confidential source).

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

At Sea Manoeuvring In Port Operation Mode (days) (days) (days)

Load 8.08 0.07 11.66

PLGR 8 0.10 0.41 0.06

Lay 2.43 9.79 1.58

Total Days 10.6 10.3 13.3

Based on the normalised installation mission of approximately 34 days, it is calculated that a total of 345 tonnes of fuel are used to install 1,000km of cable.

Air travel for personnel is assumed to have a noticeable impact and has been modelled for the installation phase. Travel to and from the vessel during “crew changes” is estimated based on 48 personnel made up of officers and crew. Air travel has been calculated at an average of approximately 6600pkm per day’s operation of the cable ship. Based on the calculate mission length of 34.18 days, the total air travel is calculated at 224830pkm per 1,000km of cable installed. Air travel is further described in Section 4.4.2.3.

The daily maintenance of the cable ship has been included due to its release of pollutants into the marine environment. This has been modelled using the LCA study of Fet et al (1996) described in Section 4.4.2.1. The surface area of the hull was estimated based on the length, breadth and draft of the studied vessel in relation to the cable ship. A factor of 1.87 was calculated and applied to the results of the study to account for the difference in ship size. The results of the study represent 10 years of total maintenance and were reduced to a daily figure. The paint itself is estimated from a process for enamel paint, assumed to represent a similar process to the antifouling paint used for ships (Chalmers, 2009).

The reference flow for the installation sub-model is 1,000km of cable.

4.4.5.2. Repeater Installation

Repeaters are laid in conjunction with the cable installation and are inseparable from that process.

4.4.5.3. Terminal Installation

The construction of the terminal building is not included in this study - further explained in section 4.4.4.4.

4.4.6. Use & Maintenance Phase The environmental impact from the use & maintenance phase results from the consumption of electricity at the terminal and the maintenance of the cable with ships. Electricity at the terminal is consumed in order to power both the cable repeaters and the terminal equipment. Power consumption figures from 2 terminal stations reveal that the cable consumes approximately 3.6% of the total energy budget for the terminal station, when normalised to 1,000km of cable. This 3.6% has been allocated to the cable sub- model with the remaining energy use allocated to the terminal sub-model. Electricity is modelled based on the average mix, as detailed in Section 4.4.1.1. Maintenance of the cable is required when a fault develops that requires repairing the cable with a purpose-built cable ship. The commercial lifetime of the cable at 13 years has been considered in this study (Ridder, 2007), however, the technical lifetime is considerably

8 Pre-Lay Grapnel Run (PLGR); used for route clearance prior to burying to cable by plough generally down to 1000 metres of water.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

greater at 25 years (Beaufils, 2000). Sensitivity analysis has been undertaken to assess the environmental impact based on the greater technical lifetime. The sensitivity analysis is detailed in Section 7.5.2.

4.4.6.1. Cable Use & Maintenance

Use & maintenance of the cable can be divided into the maintenance with ships and the electrically energy required to power the repeaters. It is assumed that cable maintenance and repair is undertaken by a similar ship as used for installation. Average engine and fuel consumption figures are presented in Table 11 and Table 12. Emission factors are presented in Section 4.4.2.1.

The typical mission lengths to load, transit to the site and repair the cable have been calculated from the operational record of four cable ships stationed on maintenance standby and covering a total of 197,000km of cable. Based on the number of repairs undertaken by these vessels during the period an annual fault factor of 0.37 was calculated when normalised to 1,000km of cable (confidential source). This figure is consistent with the work of Kordahi et al (2007). The number of vessel days on standby in port was compared with the number of vessel days on repair operations. In order to estimate the emissions from the combustion of fuel and due to lack of detailed data, it was necessary to make an assumption, based on professional judgement, of the operation mode of the vessel during repair operations. Naturally, this increases the uncertainty for this sub-model. Table 14 presents the estimated values.

Table 14: Average Cable Ship: Estimated operation mode per cable repair.

Transit Operations Loading Operation Mode (At sea) (Manoeuvring) (In port)

Assumed 50% 40% 10%

Based on the above assumptions, the annual repair mission normalised to 1,000km of cable was calculated and is presented in Table 15.

Table 15: Average Cable Ship: Annual repair mission normalised to 1,000km of cable (confidential source).

Operation Mode At Sea Manoeuvring Port

Total Days 2.1 1.7 3.6

Based on the 1,000km normalised repair mission, it is calculated that a total of 67 tonnes of fuel are used annually.

As with other sub-models, air travel is assumed to have a noticeable impact. Travel to and from the vessel during “crew changes” is estimated based on 48 personnel made up of officers and crew. Air travel has been calculated at an average of approximately 6,600pkm per day’s operation of the cable ship. Based on the annual operation time of 7.41 days, the total air travel is calculated at approximately 49,000pkm per 1,000km of cable maintained. Air travel is further described in Section 4.4.2.3.

The daily maintenance of the cable ship has been included and is explained further in Section 4.4.5.1.

Electricity is sent down the cable via the inbuilt copper conductor, in order to power the submarine repeaters. Analysis of the DC power feed readings from two terminal stations is used to calculate the average electricity requirement to power 1,000km of cable, being 1.09 kW at approximately one ampere (1A). This equates to approximately 34 GJ of electricity annually and is roughly 3.6% of the total energy

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

budget for the terminal station. It is assumed that this figure represents the common cable system. Electricity production is taken from the average mix sub-model, described further in 4.4.1.1.

The reference flow for the cable use & maintenance sub-model is 1 year of operation.

4.4.6.2. Terminal Use & Maintenance

The terminal electricity budget has been calculated from the monthly and yearly consumption of two terminal stations (confidential sources). The total consumption of a typical terminal station including, terminal equipment, climate control and lighting is calculated at approximately 191 kW, which equates to roughly 6034 GJ of electricity annually. It is assumed that this figure represents the common cable system. Electricity production is taken from the average mix sub-model, described further in 4.4.1.1.

Upgrading is likely to affect the energy use at the terminal due to efficiency gains in the equipment. Consultation with persons working within the industry suggests that 3 upgrades are likely over the course of the system’s 13 year commercial lifetime (Betts, 2009; Veverka, 2009). In order to avoid complexities in the model, upgrading is simplified and is treated as a linear input during raw material production and manufacturing. Upgrading and efficiency gains are not considered under the use & maintenance sub- model.

The reference flow for the terminal use & maintenance sub-model is 1 year of operation.

4.4.7. End-of-Life Decommissioning Phase A recovery and recycling scenario has been assumed for the end-of-life treatment of the cable based on the main mechanical materials. Terminal station recycling has been considered for the lead acid batteries and the PBA component only. It is assumed that this does not significantly affect the results. The scenario for cable recovery and recycling is based on information provided by a company whose “core business is the recovery and dismantling of redundant submarine cables” (Mertech, 2009, p1). Recently, this company recovered 350km of the SAT-1 cable, decommissioned in 1993. While SAT-1 was not a fibre optic cable, the same mechanical materials including, copper, steel and high-density polyethylene (HD-PE) plastic were recovered and are therefore representative of modern fibre optic cables (Louw, 2009).

The reference flow for the recovery and recycling scenario sub-model is 1,000km of cable.

4.4.7.1. Cable Recovery

Recovery is similar to the installation process in that the cable must be recovered with a purpose-built cable ship. Figures indicate that the recovery rate is similar to the installation rate (Louw, 2009). It is therefore assumed that recovery of the cable is the exact reverse process to installation, resulting in the same resource use and emissions. The installation sub-model described in Section 4.4.5.1 is used to estimate the impact of recovery. Based on the installation sub-model, a total of approximately 34 days ship operations and 345 tonnes of fuel are used to recover 1,000km of cable.

4.4.7.2. Cable Recycling

Cable recycling is a two part process involving dismantling of the cable into its separate mechanical materials at the shipyard and then processing of the materials into recycled raw product. Cable dismantling is undertaken using special purpose equipment drawing electrical power. Modelling of the dismantling process has been simplified based on the total electricity consumption of the shipyard per unit length of cable and is estimated at approximately 2.2 kWh per tonne (Louw, 2009). Copper, Steel and HD-PE are separated and sent to recycling plants to be smelted into billets or palletised. Recovery of these materials is

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

considered to be 100 percent (Louw, 2009). However, allowing for loss during the recovery and recycling process, it is assumed that 90 percent of the materials are available for return to the system to offset the input of virgin materials in a closed-loop process. The 10 percent loss from the system is assumed to result from the mechanical separation and smelting processes. This is likely to be a conservative estimate. No emissions or wastes are modelled for this 10 percent loss.

Steel is one of the most recycled materials and is 100 percent recyclable without loss of quality or properties. All ‘new’ steel products contain some recycled material (Corus, 2007; World Steel, 2008). Steel production technologies use two methods; the basic oxygen furnace (BOF) utilising 25 to 35 percent scrap steel and the electric arc furnace (EAF) utilising more than 80 percent scrap steel (Steel Recycling Institute, 2007). Most scrap steel is recycled using the EAF, with the main inputs being recycled steel and electricity (BlueScope, 2009; World Steel, 2008). In this study it is assumed that all steel is recycled using the EAF. The recycling process is simplified to the electricity consumption of the EAF, with no other emissions accounted for. The literature shows primary energy consumption for recycling of steel ranges from 7 to 17.4 gigajoules per tonne (Jones, 2009; BlueScope, 2009; Corus, 2007; World Steel, 2008). Averaging these figures and applying the primary to secondary energy factor, explained in Section 4.4.1.1, gives an average electricity input of 796kWh per tonne of recycled steel – presented in Table 16.

Table 16: Energy consumption – steel recycling process (Jones, 2009; BlueScope, 2009; Corus, 2007; World Steel, 2008).

Steel Recycling Energy Primary Secondary

(GJ/tonne) (kWh/tonne) (kWh/tonne)

Corus Group 17.4 4833 1267

BlueScope Steel 7 1944 510

World Steel Association. 10.8 3000 787

American Iron and Steel institute - - 620

Average Input Energy (Electricity) 796

Estimated Input Energy (Electricity) 800

For this study, 800kWh per tonne has been estimated for the recycling of steel. It is assumed that steel recycled from the cable is of equal quality to virgin material, with a 10 percent loss in the recycling process, thereby replacing 90 percent of the virgin material input. It is assumed that the zinc used for galvanizing the steel wires is lost from the system. No emissions are modelled for this loss.

Worldwide, recycling rates of copper are between 40 to 60 percent, with demand far out weighing the availability of scrap for recycling (Graedel et al , 2004; Lifset et al , 2002; Spatari et al , 2005). Losses in production are only significant in milling and smelting. Refining and fabrication losses can be readily recovered and returned to the system (Lifset et al , 2002). Giurco et al (2006) estimate that copper recycling using a reverberatory smelter consumes 1200 kWh of electricity and 480kg of fuel oil with material recovery of 88 percent. It is assumed that copper recycled from the cable is of equal quality to virgin material, with a 10 percent loss in the recycling process, thereby replacing 90 percent of the virgin material input.

High density polyethylene (HD-PE) recovered from the cable is processed into pellets for reuse. Louw (2009) describes the material as “top-quality non-virgin material”. Dodbiba et al (2008) describe this process as mechanically recycling , where the basic structure of the plastic material is unaffected. Mechanical recycling of plastics is only effective if the recovered material is of high-quality (96% purity). Furthermore,

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

they compare energy recovery from incineration to mechanical recycling and conclude that the latter has less environmental burden. Louw (2009) estimates the recycling of HD-PE to consume 400 kWh per tonne of material. It is assumed that HD-PE recycled from the cable is of equal quality to virgin material, with a 10 percent loss in the recycling process, thereby replacing 90 percent of the virgin material input.

Stainless steel has also been considered for recycling due to it containing scare and energy intensive elements and the potential for emissions reductions when compared to virgin material (Igarashi et al, 2007). It is assumed that the recycling of stainless steel uses the same energy input as steel recycling at 800 kWh per tonne. Recovery is also assumed at 90 percent.

Tests show that the recovery of the fibre itself is economically unattractive as complete gel removal is not possible. Energy recovery was therefore considered as potentially the best solution for the gel fractions (Arnaiz et al , 2008). In this study no allowance has been made to recover energy from the gel as it accounts for only 1 percent of the total weight of the lightweight cable.

The other main mechanical materials that make up the single armour (SA) and double armour (DA) cables are polypropylene yarn and bitumen. These materials are sealed in drums and sent to a licensed landfill (Louw, 2009). It is assumed that all non-recycled cable waste is treated in this way. The landfill process is modelled based on the standard GaBi database process for commercial waste. No defined time horizon is given for emissions, however, based on the literature (Baumann and Tillman, 2004), a cut-off at the surveyable time period of approximately 100 years is assumed.

4.4.7.3. Repeater Recycling

It is assumed that the main mechanical materials of the repeater are recycled, particularly the 170kg beryllium copper casing. Based on the assumptions made in Section 4.4.3.2, the repeater contains copper, steel and stainless steel. Recycling of these materials is explained above in Section 4.4.7.2. In addition, aluminium and the printed board assembly (PBA) are considered for recycling and waste handling.

Aluminium is in principle indefinitely recyclable as it retains its structure and inherent properties through the melting process (EAA, 2007). Processing plants for secondary aluminium do not generally use electrical energy for the furnaces, using mainly mineral oil and natural gas. However, the auxiliary equipment does consume substantial quantities of electrical energy (Schmitz et al , 2006). Further, Schmitz et al (2006) estimates the energy consumed by a reverberatory furnace to be between 1,000 to 1,200 kWh per tonne. Green (2007) estimates the total energy consumed to produce secondary aluminium at 2,800kW per tonne. Accepting the upper limit of Schmitz et al (2006) at 1,200kWh per tonne and assuming that this represents the fuel oil consumed, a total of 1,600kWh per tonne remains, as presented in Table 17. It is assumed that this represents the electrical energy consumed during the recycling process.

Table 17: Energy consumption – aluminium recycling process (Green, 2007; Schmitz et al , 2006).

Primary Aluminium Recycling Energy (kWh/tonne)

Green, 2007 (Total) 2800

Schmitz et al , 2006 (Furnace only) 1200

Estimated Input Energy (Fuel Oil) 1200

Estimated Input Energy (Electricity) 1600

Electricity and heavy fuel oil come from the sub-models used throughout the study, explained further in Sections 4.4.1.1 and 4.4.1.1. It is assumed that recycled aluminium is of equal quality to virgin material,

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

with a 10 percent loss in the recycling process, thereby replacing 90 percent of the virgin material input. This is a conservative figure and no emissions are accounted for in relation to this loss.

It is assumed that the PBA is incinerated and the energy recovered, providing approximately 0.1 percent return into the system. No material recovery has been considered. Emissions are accounted for by the standard database process for PBA incineration.

4.4.7.4. Terminal Recycling

Recycling of the terminal components has been considered for the lead acid batteries and the PBA elements only. These are considered to represent the most toxic substances in the sub-model. The lead acid battery recycling has been based on a study by Salomone et al (2005). It is assumed that the PBA is incinerated, which is modelled using database processes. Material recovery has not been considered. It is further assumed that the remaining components of the terminal station, the mechanical enclosures and the generators, are used for subsequent systems and therefore are lost from the system with no emissions or waste generated.

4.5. Allocation The ISO 14040 series guidelines state that where possible allocation should be avoided by increasing detail in the model or expanding the system to include those processes requiring allocation. However, when allocation cannot be avoided then, the environmental loads should be partitioned to reflect the underlying physical relationship or other relationships, such as proportioned economic value (Baumann and Tillman, 2004; ISO 14044:2006).

Data for the raw materials and manufacturing processes for the cable has been provided by a number of suppliers to Ericsson. These companies are likely to produce more than one product, hence the data must be allocated (Bergelin, 2008). In this case, companies have undertaken the allocation of materials and energy inputs, and any emissions, themselves. For confidentiality reasons, no control over this allocation procedure was possible within the scope of this study. This leaves the data open to uncertainty as the allocation method cannot be verified for the purposes of this study. The impact of this on the final result is reduced significantly by the end-of-life scenario, which replaces 90 percent of the virgin material with recycled material.

Allocation assumptions for the end-of-life phase are based on an approximation of a closed-loop recycling process. The system accounts for the recycling of cable materials in a simplified model based on energy consumption during the recycling process. This approximation is considered valid for materials, such as metals, that retain their quality after recycling (Baumann and Tillman, 2004). It is assumed that the recycled materials replace 90 percent of the virgin material input. Plastics are usually considered degraded by the recycling process and may be allocated on a 50/50 method ( ibid ). For submarine cable recycling, no loss of quality is observed (Louw, 2009) and it is assumed that recycled HD-PE also replaces 90 percent of the virgin material. In line with the ISO 14040 series guidelines, a sensitivity analysis has been undertaken on the end-of-life phase to determine the impact of this allocation assumption (ISO 14044:2006). The sensitivity analysis is presented in Section 7.5.2.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

4.6. Inventory Results and Discussion

4.6.1. Inventory Results A summary of the significant LCI data is presented in Table 18. The values represent 99 percent of the potential impact under each impact medium, thereby accounting for the most significant inputs and outputs of the system. All LCI data is stored in a database containing the complete LCA model created using the GaBi software.

Table 18: LCI result summary for submarine cable system

Commercial lifetime 13 years, 10,000km, 5.2 terminal stations Substance Impact Media Amount Unit Hard coal Resource 33537318 kg Crude oil Resource 21283296 kg Natural gas Resource 5419385 kg Lignite Resource 990209 kg Uranium natural Resource 1100 kg Copper Resource 800008 kg Copper ore Resource 5759909 kg Gold Resource 6.549 kg Lead ore Resource 309661 kg Zinc ore Resource 1242200 kg Carbon dioxide Air 270669959 kg Carbon monoxide Air 232314 kg Halon (1301) Air 0.0096 kg Methane Air 297449 kg Nitrogen oxides Air 4103853 kg Nitrous oxides (NOx) Air 293434 kg NMVOC (unspecified) Air 108425 kg R 11 (trichlorofluoromethane) Air 0.110 kg R 114 (dichlorotetrafluoroethane) Air 0.112 kg R 12 (dichlorodifluoromethane) Air 0.024 kg Sulphur dioxide Air 2658144 kg Sulphur oxides (SOx) Air 207900 kg VOC (unspecified) Air 28941 kg Cobalt Freshwater 102.0 kg Copper Freshwater 271.9 kg Nickel Freshwater 261.9 kg Polycyclic aromatic hydrocarbons (PAH, unspec.) Freshwater 168.9 kg Selenium Freshwater 258.5 kg Vanadium Freshwater 264.6 kg Barium Seawater 101.6 kg Beryllium Seawater 0.134 kg Copper (+II) Seawater 1471 kg Tributyltinoxide Seawater 1294 kg Zinc (+II) Seawater 910 kg Chromium (unspecified) Soil 1.243 kg Arsenic (+V) Soil 0.091 kg Nickel (+II) Soil 0.460 kg

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

4.6.2. LCI Discussion This section presents a discussion on important aspects of the life cycle inventory (LCI).

4.6.2.1. LCI Data Quality

In comparing the five life cycle sub-models, data for the use & maintenance phase is considered the most reliable. Notwithstanding that the data lacks transparency, as it is based on confidential information and that the HFO model is biased toward EU and US production (explained further below), data in this phase is based on measured data for terminal energy use and ship maintenance days. In addition, the electricity model is defined by previous LCA studies and emissions from ships are modelled from the literature.

The least reliable data is for the raw materials and design & manufacturing phases, where large uncertainty exists in the data due to data gaps, allocations made by suppliers and the use of material processes from the GaBi database. Data collected from suppliers to Ericsson comes with high uncertainty as, in most cases, this was presented in email form or as answers to the questionnaire, rather than official reports. Therefore, it is difficult to verify the source and quality of this data. It is assumed that this data is based on measured values, though again, the allocation methodology of this data is uncertain. Modelling of the upstream processes has been taken from the standard databases. The quality of these processes is directly affected by the quality of the source data (3GLCA, 2002).

4.6.2.2. Representativeness

The data collection process of this study is based on “voluntary information exchange” (3GLCA, 2002, p.49). It is likely that environmental data is representative of the best available technologies based on modern facilities with pro-active environmental profiles. Companies with poor or non-existent environmental monitoring often do not release data or simply ignore requests to participate. The result is that the available data may have “a tendency to reflect better-than-average conditions” (ibid ). This is illustrated by the sub-model for heavy fuel oil production, which is a 50:50 combination of EU and US production. This does not take into account the production in other regions that may have less efficient production methods or poorer environmental controls (ibid ).

Emissions from the combustion of marine fuels have been based on residual oils (ROs), which, on a global scale, the majority of ships use. However, the company used as the reference for cable ship specifications advised that marine distillates (MDs) were used in their ships (Todd, 2009). As discussed previously, MDs have significantly less emissions for NOx and SOx, which in turn will reduce the potential impacts. Therefore, the use of RO emissions factors should be considered a worst-case scenario. The difference in impact between RO and MD emissions is addressed in the sensitivity analysis in Section 7.5.2.

Data has been collected from many sources. Where only single source data was available it is assumed that this is representative of the generic process. This raises the uncertainty of the result as verification is not possible. Where similar data from multiple sources was available an average value has been calculated and used in the model.

4.6.2.3. Completeness and Consistency

One of the objectives of this study has been to collect as up-to-date data as possible from suppliers and operators. Given the complexity of the system and the time limitation of 20 weeks it was not possible to collect detailed data on all aspects of the system, particularly submarine repeaters and components of the terminal station. As such, published data from previous studies and databases of standard processes have been used to fill the data gaps.

4. Life Cycle Inventory (LCI) A life cycle assessment of fibr e optic sub marine cable systems

For the cable, data for the raw materials is near complete, with all significant processes accounted for based on information from suppliers to Ericsson and standard databases within the GaBi software. However, reliance on the published databases reduces consistency as it is not possible to verify the age, uncertainties, system boundaries and allocation procedures for each standard process (3GLCA, 2002). Manufacturing of the cable is based on the effect rating of the plastic extrusion and armouring stations. This is likely to be an under estimation of the total energy requirement to manufacture the cable. Emissions from the cable manufacturing process and the end-of-life recycling process are limited to the emissions from electricity production. Therefore, these emissions are likely to be under estimated. Data for the installation, use & maintenance and end-of-life phases for the cable is considered to be complete, though comes with some uncertainty as the data is not verifiable within the scope of this study.

For the repeater and terminal station, it was not possible to source data for the raw material and manufacturing phases within the study timeframe, as such, these processes have been represented by similar processes from the literature. The effect of this on the results is explored in the sensitivity analysis in Section 7.5.1. The consequence of this data gap and the results of the sensitivity analysis must be taken into account when considering the final result. The installation phase for the terminal station has been assumed to have little effect on the result and has not been modelled. Data for the use & maintenance phase for the terminal station has been taken from measured values at a number of terminal stations, which agree with good consistency. Data for the end-of-life phase is again affected by the data gap described above and subsequent assumptions.

For consistency, the average electricity mix and heavy fuel oil (HFO) sub-models have been used for all respective electricity and HFO inputs throughout the model.

5. Life Cycle Impact Assessment (LCIA) A life cycle assessment o f fibre optic submarine cable systems

5. Life Cycle Impact Assessment (LCIA)

This section describes the life cycle impact assessment (LCIA), which “aims at describing the environmental consequences of the environmental loads quantified in the inventory analysis” (Baumann and Tillman, 2004, p.129). Impact categories, category indicators and the characterisation models are based on the CML problem-oriented approach that focuses on environmental problems or the so-called midpoint of the cause- effect chain (Guinée et al, 2004).

5.1. General Allocation Procedure Classification of the LCI data can result in parameters being assigned to more than one impact category.

For example, emission of nitrogen oxides (NO x) can be assigned to the acidification, eutrophication and, in some cases, the photo-oxidant formation categories. Multiple assignments should only be made if the effects are independent of each other, otherwise double-counting arises (Baumann and Tillman, 2004).

The case of NO x can be considered a serial mechanism, whereas, in the case of parallel mechanisms, such as sulphur dioxide (SO x), the parameter should be apportioned between impact categories, such as, human health and acidification (ISO 14044:2006). Previous studies show that the impact of SO x on human health is negligible in comparison to the contribution to acidification. Leading to a general conclusion “…that other uncertainties probably will overshadow the need for partitioning or allocation” (3GLCA, 2002, p.53). Thus, in this study, the problem of allocation between impact categories has not been addressed.

5.2. Definition of Impact Categories and Characterisation Factors The characterisation models used in this study follow the method developed by the Institute of Environmental Sciences (CML) at the University of Leiden, in the Netherlands. This CML methodology uses a problem-oriented approach that focuses on environmental problems or the so-called midpoint of the cause-effect chain (Guinée et al, 2004). Impact categories, category indicators and the characterisation models, along with the methodology and scientific rationale for the CML impact assessment models are described in detail in Guinée et al, (2004). In this study, ten baseline impact categories (group A categories) have been selected using the CML 2001 characterisation database supplied with the GaBi software. Each impact category is described in the following sections and summarised below in Table 19.

Table 19: CML impact categories used for the life cycle impact analysis (Adapted from Guinée et al, 2004).

Impact Category Year Indicator

Abiotic Resource Depletion 2001 kg Sb eq. 9

Acidification Potential 2001 kg SO 2 eq.

Freshwater Ecotoxicity Potential 2001 kg 1,4DCB eq.

Terrestrial Ecotoxicity Potential 2001 kg 1,4DCB eq.

Marine Aquatic Ecotoxicity Potential 2001 kg 1,4DCB eq.

Global Warming Potential 2001 kg CO 2 eq.

Photochemical Ozone Creation Potential 2001 kg C 2H4 eq.

Ozone Depletion Potential 2001 kg CFC-11 eq.

Eutrophication Potential 2001 kg PO 4 eq.

Human Toxicity Potential 2001 kg 1,4DCB eq.

9 eq. - abbreviation for ‘equivalent’.

5. Life Cycle Impact Assessment (LCIA) A life cycle assessment o f fibre optic submarine cable systems

5.2.1. Abiotic Resource Depletion Depletion of abiotic resources relates to the irreversible use of natural non-living resources such as metals, minerals and fossil fuels. The abiotic depletion potential (ADP) for each resource is determined from the extraction rate and the remaining reserves. These are then compared to the reference case for depletion of the rare metal antimony (Sb). The reference unit is kilograms Sb equivalent (kg Sb eq.) (BRE, 2005; Guinée et al, 2004).

5.2.2. Acidification Potential Acidification results from acidifying pollutants reacting with water in the atmosphere to form “acid rain”. This has a detrimental effect on biological organisms, ecosystems and materials, such as buildings. The characterisation model is adapted to LCA based on the RAINS10 model describing the deposition of acidifying substances. The major pollutants are sulphur dioxide (SO 2), nitrogen oxides (NO X) and ammonia (NH 3). The acidification potential (AP) of each pollutant is expressed in the reference unit of kilogram emissions of SO 2 equivalents (kg SO 2 eq.) (BRE, 2005; Guinée et al, 2004).

5.2.3. Ecotoxicity Potential to Freshwater, Land and Seawater Ecotoxicity relates to emissions of toxic substances having a detrimental effect on ecosystems. The characterisation model is adapted to LCA based on the USES 2.0 model which describes fate, exposure and effects of toxic substances. Freshwater aquatic ecotoxicity potential (FAETP) relates to the impact on freshwater aquatic ecosystems, marine aquatic ecotoxicity potential (MAETP) relates to marine aquatic ecosystems and terrestrial ecotoxicity potential (TETP) relates to land-based ecosystems. Emissions are related to the reference unit of kilograms of 1,4-dichlorobenzene equivalents (kg 1,4DCB eq.) (BRE, 2005; Guinée et al, 2004).

5.2.4. Global Warming Potential Global warming potential or “climate change” relates to the impact of human emissions of greenhouse gases (GHGs) and the resultant radiative forcing of the atmosphere. The characterisation model has been developed by the Intergovernmental Panel on Climate Change (IPCC) describing the global warming potential for a 100-year time horizon (GWP100). Emission factors are measured against the reference unit of kilograms of carbon dioxide equivalents (kg CO 2 eq.) (BRE, 2005; Guinée et al, 2004).

5.2.5. Photochemical Ozone Creation Potential Photochemical ozone creation is the effect of emissions such as carbon monoxide (CO) and volatile organic compounds (VOCs) that cause ozone to be created in the presence of sunlight, commonly seen as summer smog. Winter smog, by contrast, is considered under human toxicity. Photochemical ozone creation potential (POCP) relates to the detrimental effects on human health and is modelled by the UNECE Trajectory model. Emission factors are measured against the reference unit of kilograms of ethylene equivalents (kg C 2H4 eq.) (BRE, 2005; Guinée et al, 2004).

5.2.6. Ozone Depletion Potential Stratospheric ozone depletion refers to the breakdown of ozone in the stratosphere resulting in a thinning of the ozone layer. This is caused by the anthropogenic emission of ozone-depleting gases to air. The ozone depletion potential (ODP) of different gases is modelled by the World Meteorological Organisation (WMO) and relates to human health and ecosystem effects caused by ozone depletion. Emissions of gases are measured against the reference unit of kilograms of chloroflurocarbon-11 equivalents (kg CFC-11 eq.) (BRE, 2005; Guinée et al, 2004).

5. Life Cycle Impact Assessment (LCIA) A life cycle assessment o f fibre optic submarine cable systems

5.2.7. Eutrophication Potential Eutrophication relates to the concentration of high levels of macronutrients in the environment. Nitrates and phosphates are essential for life, however increased levels can lead to detrimental shifts in species composition and elevated biomass concentrations, such as excessive algae growth in aquatic ecosystems. The eutrophication potential (EP) of emissions are modelled by the stoichiometric procedure and are measured against the reference unit of kilograms of phosphate equivalents (kg PO 4 eq) (BRE, 2005; Guinée et al, 2004).

5.2.8. Human Toxicity Potential Human toxicity relates to emissions of toxic substances having a detrimental effect on human health. The characterisation model is adapted to LCA based on the USES 2.0 model which describes fate, exposure and effects of toxic substances. The human toxicity potential (HTP) of each toxic substance emitted to air, water and soil is measured against the reference unit of kilograms of 1,4-dichlorobenzene equivalents (kg 1,4DCB eq.) (BRE, 2005; Guinée et al, 2004).

5.3. Classification and Characterisation Summary Classification involves the qualitative process of assigning the inventory data to each of the impact categories. In this case, no classification is needed as this process is defined by the CML methodology and factors are assigned in the characterisation database (Guinée et al, 2004).

Characterisation is the quantitative process of calculating the indicator results for each impact category based on the conversion of the LCI data to common units and the aggregation of the results (ISO 14040:2006). The result is a category indicator for each impact category, which together represent the environmental profile of the studied system (Guinée et al, 2004). Again, characterisation factors are defined in the CML 2001 characterisation database.

5.4. Definition of Normalisation Factors The aim of normalisation is to relate the characterisation results to a background or reference value that reflects the actual magnitude of the impact. This identifies if the impact is significant in relation to the total impacts of the studied area, which may have global or regional consequences (Baumann and Tillman, 2004). Normalisation factors used in this study are taken from Sleeswijk et al (2008), where the annual world reference values for the emissions and consumption of the significant substances under each impact category are collate for the year 2000. This then helps to frame the environmental profile of a submarine cable system within the greater economic system that it is a part of (Sleeswijk et al , 2008).

The resource and emission data presented by Sleeswijk et al (2008) have been divided by the world population for the year 2000, estimated at 6.12 billion (UN, 2009), resulting in an environmental impact expressed as annual person equivalents.

In attempting to quantify the global environmental impact, normalisation itself may result in uncertainties due to doubt in the emissions data and in the characterisation factors (Sleeswijk et al 2008). As such, qualitative assessment of the uncertainty has been undertaken by the authors. They note that the highest level of uncertainty is associated with the toxicity categories due to the scarcity of data and the uncertain fate modelling of heavy metals. While the lowest uncertainty from a global perspective is associated with global warming, acidification and photochemical ozone creation. To avoid introducing greater uncertainty into the results, only these three categories will be analysed for the purposes of normalisation. The normalisation factors are presented in Table 20.

5. Life Cycle Impact Assessment (LCIA) A life cycle assessment o f fibre optic submarine cable systems

Table 20: World normalisation factors, person equivalents per year (Adapted from Sleeswijk et al, 2008).

Normalization Factor Impact Category Unit (per capita per year)

Acidification (500 years) kg SO 2 eq. 61.8

Global Warming (100 years) kg CO 2 eq. 6830

Photochemical Ozone Creation kg NMVOC eq. 57.4

These normalisation factors are compared to the results based on the functional unit and the estimated per capita data annual traffic, to provide a percentage value relating to the annual environmental impact of one person per year. The results of the normalisation calculation are presented in Section 7.6.

6. Calculation Procedure A life cycle assessment o f fibre optic su bmarine cable systems

6. Calculation Procedure

This section describes the calculation procedure for taking the results of the LCIA and presenting them in relation to the functional unit of the study.

In order to generate the LCIA results for each impact category, the LCI data must be multiplied with the corresponding characterisation factor. The characterisation factors used in this study are taken from the CML database - year 2001.

The modular system of the GaBi software allows for the model to be analysed as a whole or by sub- model, down to individual processes. The model for the generic cable system is built up of two main sub- models, the cable and the terminal station. These two sub-models allow for the flexibility in analysing the system, whereby scaling of the cable length and the number of terminals can, in principle, allow modelling of any particular repeated submarine cable network.

Cable systems are decidedly variable in architecture and can be built as point-to-point, branched or ringed systems, as shown in Section 2.2.3. Consequently, the length of cable and number of terminal stations can vary considerably. In order to estimate the architecture of the generic system, a total of 24 networks were averaged from three key regions; Transatlantic, Transpacific and South Asia (Europe to the Far East). A summary of each region is presented in Table 23, with full details presented in Appendix A.

Table 21: Calculation summary for the generic cable system (Adapted from Ruddy, 2006).

Design Lit Capacity Number of Total Length Lit Capacity Total Cable Region Capacity, est. as % of Systems (km) (Gbps) Landings (Gbps) Design Transatlantic 11 128201 2707 12217 22.16% 49

Transpacific 5 111536 1320 9060 14.57% 30

South Asia 8 149213 829 15636 5.30% 125

Total 24 388950 4856 36913 13.16% 204

Average generic system 1 16206 202 1538 13.16% 8.5

Normalised to 10000km of cable 10000 5.2

The results of the LCIA for the 13 year life cycle of the cable are based on a system normalised to 10,000 kilometres of cable, in order to relate to the functional unit. This then allows for the calculation of the potential impacts to send 1 gigabit (Gb) of data over 10,000km of cable. However, in order to complete the functional unit calculation, an additional calculation is necessary based on system capacity and actual data traffic (or bandwidth usage). System capacity is defined by three figures; design capacity: the technical limits to system capacity, lit capacity: the current installed capacity at the terminal station and, bandwidth usage: the actual usage of the system (Telegeography, 2009). Table 21 shows that on average only 13 percent of the design capacity is currently lit. The fact that over 85 percent of the capacity remains unlit highlights the significant potential for system upgrading, without the need to build new systems (Kidorf, 2006). Telegeography (2009) note that used capacity should not be interpreted as actual data traffic, however, in this study it is assumed that used capacity approximates data traffic.

6. Calculation Procedure A life cycle assessment o f fibre optic su bmarine cable systems

The capacity calculation is based on an assumption of the modern system, whereby the average capacity is calculated from systems installed on, or after, the year 2000 and with a capacity greater than 100Gbps. A total of 11 systems fulfil this criterion, as detailed in Appendix A. If all 24 systems are used in the calculation, then the lit capacity is averaged at approximately 200Gbps (see Table 21). However, this data is then biased by the older systems, in the worst case having only 1 or 2Gbps total capacity. Therefore, the older systems have not been used in this calculation in order to provide a reasonable estimate of cable capacity based on current technology. Using the average capacity of the 11 systems, it is estimated that the current lit capacity of the modern cable system is 400Gbps. However, research by Telegeography (2009) shows that actual used bandwidth is, on average, only 25 percent of the lit capacity. Assuming bandwidth usage approximates data traffic then the average data traffic is calculated at 100Gbps. This figure should be considered an average for the generic system and individual systems may vary.

If this 100Gbps is multiplied by the number of seconds in a year and then by the 13 year lifetime of the system, the annual and lifetime data traffic is estimated. A summary of this calculation is presented below in Table 22.

Table 22: Summary of capacity calculation for the generic cable system (Adapted from Ruddy, 2006).

Capacity Calculation

Average lit capacity of modern system (>year2000 and >100Gbps) 381 Gbps

Estimated current lit capacity of modern system 400 Gbps

Assumed bandwidth usage of modern system at 25% 100 Gbps

Estimated total annual data traffic of modern system 3155692600 Gb/year

Estimated total lifetime data traffic of modern system 41024003800 Gb/lifetime

This gap between used capacity and lit capacity reflects the structure of the cable market influenced by a number of factors; reserved restoration capacity and allocation for future needs by bandwidth purchasers, market inefficiencies, contract structures and reserved spare inventory by the bandwidth suppliers (Telegeography, 2009).

The results of the LCIA are based on a 13 year system lifetime and normalised to 10,000km of cable and 5.2 terminal stations (Table 21). These results are then divided by the estimated lifetime data traffic (Table 22) to give the resultant potential impact of 1Gb of data sent over 10,000km of cable, thereby relating to the functional unit of 10,000Gb·km . The results and life cycle interpretation are presented in Section 7.

It must be acknowledged that this calculation assumes a steady state and does not allow for capacity upgrading. However, upgrading will work to reduce any future potential environmental impacts. The results must therefore be viewed in the context of this data traffic calculation. Furthermore, this calculation is linear, and for future application, the results of this study can be scaled by a factor relating to the estimated annual data traffic of this study and actual annual traffic of the system under investigation. Though care must be taken to account for the difference between one data bit and one byte of data.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7. Results of Life Cycle Interpretation

This section details the results of the life cycle interpretation, which are presented in relation to the functional unit using a variety of bar charts in the following sections.

Firstly, the results based on the functional unit are summarised for each impact and energy category. Secondly, each impact/energy category is analysed individually based on the life cycle phases. Thirdly, each life cycle phase is analysed based on the components of the system; cable, repeaters and terminal station. This section gives particular attention to the climate change potential, as this is of particular interest to Ericsson in their research into the total carbon footprint of the global ICT network.

The results are based on what is considered to be world average values and no attempt has been made to model regional differences. It is not possible to extract the electricity or fuel oil consumption from the primary energy value for each GaBi database process, therefore, electricity and fuel oil consumption will be under estimated for the raw materials phase. All other phases are near complete.

It is important to remember when interpreting the results, that an LCA is based on models and that these models are simplifications of reality (3GLCA, 2002).

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.1. Summary of Results

Commerical lifetime 13 years, 10000km and 5.2 terminal stations

Primary Energy

Electricity

Heavy Fuel Oil

Abiotic Resource Depletion

Acidification Potential

Freshwater Aquatic Ecotoxicity Potential

Terrestric Ecotoxicity Potential

Marine Aquatic Ecotoxicity Potential

Global Warming Potential

Photochemical Ozone Creation Potential

Ozone Depletion Potential

Eutrophication Potential

Human Toxicity Potential

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Raw Materials Manufacture Installation Use & Maintenance End of Life

Figure 18: Summary of results – per 10,000Gb·km.

The results are presented in Figure 18 and show that the use & maintenance phase clearly dominates all impact categories at an average of 66 percent , with the exception of the ozone depletion potential (ODP) category. By comparison, the raw materials and design & manufacturing phases account for, on average, only 6 percent of the total potential impact. Again, this is with the exception of ODP, which represent 67 percent of the impact. The result is not unexpected as la rge amounts of electricity and fuel oil are consumed during the 13 year use & maintenance of the cable. By comparison, it has been noted that the planning, design, manufacture and installation phases take approximately 18 months. The end -of-life phase includes recycling of the recovered materials , which offsets 90 percent of the virgin material input in an assumed closed-loop cycle, thereby reducing the impact of the raw materials phase significantly. Ozone depletion is significantly higher for the raw mate rials phase due to the release of halogenated organic emissions during material processing.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

Table 23: Summary of results – per 10,000Gb·km.

Total Raw Design & Use & Impact Category Units Installation End of Life 10000Gb·km Materials Manufacture Maintenance

Primary Energy MJ 6.580E-02 4.620E-03 2.254E-03 3.892E-03 4.818E-02 6.849E-03

Electricity kWh 3.090E-03 2.206E-05 9.817E-05 2.390E-09 2.792E-03 1.777E-04

Heavy Fuel Oil kg 4.041E-04 0.000E+00 1.549E-05 8.415E-05 2.124E-04 9.205E-05

Abiotic Resource kg Sb eq. 2.390E-05 1.623E-06 8.228E-07 1.861E-06 1.677E-05 2.819E-06 Depletion

Acidification kg SO eq. 1.402E-04 7.103E-07 5.103E-06 2.763E-05 7.852E-05 2.823E-05 Potential 2

Freshwater Aquatic kg 1,4DCB eq. 1.591E-04 2.323E-06 5.341E-06 6.619E-06 1.309E-04 1.395E-05 Ecotoxicity Potential

Terrestrial Ecotoxicity kg 1,4DCB eq. 1.948E-05 1.642E-07 7.495E-07 2.757E-06 1.264E-05 3.172E-06 Potential

Marine Aquatic kg 1,4DCB eq. 5.914E-01 2.029E-02. 1.905E-02 5.603E-02 4.218E-01 7.428E-02 Ecotoxicity Potential

Global Warming kg CO eq. 6.852E-03 1.790E-04 2.676E-04 9.225E-04 4.409E-03 1.075E-03 Potential 2

Photochemical Ozone kg C H eq. 7.423E-06 7.399E-08 2.954E-07 1.452E-06 4.111E-06 1.491E-06 Creation Potential 2 4

Ozone Depletion kg CFC11 eq. 8.308E-12 4.227E-12 1.358E-12 5.841E-13 1.484E-12 6.547E-13 Potential

Eutrophication kg PO eq. 1.417E-05 9.456E-08 4.961E-07 2.796E-06 7.900E-06 2.879E-06 Potential 4

Human Toxicity kg 1,4DCB eq. 2.811E-03 1.881E-05 1.051E-04 4.813E-04 1.694E-03 5.122E-04 Potential

The results for each impact category, based on the functional unit, are presented in Table 23. The results show that over the 13 year commercial lifetime of the system, an equivalent of 3Wh of electricity and 0.4 grams of fuel oil are consumed per 10,000Gb·km. Impacts of interest are Climate Change and

Acidification. The results reveal that 6.9 grams of carbon dioxide equivalents (CO 2 eq.) and 0.14 grams of

sulphur dioxide equivalents (SO 2 eq.) are released per 10,000Gb·km.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.2. Application of Results In order to place the results of the previous section into context, a number of examples are presented in this section based on the global warming potential category.

Referring back to the telepresence example presented in Section 3.2.2, a telepresence conference between Stockholm and New York would equate to the functional unit at 1.25Gb of data sent over 8,000km of cable. If this conference lasted 60 minutes then a total of 518,400Gb·km would result, giving a functional unit factor of 51.8. Applying this factor to the global warming potential impact result of 7 grams of carbon dioxide (CO 2) equivalents, then a total of 355 grams of CO 2 equivalents would potentially be released for the data transfer by submarine cable, per 60 minute telepresence conference. Studies show that 120 grams of CO 2 is released per person kilometre of air travel (see Section 4.4.2.3) and 160 grams of CO 2 are released per kilometre for the average new EU-15 passenger car in 2006 (3GLCA, 2002; European Commission, 2007). If the above example is then compared to the alternative of a face-to-face meeting requiring air travel, then 355 grams of CO 2 would potentially be released by a single person flying only 3 kilometres. If compared to the average passenger car, then this would equate to 2.2 kilometres. Table 24 shows a summary of this calculation.

Table 24: Comparison of results – Climate change: Example 1 (Adapted from 3GLCA, 2002; European Commission, 2007; Jonsson, 2009b).

Data transfer via subsea cable Comparison Factor Air Travel Car Travel Telepresence System

Impact 355 grams CO 2 eq. 355 grams CO 2 355 grams CO 2

Utility 60 minutes use of system 3 km 2.2 km

Assumptions: Telepresence bandwidth = 18Mbps transferred over 8,000km of cable. Note: Submarine cable data transfer only.

Further expansion of this example can assume a two day meeting with a total of 16 hours of telepresence use. Data transfer via the submarine cable would potentially release a total of 5.7 kilograms of CO 2 equivalents in 16 hours. By comparison, the air travel for a single person roundtrip would amount to

16,000 kilometres, resulting in 1920 kilograms of CO 2 emissions. The comparison is shown in Table 25.

Table 25: Comparison of results – Climate change: Example 2 - Single person, 2 day meeting. (Adapted from 3GLCA, 2002; Jonsson, 2009b).

Data transfer via subsea cable Comparison Factor Air Travel Telepresence System

Impact 5.7 kg CO 2 eq. 1920 kg CO 2

Utility 16 hours use of system 16,000 km

Assumptions: Telepresence bandwidth = 18Mbps transferred over 8,000km of cable. Note: Submarine cable data transfer only.

These examples clearly show the environmental benefits of using the ICT network, however, it must noted that they compare only data transfer by submarine cable to air travel. In reality the telepresence system is more complex and has a greater environmental impact as a whole (Jonsson, 2009b) as does international travel for a face-to-face meeting. It is not considered that air travel will be replaced entirely by submarine cable data transfer.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3. Results by Environmental Impact Category The following section presents the results by individual impact category and in relation to the functional unit of 10,000Gb·km. Overall, the results clearly highlight the dominance of the use & maintenance phase and the significant impact that electricity production and the combustion of heavy fuel oil (HFO) have on the outcome.

7.3.1. Energy Resources

Electricity (kWh) 0 0,0005 0,001 0,0015 0,002 0,0025 0,003

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 0,01 0,02 0,03 0,04 0,05

Energy (MJ)

Primary Energy Electricity

Figure 19: Primary Energy vs. Electricity consumption - per 10,000Gb·km.

Electricity consumption by the terminal station during the use & maintenance phase dominates at 90 percent, as shown in Figure 19. The link between electricity consumption and total primary energy consumption is also apparent, representing over 99 percent of the primary energy count for the use & maintenance phase. Very little electricity is accounted for during the raw materials phase as it is not possible to extract the electricity consumption from the primary energy value for each GaBi database process. However, the total primary energy use for raw materials represents only 7 percent of the total energy use for the system. Electricity used during the end-of-life phase represents the material recycling process.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

HFO (kg) 0 0,0002 0,0004 0,0006 0,0008 0,001 0,0012

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 0,01 0,02 0,03 0,04 0,05

Energy (MJ)

Primary Energy Heavy Fuel Oil

Figure 20: Primary Energy vs. Heavy Fuel Oil consumption – per 10,000Gb·km.

Figure 20 shows that heavy fuel oil (HFO) consumption by ships during the maintenance of the cable presents the largest impact at 53 percent. The link between HFO consumption and total primary energy consumption for the installation phase is apparent. The use & maintenance phase shows that the primary energy count for HFO is much less in relation to electricity. HFO is consumed by the maintenance of the cable with no HFO used by the terminal. No HFO is accounted for during the raw materials phase as it is not possible to extract this from the primary energy value for each GaBi database process. However, again, the total primary energy use for raw materials represents only 7 percent of the total energy use for the system. The relatively high values for HFO for the installation and end-of-life phases at 21 and 23 percent respectively, result from HFO consumption by the cable ship operations in laying and recovering the cable.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3.2. Resource Depletion

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 0,000005 0,00001 0,000015 0,00002

ADP (kg Sb eq.)

Figure 21: Abiotic resource depletion potential - per 10,000Gb·km.

Figure 21 shows that the use & maintenance phase clearly dominates resource depletion at 70 percent. The principal resources for this indicator are crude oil used to produce heavy fuel oil (HFO) for cable maintenance and hard coal (with natural gas impacting to a lesser extent) used to produce electricity consumed at the terminal station. Electricity has the greatest effect on the use & maintenance phase at 71 percent. The end-of-life phase makes a positive contribution due to the recovery and recycling process which consumes both HFO and electricity. Recycling of the materials is assumed to replace 90 percent of the virgin material input, hence the relatively small impact resulting from the raw materials phase.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3.3. Acidification

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 0,00002 0,00004 0,00006 0,00008 0,0001

AP (kg SO eq.) 2

Figure 22: Acidification potential - per 10,000Gb·km.

The two important indicators for Acidification are sulphur dioxide (SO 2) and nitrogen oxides (NO x). Again, the production of electricity and the combustion of heavy fuel oil (HFO) during the use & maintenance phase represent the greatest impact at 56 percent, as presented in Figure 22. HFO combustion during cable maintenance dominates the impact by a factor of 10, over the production of electricity used at the terminal, confirming the high acidification impact of shipping, as presented in the literature. China represents the greatest acidification impact for electricity production by a factor of 12 in relation to US production and 25 to Japan, indicating the high use of coal in China. The relatively high values for the installation and end-of-life phases at 20 percent, are linked directly to the HFO consumption by the cable ship operations, presented in Figure 20.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3.4. Ecosystem Toxicity

MAETP (kg 1,4DCB eq.)

0 0,1 0,2 0,3 0,4 0,5

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

-6,78E-19 0,00003 6E-05 9E-05 0,00012 0,00015

TETP & FAETP (kg 1,4DCB eq.) Terrestric Ecotoxicity Potential Freshwater Aquatic Ecotoxicity Potential Marine Aquatic Ecotoxicity Potential

Figure 23: Ecotoxicity potential - per 10,000Gb·km.

Emissions of heavy metals to water and air represent the major contribution to Ecosystem Toxicity. As shown in Figure 23, the significant use of heavy fuel oil (HFO) and electricity during the use & maintenance phase results in the greatest impact at an average of 73 percent. Selenium and vanadium are the significant contributors from the production of electricity, representing on average 65 percent of the emissions from the use & maintenance phase in relation to terminal energy use. Vanadium and Barium are the major contributors from the production of HFO, while combustion of HFO releases nickel in the greatest amount. Copper, zinc and the pesticide tributyltinoxide are the principal contributors from the maintenance of the ship’s hull. These factors together combine to represent an average of 35 percent of the use & maintenance impact associated with cable maintenance by ship.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3.5. Climate Change

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 0,001 0,002 0,003 0,004 0,005

GWP100 (kg CO eq.) 2

Figure 24: Climate change potential - per 10,000Gb·km.

As shown in Figure 24, actual carbon dioxide (CO 2) emissions to air, from the production of electricity, used at the terminal and the combustion of heavy fuel oil (HFO), consumed by cable maintenance, for the use & maintenance phase, represent the greatest impact at 64 percent. The emission of other climate change gases is insignificant by comparison. The impact from use (electricity) and maintenance (HFO) is relatively even at 47 and 53 percent respectively. This highlights the significantly greater impact on climate change of HFO combustion in relation to electricity, if the primary energy count is compared. The installation and end-of-life phases are characterised by the same impacts to a lesser extent at 13 and 16 percent respectively. Analysis of the electricity mix used in this study, shows that production related CO 2 emissions to air are greater for the US and China by a factor of 1.5 and 2.5 respectively, when compared to production in the EU and Japan.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3.6. Photochemical Ozone Creation

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 0,000001 0,000002 0,000003 0,000004 0,000005

POCP (kg C H eq.) 2 4

Figure 25: Photochemical ozone creation potential - per 10,000Gb·km.

As shown in Figure 25, photochemical ozone creation, or “summer smog”, is linked directly to the combustion of heavy fuel oil (HFO) by ships and hence the use & maintenance phase dominates at 55 percent. As such the impact for the use & maintenance phase is related fully to cable maintenance. The installation and end-of-life phases are similarly affected at 20 percent. Sulphur dioxide (SO 2), nitrogen oxides (NO x) and unspecified non-methane volatile organic compounds (NMVOC) are the significant contributors from HFO combustion.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3.7. Stratospheric Ozone Depletion

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 1E-12 2E-12 3E-12 4E-12 5E-12

OD (kg CFC11 eq.)

Figure 26: Stratospheric ozone depletion potential - per 10,000Gb·km.

The release of halogenated organic (HO) emissions or CFCs, is the indicator for ozone layer depletion. The production of raw materials dominates here at 51 percent, due to the release of HO emissions during raw material processing, as shown in Figure 26. The production of silicon makes the significant contribution (R11, R12 and R114) and, to a lesser extent, copper. The aluminium used for terminal components releases halon during production. HO emissions are also released during the production of heavy fuel oil (HFO), hence the relatively high impact of the installation, use & maintenance and end-of- life phases. Manufacturing, at 16 percent, is impacted by HFO production and also by halon released during the production of thermal energy from gas. The cable and repeaters dominate the raw materials phase accounting for 70 percent of the impact, while the cable dominates the use & maintenance phase accounting for 100 percent of the impact.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3.8. Eutrophication

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 0,000002 0,000004 0,000006 0,000008 0,00001

EP (kg PO eq.) 4

Figure 27: Eutrophication potential - per 10,000Gb·km.

As presented in Figure 27, the use & maintenance phase dominates eutrophication at 56 percent, though not as significantly as other categories. The principal indicator is nitrogen oxide (NO x) emissions from the combustion of heavy fuel oil (HFO). This is reflected in the use & maintenance phase, where cable maintenance by ship represents 90 percent of the impact. Similar to Acidification, the relatively high values for the installation and end-of-life phases, at 20 percent, are linked directly to the HFO consumption by the cable ship operations, presented in Figure 20.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.3.9. Human Toxicity

End of Life

Use & Maintenance

Installation

Manufacture

Raw Materials

0 0,0005 0,001 0,0015 0,002

HTP (kg 1,4DCB eq.)

Figure 28: Human toxicity potential - per 10,000Gb·km.

Emissions of heavy metals to water and air represent the major contribution to Human Toxicity. The significant use of heavy fuel oil (HFO) and electricity during the use & maintenance phase results in the greatest impact at 60 percent, as shown in Figure 28. Nickel and arsenic are the major contributors resulting from the combustion of HFO during cable maintenance. Polycyclic aromatic hydrocarbon (PAH) emissions to seawater, from the maintenance of the ship’s hull, have a lesser impact. These combine to represent 72 percent of the impact for the use & maintenance phase. Selenium is the significant contributors from the terminal station due to the production of electricity, though has a lesser impact to cable maintenance by a factor of three. Again, the relatively high values for the installation and end-of-life phases at 17 and 18 percent respectively, are linked directly to the HFO consumption by the cable ship operations, presented in Figure 20.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.4. Results by Life Cycle Phases The following section presents the results by life cycle phase and in relation to the functional unit of 10,000Gb·km . For each life cycle phase, the results are firstly presented for all impact categories in relation to the sub-models of the cable, repeaters and the terminal station. Secondly, a more detailed analysis of the components of these sub-models is undertaken based on the energy and resource use, acidification and climate change. These areas are of particular interest to Ericsson in their research into the total carbon footprint of the global ICT network and therefore, are the focus of this section. The results highlight the general dominance of the impact from the cable in relation to the terminal station and the relatively insignificant impact of the repeaters.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.4.1. Raw Materials

Primary Energy

Electricity

Heavy Fuel Oil

Abiotic Resource Depletion

Acidification Potential

Freshwater Aquatic Ecotoxicity Potential

Terrestric Ecotoxicity Potential

Marine Aquatic Ecotoxicity Potential

Global Warming Potential

Photochemical Ozone Creation Potential

Ozone Depletion Potential

Eutrophication Potential

Human Toxicity Potential

0% 20% 40% 60% 80% 100%

Cable Repeaters Terminal

Figure 29: Impact distribution for Raw Material sub-models.

Figure 29 shows that the raw materials sub-model for the cable clearly dominates the selected impact categories at over 85 percent. This is traced to the bulk weight of the cable in relation to repeaters and the terminal station. Resource depletion, energy consumption and climate change are shown to be closely linked and relate primarily to the use of primary energy resources (oil, coal and natural gas) during raw material production.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,002 8,00E-07

0,0018 7,00E-07 0,0016 6,00E-07 0,0014 0,0012 5,00E-07 0,001 4,00E-07

0,0008 3,00E-07 0,0006 Sbeq.) (kg ADP Primary Energy (MJ) Energy Primary 2,00E-07 0,0004 0,0002 1,00E-07 0 0,00E+00

Primary Energy Resources

Figure 30: Primary Energy verses Resource Depletion for selected raw material sub-models.

The link between energy consumption and resource depletion is clearly shown in Figure 30, and, is directly linked to the use of primary energy resources (oil, coal and natural gas) during raw material production. The cable visibly dominates over the repeater and terminal station sub-models. At only 3 percent of a cable system, the relatively high energy and resource use for double armour (DA) cable reflects the large material input, as shown in Figure 14.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,002 7,00E-05 0,0018 6,00E-05 0,0016 0,0014 5,00E-05 eq.) 0,0012 2 4,00E-05 0,001 3,00E-05 0,0008 0,0006

Primary Energy (MJ) Energy Primary 2,00E-05 GWP100 (kg CO (kg GWP100 0,0004 1,00E-05 0,0002 0 0,00E+00

Primary Energy Climate Change

Figure 31: Primary Energy verses Climate Change for selected raw material sub-models.

Carbon dioxide equivalent (CO 2 eq.) emissions are directly linked to the use of primary energy resources for the production of raw materials, as shown in Figure 31 . Silicone gel and galvanised steel wire production, particularly, release large amounts of CO 2. Again, the emissions from the cable raw materials clearly dominates.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.4.2. Design & Manufacturing

Primary Energy

Electricity

Heavy Fuel Oil

Abiotic Resource Depletion

Acidification Potential

Freshwater Aquatic Ecotoxicity Potential

Terrestrial Ecotoxicity Potential

Marine Aquatic Ecotoxicity Potential

Global Warming Potential

Photochemical Ozone Creation Potential

Ozone Depletion Potential

Eutrophication Potential

Human Toxicity Potential

0% 20% 40% 60% 80% 100%

Cable Repeaters Terminal

Figure 32: Impact distribution for Design & Manufacturing sub-models.

The result of the design & manufacturing sub-model reveals no clear structure across the selected impact categories, as shown in Figure 32. Primary energy consumption is balanced between cable and terminal and relates to the production of electricity for manufacturing of the cable and the terminal components, particularly the printed board assembly (PBA). Manufacturing of the PBA consumes the greatest amount of electricity, hence the dominance of the terminal sub-model at 45 percent and the relatively large consumption of electricity (15 percent) by the repeater manufacture. Heavy fuel oil (HFO) is used primarily for the cable route survey and a very small amount for the manufacturing of the terminal lead- acid batteries. The combustion of HFO by the research ship is linked to climate change (global warming potential) and is the principal cause for the greater impact of the cable sub-model at 78 percent.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,0008 5,00E-05

0,0007 4,50E-05 4,00E-05 0,0006 3,50E-05 0,0005 3,00E-05 0,0004 2,50E-05

0,0003 2,00E-05 Electricity (kWh) Electricity

Primary Energy (MJ) Energy Primary 1,50E-05 0,0002 1,00E-05 0,0001 5,00E-06 0 0,00E+00

Primary Energy Electricity

Figure 33: Primary Energy verses Electricity for selected manufacturing sub-models.

Figure 33 shows that the link between electricity consumption and primary energy is clearly apparent and relates to the use of primary energy resources (oil, coal and natural gas) for electricity production. The large energy consumption of the submarine line terminal equipment (SLTE) is related to the manufacture of the PBA, particularly the integrated circuit (IC) chips. As mentioned in Section 4.4.3.3, this is likely to be an over-estimation, though clearly dominates over the other components. The high primary energy consumption during the route survey relates to heavy fuel oil (HFO) consumption.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,0012 2,00E-04 1,80E-04 0,001 1,60E-04 1,40E-04 0,0008 eq.) 1,20E-04 2 0,0006 1,00E-04 8,00E-05 0,0004

Primary Energy (MJ) Energy Primary 6,00E-05 GWP100 (kg CO (kg GWP100 4,00E-05 0,0002 2,00E-05 0 0,00E+00

Primary Energy Climate Change

Figure 34: Primary Energy verses Climate Change for selected manufacturing sub-models.

The large impact of the route survey on climate change highlights the significant effect of CO 2 equivalent emissions from the combustion of heavy fuel oil (HFO), as shown by Figure 34. By comparison, the consumption of electricity has no impact on climate change. The production of electricity on the other hand does have an impact, as highlighted by the lesser effect for all other components.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.4.3. Installation

Primary Energy

Electricity

Heavy Fuel Oil

Abiotic Resource Depletion

Acidification Potential

Freshwater Aquatic Ecotoxicity Potential

Terrestrial Ecotoxicity Potential

Marine Aquatic Ecotoxicity Potential

Global Warming Potential

Photochemical Ozone Creation Potential

Ozone Depletion Potential

Eutrophication Potential

Human Toxicity Potential

0% 20% 40% 60% 80% 100%

Cable Repeaters Terminal

Figure 35: Impact distribution for the Installation sub-models.

As discussed earlier, the installation of the terminal building has not been accounted for as it is assumed that it does not make a significant contribution to the results. Repeaters are installed as part of the processes of cable installation and cannot be accounted for separately. As such, cable installation by purpose-built cable ship is the only sub-model accounted for under installation and hence represents the total impact, as shown in Figure 35. Emissions from the combustion of heavy fuel oil (HFO) represent the principal effect for the installation sub-model, with climate change and acidification as important indicators due to the release of carbon dioxide (CO2) , sulphur dioxide (SO 2) and nitrogen oxides (NO X).

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,0018 4,50E-04

0,0016 4,00E-04

0,0014 3,50E-04

0,0012 3,00E-04 eq.) 2 0,001 2,50E-04

0,0008 2,00E-04

0,0006 1,50E-04 Primary Energy (MJ) Energy Primary GWP100 (kg CO (kg GWP100 0,0004 1,00E-04

0,0002 5,00E-05

0 0,00E+00 In Port Manoeuvring Transit Air travel Ship Hull

Primary Energy Climate Change

Figure 36: Primary Energy verses Climate Change for selected installation sub-models.

Figure 36 shows that primary energy consumption is related to the production of heavy fuel oil (HFO), whilst climate change is related to the combustion of HFO. The greatest effect on climate change comes from the emission of carbon dioxide (CO 2) during the HFO combustion process. Air travel is shown to have only a minor impact by comparison.

0,0018 1,40E-05

0,0016 1,20E-05 0,0014 1,00E-05 0,0012

8,00E-06 eq.) 0,001 2

0,0008 6,00E-06

0,0006 SO (kg AP

Primary Energy (MJ) Energy Primary 4,00E-06 0,0004 2,00E-06 0,0002

0 0,00E+00 In Port Manoeuvring Transit Air travel Ship Hull

Primary Energy Acidification

Figure 37: Primary Energy verses Acidification for selected installation sub-models.

Similar to climate change, Figure 37 shows that the acidification effect of heavy fuel oil (HFO) combustion is clearly apparent. The main contribution comes from the emissions of sulphur dioxide (SO 2) and nitrogen oxides (NO X).

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.4.4. Use & Maintenance

Primary Energy

Electricity

Heavy Fuel Oil

Abiotic Resource Depletion

Acidification Potential

Freshwater Aquatic Ecotoxicity Potential

Terrestrial Ecotoxicity Potential

Marine Aquatic Ecotoxicity Potential

Global Warming Potential

Photochemical Ozone Creation Potential

Ozone Depletion Potential

Eutrophication Potential

Human Toxicity Potential

0% 20% 40% 60% 80% 100%

Cable Repeaters Terminal (Maintenance) (Use) (Use) Figure 38: Impact distribution for the Use & Maintenance sub-models.

The results of the use & maintenance sub-model reveals the distinction between heavy fuel oil (HFO) consumption during cable maintenance and electricity use at the terminal, as presented in Figure 38. By comparison, electricity consumption by the cable is minor. Primary energy consumption at the terminal dominates at 67 percent. However, the combustion of HFO results in a much greater impact on climate change and more significantly, acidification. The global warming potential category (GWP100) shows that emissions of carbon dioxide equivalents (CO 2 eq.) are equally shared between use at the terminal (47 percent) and maintenance of the cable (53 percent). This is due to the greater emissions of carbon dioxide

(CO 2), sulphur dioxide (SO 2) and nitrogen oxides (NO X) from the combustion of HFO.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,04 2,50E-03

0,035 2,00E-03 0,03 eq.) 0,025 1,50E-03 2 0,02

0,015 1,00E-03 Primary Energy (MJ) Energy Primary 0,01 CO (kg GWP100 5,00E-04 0,005

0 0,00E+00

Primary Energy Climate Change

Figure 39: Primary Energy verses Climate Change for selected use & maintenance sub-models.

The greater impact on climate change from the combustion of heavy fuel oil (HFO) is apparent in Figure 39. Whilst the terminal clearly consumes more primary energy, ship operations release more carbon dioxide (CO 2) emissions per unit of energy. Again, air travel results in a relatively minor impact.

0,04 3,50E-05

0,035 3,00E-05 0,03 2,50E-05 0,025 eq.)

2,00E-05 2 0,02 1,50E-05 0,015 AP (kg SO (kg AP

Primary Energy (MJ) Energy Primary 1,00E-05 0,01

0,005 5,00E-06

0 0,00E+00

Primary Energy Acidification

Figure 40: Primary Energy verses Acidification for selected use & maintenance sub-models.

The impact of heavy fuel oil (HFO) combustion per unit energy input is even more apparent for acidification, as shown in Figure 40. Ship operations clearly dominate the acidification impact. The main contribution comes from the emission of sulphur dioxide (SO 2) and nitrogen oxides (NO X) during the combustion of HFO.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,01 0,000006

0,000005 0,008

0,000004 0,006 0,000003

0,004 (MJ) HFO

Electricity (MJ) Electricity 0,000002

0,002 0,000001

0 0 Cable Repeaters Terminal

Electricity Heavy Fuel Oil

Figure 41: Electricity verses Heavy Fuel Oil for the use & maintenance sub-models.

The distinction between heavy fuel oil (HFO) consumption during cable maintenance and electricity use at the terminal is apparent in Figure 41. By comparison, the cable itself uses little electricity.

0,04 4,00E-06

0,035 3,50E-06

0,03 3,00E-06

0,025 2,50E-06 eq.) 4 0,02 2,00E-06

0,015 1,50E-06 EP (kgPO EP

Primary Energy (MJ) Energy Primary 0,01 1,00E-06

0,005 5,00E-07

0 0,00E+00

Primary Energy Eutrophication Potential

Figure 42: Primary Energy verses Eutrophication the use & maintenance sub-models.

Figure 42 shows that the impact of heavy fuel oil (HFO) combustion per unit energy input is dominant for eutrophication. Ship operations clearly dominate the eutrophication impact. The main contribution comes from the emission of nitrogen oxides (NO X) during the combustion of HFO. Similarly the emission of NO X from electricity production represents the main impact from the terminal station.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,04 3,00E-01

0,035 2,50E-01 0,03 2,00E-01 0,025

0,02 1,50E-01

0,015 1,00E-01 Primary Energy (MJ) Energy Primary

0,01 eq.) 1,4DCB (kg MAETP 5,00E-02 0,005

0 0,00E+00

Primary Energy Marine Aquatic Ecotoxicity Potential

Figure 43: Primary Energy verses Marine Aquatic Ecotoxicity Potential for the use & maintenance sub-models.

Emissions of heavy metals represent the major contribution to marine aquatic ecotoxicity potential (MAETP) and are dominant for the terminal station, as shown in Figure 43. Heavy metal emissions to freshwater from electricity production are selenium in the greatest amount, with vanadium, nickel and cobalt following. Maintenance of the ship’s hull releases copper in the greatest amount to seawater. The impact from shipping activities is related primarily to nickel released to air from the combustion of heavy fuel oil (HFO).

A review of the freshwater aquatic ecotoxicity potential (FAETP) results reveals a similar picture as MAETP. Emissions of the same heavy metals to freshwater from electricity production represent the major contribution from the terminal energy use. Again, the impact from shipping activities is related primarily to nickel released to air from the combustion of HFO.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.4.5. End-of-Life Decommissioning

Primary Energy

Electricity

Heavy Fuel Oil

Abiotic Resource Depletion

Acidification Potential

Freshwater Aquatic Ecotoxicity Potential

Terrestrial Ecotoxicity Potential

Marine Aquatic Ecotoxicity Potential

Global Warming Potential

Photochemical Ozone Creation Potential

Ozone Depletion Potential

Eutrophication Potential

Human Toxicity Potential

0% 20% 40% 60% 80% 100%

Recovery Cable Materials Repeater Materials Terminal Materials

Figure 44: Impact distribution for the end-of-life sub-models.

The end-of-life phase considers both recovery and recycling of the cable and recycling of some terminal components. The system includes a simplified recycling model based on electricity consumption for material reprocessing. Recovery is assumed to be the exact opposite of the installation process and therefore has the same impacts as discussed in Section 7.4.3. The recycled cable materials are assumed to offset 90 percent of the virgin material input. The results, presented in Figure 44, show the distinction between the combustion of heavy fuel oil (HFO) during recovery and the use of electricity for material reprocessing. As observed in the results for the use & maintenance phase, climate change and acidification impacts are clearly affected by the combustion of HFO and the emissions of carbon dioxide (CO 2), sulphur dioxide (SO 2) and nitrogen oxides (NO X). Recycling of the repeaters and terminal station components has a minor effect by comparison.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,0045 1,00E-03

0,004 9,00E-04

0,0035 8,00E-04 7,00E-04 0,003 eq.) 6,00E-04 2 0,0025 5,00E-04 0,002 4,00E-04 0,0015

Primary Energy (MJ) Energy Primary 3,00E-04 GWP100 (kg CO (kg GWP100 0,001 2,00E-04 0,0005 1,00E-04 0 0,00E+00 Recovery Cable Materials Repeater Materials Terminal Materials

Primary Energy Climate Change

Figure 45: Primary Energy verses Climate Change for selected end-of-life sub-models.

Figure 45 shows that again, the greater impact on climate change from the combustion of heavy fuel oil (HFO) is apparent. Whilst the recycling of cable materials consumes a relatively large amount of primary energy, ship operations during recovery release significantly more carbon dioxide (CO 2) emissions per unit of energy. By comparison, the repeaters and terminal station recycling has little impact.

0,0045 3,00E-05

0,004 2,50E-05 0,0035

0,003 2,00E-05 eq.)

0,0025 2 1,50E-05 0,002

0,0015 1,00E-05 SO (kg AP Primary Energy (MJ) Energy Primary 0,001 5,00E-06 0,0005

0 0,00E+00 Recovery Cable Materials Repeater Materials Terminal Materials

Primary Energy Acidification

Figure 46: Primary Energy verses Acidification for selected end-of-life sub-models.

Similar to the use & maintenance phase, the impact of heavy fuel oil (HFO) combustion per unit of primary energy input is even more apparent for acidification and is clearly dominated by the ship operations, as shown in Figure 46. The main contributors are sulphur dioxide (SO 2) and nitrogen oxides

(NO X) emissions.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

0,0007 0,0000025

0,0006 0,000002 0,0005

0,0000015 0,0004

0,0003 0,000001 (MJ) HFO Electricity (MJ) Electricity 0,0002 0,0000005 0,0001

0 0 Recovery Cable Materials Repeater Materials Terminal Materials

Electricity HFO

Figure 47: Electricity verses heavy fuel oil for the end-of-life sub-models.

The distinction between heavy fuel oil (HFO) consumption during recovery operations and electricity consumption during recycling is apparent in Figure 47. By comparison, repeater and terminal station recycling uses little energy.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

7.5. Data Quality Analysis This section details the data quality analysis undertaken to assess how uncertanties in the model may affect the result. The data quality analyisis thereby gives an indication of the accuracy and robustness of the model. Various scenarios were constructed based on the perceived importance of the process, data gaps or where the quality of the data did not reach the required standard (ISO 14044:2006).

The results of the data quality analysis show that the model is relatively robust. The largest data gaps and uncertainties are presented by the raw material and manufacturing phases for the repeaters and terminal station equipment, which are shown to have no significant effect on the results. The greatest effect on the model results from the use & maintenance phase relating to fuel combustion from ship activities and electricity production. These processes are modelled on measured data from the literature and are assumed to represent the system appropriately.

7.5.1. Sensitivity Analysis – Data gaps and uncertainties The first part of the sensitivity analyisis was undertaken to assess how uncertanties and gaps in the data may affect the result. Five scenarios were constructed and are detailed in Table 26.

Table 26: Description of sensitivity analysis (uncertainties and data gaps) scenarios

Name Description Motivation

Road transportation of raw materials was increased by 500% from 1000km to Assumption/ Raw Transport at 500% 5000km to assess the impact of any likely under estimation. Uncertainty

Road transportation of materials for recycling was increased by 1000% from Assumption/ Recycling Transport at 1000% 100km to 1000km to assess the impact of any likely under estimation. Uncertainty

Repeaters internal electronics The internal electronics modelled by an assumed comparable process of a printed Data gap / at 400% board assembly was increased from 8.5kg to 35kg to assess the impact. Uncertainty

Repeaters raw materials, Repeaters were removed entirely from the model to test the impact of repeaters Data gap Manufacturing and E-o-L at 0% on the final result.

Terminal raw materials, The raw materials and manufacturing of the terminal station were moved from Data gap / Manufacturing and E-o-L at 0% the model. Uncertainty

For each scenario, the studied input parameter was varied and the result recalculated. The results are given as a percentage deviation for each impact category from the original result. The results are presented graphically in Figure 48 and are tabluated in full in Appendix F.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

100%

80%

60%

40%

20%

-20%

-40%

-60%

-80%

-100%

Primary Energy Electricity Heavy Fuel Oil Abiotic Resource Depletion Acidification Potential Freshwater Aquatic Ecotoxicity Potential Terrestric Ecotoxicity Potential Marine Aquatic Ecotoxicity Potential Global Warming Potential Photochemical Ozone Creation Potential Ozone Depletion Potential Eutrophication Potential Human Toxicity Potential

Figure 48: Results of sensitivity analysis (uncertainties and data gaps)

The results of the sensitivity analysis reveal that the processes considered to have the greatest uncertainty for the raw materials, design & manufacturing and end-of-life phases have limited effect on the final result.

Transportation of the raw materials was examined due to the uncertainty introduced by the assumption that all raw materials were road transported 1,000 kilometres. Given the nature of the specialised components, it is possible that this is an under estimate. The analysis shows that transport of raw materials does not have a significant effect on the result. Increasing the road transport distance by 500 percent to 5,000 kilometres, results in an increase in impact from 0.05 to 2.1 percent over all categories. A similar result is returned for the transport of materials to be recycled. Increasing the road transport distance by 1,000 percent from 100 to 1,000 kilometres, to account for any under estimate, results in an increase in impact from 0.1 to 5.0 percent.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

With an estimated 200 submarine repeaters per 10,000 kilometres, it was assumed that repeaters would make a significant contribution to the overall environmental impact of the cable. Repeaters represent a data gap that resulted in assumptions being made from similar processes for all materials other than the repeater housing, which was assumed to be beryllium copper alloy. Firstly, the assumption based on the printed board assembly (PBA) was explored. Due to uncertainty in the assumption and a possible under estimate, the amount of PBA in the repeater was increased from 8.5 kilograms to 35 kilograms. The results show a minimal effect of approximately one percent. Secondly, the repeaters were removed from the model entirely to test the effect on the result. Ozone depletion potential is reduced by 16 percent, while all other impact categories are affected by less than 0.6 percent. The change in ozone depletion is related mainly to the aluminium production process and the reduced raw material input. The process of aluminium production is taken from the GaBi database and is assumed to be accurate.

A similar case is apparent for the removal of the terminal raw materials, design & manufacturing and end- of-life phases. Ozone depletion potential is affected by 19 percent and linked mainly to the cabinets, both the reduced aluminium input and to a lesser extent the thermal energy used during the manufacture process. Again, aluminium and thermal energy production are taken from the GaBi database and assumed to be accurate. All other categories were affected by less than 1.5 percent. Other than ozone depletion, the sensitivity analysis shows that the assumptions made for the weight of PBA and cabinets contained within a terminal do not significantly affect the results.

7.5.2. Sensitivity Analysis – Methodological Choices The second part of the sensitivity analyisis was undertaken “to determine how changes in data and methodological choices” may affect the result (ISO 14044:2006, p22). Various scenarios were constructed based on identified key processes, particullary during the use & maintenance phase. The sensitivity analysis scenarios are detailed in Table 27.

Table 27: Description of sensitivity analysis (methodological choices) scenarios

Name Description Motivation Terminal electricity consumption during the use phase was reduced by 30% to Terminal Electricity at 70% assess the impact of any likely over estimation due to upgrading to more efficient Uncertainty components. The electricity model was replaced by the standard database process for EU-25 electricity production electricity production in the EU-25 countries to assess the validity of the regional Key process model used in this study.

Cable Maintenance by ship at Annual maintenance (per 1000 kilometre) of the cable was increased from 7.41 Key process 150% ship days to 11.12 ship days.

Ship Emissions to Marine Emission factors for ships were changed from heavy residual oil (RO) to the less Key process Distillate fuel utilised, yet cleaner marine distillates (MD).

Lifetime 25 years at The results of the 25 year technical lifetime of the cable were assessed relative to Key process 10000Gb.km relative the functional unit of 10000Gb·km of cable and compared to the original case.

The end-of-life phase undertaking recovery and recycling of the cable was No recovery at E-o-L Key process removed from the model, thereby increasing the raw material input to 100%.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

100%

80%

60%

40%

20%

-20%

-40%

-60%

-80%

-100%

Figure 49: Results of sensitivity analysis (methodological choices)

The results of the sensitivity analysis reveal that the activities and methodological assumptions made for the use & maintenance phase affect the model significantly.

Electricity consumption at the terminal during the use & maintenance phase was examined due to the large impact generated, as shown in Section 7.4.4. Whilst the energy consumption is considered to be well defined, based on the actual usage at two terminals (with good agreement), it was interesting to consider the impact of potential energy savings due to the upgrading of terminal components. As such, electricity consumption at the terminal was reduced by 30 percent. All resultant changes are directly related to the production of electricity. The largest effect was to freshwater aquatic ecotoxicity at 21 percent, which is linked to the release of heavy metals to freshwater. Marine aquatic ecotoxicity at 14 percent and terrestrial ecotoxicity at 9 percent change are also directly affected by the release of heavy metals. Resource depletion is affected by 15 percent, linked to the use of primary energy resources, such as crude oil and coal. Climate change potential is affected by 9 percent related to carbon dioxide and methane emissions. Less significant change is observed in all other impact categories.

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

In order to assess the validity of the regional electricity mix modelled by a previous study at Ericsson, the electricity sub-model was changed to the standard database process for the EU-25 countries. No process was available in the GaBi software that modelled a world electricity mix, therefore the EU-25 production process was selected. The most significant change relates to ozone depletion potential category at a relative increase of 5000 percent and relates directly to the emissions of CFC gases. Secondly, marine aquatic ecotoxicity potential (MAETP) results in a 65 percent increase. This relates primarily to the emission to air of hydrogen fluoride, which dominates the EU-25 impacts and is not modelled in the Ericsson regional sub-models. In contrast, the Ericsson study models a far greater amount of heavy metal emissions to freshwater based on regional differences in production. Freshwater aquatic ecotoxicity potential (FAETP) is reduced by 76 percent, which relates directly to the modelling of heavy metal emissions to freshwater. These emissions are greater by a factor of 45 in the Ericsson study when compared to the EU-25 production process. Primary energy is reduced by 13 percent and abiotic resource depletion by 19 percent. With the exception of terrestrial ecotoxicity potential (TETP) at 21 percent, all other categories show a change of less than 15 percent. The Ericsson study is based on extensive research into the regional differences in electricity production (3GLCA, 2002), while the assumptions and data source of the EU-25 standard database process are unknown. The differences in impact between the two electricity production methods likely reflects these regional differences and the differences in data sources.

Shipping activities are identified to be key processes over the 13 year lifetime of the cable. The analysis shows that a 50 percent increase in the annual maintenance (per 1,000 kilometres) with ships from 7.41 ship days to 11.12 ship days, results in significant increase in almost all impact categories. Acidification and eutrophication are both affected by 25 percent and are directly linked to fuel combustion and the emissions thereof, particularly sulphur dioxide (SO2) and nitrogen oxides (NO X). Climate change is affected by 17 percent directly linked to carbon dioxide (CO 2) emissions. Emission factors are based on the more conservative figures for the heavier residual oil (RO) emissions. When these are replaced with emissions factors for the cleaner marine distillates (MD), the results are again significantly affected. A comparison of the emission factors can be found in Cooper and Gustafsson (2004). Human toxicity, terrestrial ecotoxicity and photochemical ozone creation are reduced by between 43 to 69 percent, linked to the reduced emissions of heavy metals. Acidification is reduced by 47 percent, primarily linked to the reduction of SO 2 and NO X. In contrast, the impact on climate change is increased by 2% given the slight increase in the emission of CO 2. The results of these two sensitivity analyses highlight the significant affect of shipping on the overall environmental impact of the cable system. The number of maintenance days (normalised to 1000 kilometres of cable) is calculated from actual data for standby and operation times. An assumption is made for the operation mode (transit, manoeuvring, in port) during repair operations, however this is not considered to introduce significant uncertainty into the results. Emission factors for ships are well established from the literature. The use of RO emission factors increases the impact significantly for the majority of categories, however RO is the most frequently used fuel in the shipping industry and accounts for the worst case scenario, therefore, is considered appropriate.

The use & maintenance period for the cable is shown to dominate the impact in almost all categories. Consequently, it is appropriate to investigate how the result is affected if the lifetime of the cable is increased from the commercial lifetime of 13 years to the documented technical lifetime of 25 years. All impacts, excluding ozone depletion, were reduced by between 4 and 23 percent. Ozone depletion was affected by 39 percent due to the high influence of raw material production on this impact category. The results clearly highlight the reduced environmental impact of increasing the in-service lifetime of the cable.

A sensitivity analysis was undertaken on the effect of simply abandoning the cable at decommissioning, resulting in no end-of-life impacts, yet requiring 100 percent virgin material input. The analysis shows that ozone depletion is affected significantly at 126 percent, due the influence of the processing of raw materials on this impact category. Surprisingly, with an increase in virgin materials from 10 to 100 percent,

7. Results of Life Cycle Interpretation A life cycle assessment o f fibre optic submarine cable systems

resource depletion is increased by only 6 percent and primary energy by 3 percent. Electricity is reduced by 3 percent and heavy fuel oil (HFO) is reduced by 23 percent. Electricity and HFO are somewhat problematic for the raw materials phase as they are not tracked in the standard database processes. What is further surprising is that the impacts for all other categories are reduced by between 1 and 19 percent. This reduction is related to the removal of the end-of-life phase and the reduction of impacts from fuel combustion during recovery and electricity consumption during the recycling process, which appear to have a greater impact than raw material processing.

7.6. Normalisation The results of the normalisation calculation illustrate the comparison between the annual environmental impact of one person’s data traffic in relation to the total annual environmental impact determined by the normalisation factors presented in Section 5.4.

In order to perform the normalisation calculation, based on the annual environmental impact, the annual per capita usage of a submarine cable system must first be estimated. Cisco Systems track and forecast the global IP traffic and estimate that 50 exabytes (50 x 10 18 EB) of data was sent during 2006 (Cisco, 2008). They also predict that global IP traffic will grow rapidly, driven by internet video. Dividing the global IP traffic by the total world population of 6.12 billion (UN, 2009), an average figure of 8.7 gigabytes (GB) of data are sent annually per person. While it is unlikely that all of this data would pass through the submarine network, for the purpose of normalisation, it is assumed that the total 50EB of data is transferred via a submarine cable. As a comparison, the data traffic generated by the average US adult is also estimated. Cisco (2008, p14) state that the “average US adult consumes the equivalent of nearly 120 GB per month.” This translates to an annual consumption of 1440GB of data. Again, it is assumed that all of this data is sent, at some point, via a submarine cable. These two usage figures (8.7GB and 1440GB) have then applied been to the results to estimate the potential environmental impact of one person’s annual use of a submarine cable. It is assumed that this data is sent over 10,000 kilometres of cable, thereby relating to 1Gb or 0.125GB of data (accepting 1 byte equals 8 bits).

The results of the normalisation calculation are presented in Table 28 and are expressed as a percentage of the total annual environmental impact per capita, globally. Again, the results are only presented for the three impact categories with the greatest certainty, as presented by Sleeswijk et al (2008).

Table 28: Normalisation results per person per year for 10,000km of cable.

Worldwide average US average Impact Category (8.7GB/person/year) (1440GB/person/year)

Acidification (500 years) 0.16% 2.6%

Global Warming (100 years) 0.007% 1.2%

Photochemical Ozone Creation 0.001% 0.15%

The results of the normalisation show a relatively low normalised impact. If the annual data consumption of the average US adult is sent via submarine cable, then the potential acidification impact is calculated at 2.6 percent of their total potential annual impact. For climate change and photochemical ozone creation, the figure is even less at 1.2 and 0.15 percent respectively. Accepting the limitation of the normalisation calculation, these figures indicate that sending data via submarine cables has a very low potential environmental impact in relation to the background impact of the two groups.

8. Discussion A life cycle assessment o f fibre optic submarine cable systems

8. Discussion

This section presents a discussion of the important issues based on the results of this study.

The results show that the use & maintenance phase clearly dominates all impact categories at an average of 66 percent. By comparison, the raw materials and design & manufacturing phases account for, on average, only 6 percent of the total potential impact. This clearly highlights that the greatest impact over the life cycle of a submarine cable system comes from the use & maintenance activities. Namely, electricity use at the terminal to power the terminal equipment and the combustion of marine fuel during cable maintenance with purpose-built ships. These are two key activities relating to the environmental performance of the cable system. Ericsson has a particular focus on climate change and analysis of the global warming potential (GWP100) impact category for the use & maintenance phase, shows that the emissions of carbon dioxide equivalents (CO 2 eq.) are equally shared between use at the terminal (47 percent) and maintenance of the cable (53 percent). However, the impact, per unit of primary energy input, from the combustion of marine fuel oil has a far greater impact on climate change than the impact from electricity use. This reflects the disparity in the environmental impacts of electricity and fossil fuel consumption. The process of electricity production is modelled on the four regional models from the Ericsson database. The effect on the result of using this mix has been explored in the sensitivity analysis, which shows that regional differences in electricity production do have a significant weight on the final result. Heavy use of coal for production, in both China and the US, is likely to be balanced against the cleaner production of Japan and the EU. The regional mix from the Ericsson database is considered robust and appropriate for this study. If however, the aim is to reduce the environmental burden of a cable system further, then electricity produced from renewable sources could be considered by cable owners. Submarine cable terminal stations are, by nature, close to the coast and in some regions, there may be a possibility to take advantage of renewable energy generated by wind, wave and tidal streams. It is not within the scope of this study, to assess the potential reduction in impact this could bring, however, this could be an area for future study.

The production of heavy fuel oil (HFO), on the other hand, is biased toward US and EU production and is likely to be a best case scenario. Adding production processes from other regions would strengthen the LCA model, however this data was not available. Nevertheless, the results show that it is emissions from the combustion of HFO that have a greater impact on the final result. Combustion emissions are well modelled in the literature and it is more the methodological choice of using residual oil (RO) emission factors than the cleaner marine distillates (MDs) factors that significantly affects the results. Acidification and photochemical ozone creation are two impact categories, particularly affected by a reduction of over 40 percent, while climate change is relatively unchanged. This is linked directly to the low sulphur content of the MD fuels, while the release of carbon dioxide (CO 2) from combustion remains largely the same. The calculation of fuel consumption for the average cable ship is based on the fleet of one particular company. This company actually uses the cleaner MD fuels, therefore, emission reductions are currently being made in this area. However, the majority of the world’s shipping fleet use the heavier residual oil (RO) fuels and therefore, the emission factors for RO fuel have been used in this study, allowing for a worst case scenario with regard to the impacts from shipping. Nitrogen oxide (NOx) controls on ships are also another solution to reduce ship emissions, though they have not been considered by this study as they are not common in older ships. If however, the focus is on climate change, then both fuels have similar emissions of carbon dioxide equivalents (CO 2 eq.) and hence similar environmental burdens.

8. Discussion A life cycle assessment o f fibre optic submarine cable systems

These processes of electricity production, HFO production and HFO combustion have the greatest affect on the result and as such, are the areas where future improvement of the model could be focused and the greatest gains could be made in improving the environmental performance of the cable systems per unit of data.

By comparison, the low impact for the raw materials and the design & manufacture phases, at only 6 percent of the total impact, is reflected partly in the recycling of the cable materials and partly in the simplified manufacturing sub-model for the cable. Recycling of the cable materials accounts for 90 percent of the virgin material input, thereby, providing a significant reduction in the impact from extraction and processing of the raw materials. However this gain is offset by the recovery and recycling processes, which have a greater environmental burden in relation to the extraction and processing of the equivalent 90 percent of raw materials. The sensitivity analysis for the end-of-life phase highlights this slightly poorer environmental performance for the recovery and recycling scenario in relation to simple decommissioning and abandonment. With increasing environmental accountability and likely increases in commodity prices, it is expected that recovery and recycling of cables will become more common. Also, tests have shown that recovery and reuse of the system is achievable, thereby substituting the raw materials and manufacturing process completely. Some uncertainty does exist in this sub-model as the recycling process is simplified to the electricity usage for recycling of the various materials. No emissions, other than those from electricity production, are accounted for. This represents a possible under estimation, however, it is likely that the emissions from electricity production represent the greatest impact on a global scale. Given the slightly poorer environmental performance of cable recovery and recycling, it is still considered appropriate to base the model on this scenario. Likewise, the manufacturing process of the cable is simplified to the electricity consumption of the specific plant used to assemble the cable. No allocation of the total emissions from the cable factory has been attempted, therefore, the manufacturing process is also likely to be an underestimation. Though, based on the site visit to the cable manufacturing plant, cable assembly is a relatively simple and benign process with few emissions. Again, it is likely that the emissions from electricity production represent the greatest impact. Emissions from the manufacturing and the end- of-life recycling processes are another area where future improvement of the model could be achieved.

Upgrading plays a significant role in magnitude of the potential impacts and is connected to the usage of the cable. Modern systems are designed to be upgraded without replacing the cable itself. Upgrading increases the system capacity by replacing transmission and receive components at the terminal station. This study accounted for three upgrades over the system lifetime in a simplified linear model based solely on the material and energy consumption of component manufacturing. While the literature shows that, on average, 85 percent of the design capacity is currently not lit, no attempt was made in this study to assess the impact based on an increased capacity due to this significant future upgrading potential. It has been stated that technology is moving rapidly from 10Gbps to 40Gbps transmission, with commercialisation of the later likely soon. While the overall environmental impact of the cable system over its lifetime would not change significantly, clearly this will have a significant effect on reducing the environmental burden per unit of data. Assuming that the energy consumption remains the same at the terminal station, then the impacts would be reduced by a factor of four. With the leap to 40Gbps, it may be that cables remain in service longer than 13 years as the maintenance costs per unit of data are reduced. Or perhaps, the extra bandwidth may simply be soaked up by the expanding digital video market and greater consumer demand as predicted by Cisco. With greater capacity the impact in relation to the functional unit of 10,000Gb·km would be greatly reduced, however, the actual normalised impact per capita may not change significantly as our personal data consumption is likely to increase in line with the available capacity.

As the functional unit is based on usage, the gap between lit capacity and usage affects the environmental performance of the system. It is shown that on average, 25 percent of the current installed capacity is actually used. The same energy is consumed if the cable is utilised at 25 percent capacity, or if it is

8. Discussion A life cycle assessment o f fibre optic submarine cable systems

exploited to its full capacity. The maintenance requirements with ships are the same, as are the energy requirements of the land terminal station. Therefore, subject to technical and commercial limitations, if this gap between usage and lit capacity was reduced, a subsequent gain in environmental performance per data unit would be achieved. Furthermore, increased system usage, in this case increased total data traffic, reduces the resulting potential environmental impacts. The sensitivity analysis supports this conclusion and shows that increasing system usage over the 25 year technical lifetime of a submarine cable system reduces the potential environmental impact. From a life cycle perspective, the longer a cable remains in service, the superior the environmental performance per unit of data. Used capacity and service life therefore have a significant effect on the results and are the two key areas that affect the resulting environmental impacts per unit of data.

The sensitivity analysis reveals how the limitations of the study affect the final result. Data gaps, data uncertainty and methodological choices have been analysed to determine their effect. Repeaters represent a total data gap, which has been filled by assumptions with regards to similar processes. The terminal station model was constructed from similar processes from previous studies. Furthermore, the components of the terminal have been estimated from a terminal specification from the year 2000. As such, there is likely to be some over estimation, principally in the construction of the submarine line terminal equipment (SLTE) sub-model. It has been shown that the technology is moving fast, and it is likely that during the last nine years, efficiencies in equipment size and energy consumption have been made. Methodological choices made in this study have an effect with regard to electricity production, heavy fuel oil combustion (HFO) and commercial lifetime of the cable. The sensitivity analysis shows however, that the effect of the most significant assumptions and limitations on the final result are not particularly significant and that the effect of the methodological choices is approximately 20 percent over all categories. This indicates that the model is relatively robust.

The results of the normalisation calculation show that the relative per capita environmental burden is small for global warming and photochemical ozone creation. Acidification is slightly higher and is related to the combustion of marine fuel. As explained earlier in the discussion, the methodological choice of using the high sulphur content fuels directly affects the acidification impact category and thus similarly affects the normalisation results.

9. Conclusion A life cycle assessment o f fibre optic submarine cable systems

9. Conclusions

This Life Cycle Assessment (LCA) has been undertaken in order to fill, what appears to be, a gap in knowledge regarding the environmental impacts of a submarine cable system. Submarine cable systems make it possible to transfer large amounts of data around the globe almost instantaneously, yet, little was known about the potential environmental impacts from a life cycle perspective. The main conclusions from this study are:

• The use & maintenance phase of the life cycle of a submarine cable system clearly dominates the potential environmental impacts at an average of 66 percent.

• The key potential impacts result from electricity use at the terminal and the combustion of marine fuel onboard the cable ship during the cable maintenance . For climate change, these are roughly balanced at 47 and 53 percent respectively.

• Maintenance of the cable has the greatest impact on climate change per unit of primary energy due to the emissions from the combustion of marine fuel onboard the cable ship.

• Seven grams of carbon dioxide equivalents (7g CO 2 eq.) are released per 10,000 gigabit kilometres (Gb.km).

• Based on the functional unit of sending data over the cable system, increasing the commercial lifetime of a system or increasing data traffic through greater used bandwidth, reduces the environmental impact per unit of data .

• The data quality analysis shows that the most significant data gaps and uncertainties in the LCA model do not affect the result significantly. Methodological choices affect the LCA model by approximately 20 percent. This indicates that the model is relatively robust.

• Normalisation shows a small relative impact when compared to the annual global impact per person.

10 . Recommendations A life cycle assessment o f fibre optic submarine cabl e systems

10. Recommendations and Future Improvements and Use of the Model

The results show that sending data via submarine cable has a low environmental impact in relation to the normalised impact on a global scale. However, if the aim is to reduce the environmental impact of these systems further, then the use & maintenance phase is the area where companies should focus their efforts. In particular, electricity use at the terminal and the emissions from the cable ships. The greatest gain is likely to be achieved with the reduction of ship emissions as these appear to have the most significant impact per unit of primary energy. Service lifetime and used bandwidth are also key parameters. An increase in either results in a corresponding decrease in the potential impact per unit of data. These are particular areas where cable owners could focus if environmental improvement is a goal.

The regional electricity models from the 3GLCA database could be revised to assess if the emissions of CFC gases resulting in potential ozone depletion are modelled sufficiently. This impact category presented the greatest change in the sensitivity analysis at 5000 percent. These models could also be updated with current production figures. All other changes in relation to the sensitivity variables were within acceptable and reasonable limits.

Improvements in the LCA model, for this study, should focus on improving the electricity production and the heavy fuel oil (HFO) production sub-models. Modelling of HFO could be expanded to provide a more reliable result based on a world production mix. Daily fuel consumption figures could also be strengthened with data from other companies operating cable ships. These are two key processes within the LCA model. Other areas include the data gaps and the assumptions made regarding similar processes from previous LCA studies, which represent uncertainty in the model. The most significant data gap relates to the submarine repeaters. Uncertainty in the terminal equipment could be reduced with a more up-to-date specification of a modern terminal system. The current terminal specification was from the year 2000 and much is likely to have changed in the following decade regarding size and efficiency of the submarine line terminal equipment (SLTE). The recycling model could be expanded to include wastes and emissions from the various material recycling processes. This would reduce uncertainty in the end-of-life phase and provide a more certain comparison to the “no recovery” scenario, where 100 percent virgin material is used for the system. Likewise, modelling of emissions and wastes from cable manufacturing could be enhanced.

Accepting the assumptions and delimitations of this study, the submarine cable LCA model can be used to predict the potential impacts of more specific cable systems. Two sub-models for the cable and the terminal station were developed in parallel to allow for versatility and scaling of the system based on the length of cable and the number of terminals. With this achieved, the actual used capacity of the specific system could be applied to provide a more precise impact result per functional unit.

This study represents a snapshot of the potential environmental impacts based on today’s technology. It is also a simplification of reality and these points must be taken into consideration when interpreting the results.

11 . Terminology A life cycle assessment o f fibre optic submarine cable systems

11. Terminology

1,4DCB 1,4 Dichlorobenzene Mbps Megabits per second

3G Third Generation (mobile telephone network) MD Marine Distillates

A Ampere ME Main Engine

AC Alternating Current MGO Marine Gas Oil

ADP Abiotic Depletion Potential MJ Megajoules

AE Auxiliary Engine NME Network Management Equipment

AP Acidification Potential NMVOC Non -Methane Volatile Organic Compound

BeCu Beryllium Copper Alloy NO x Nitrogen Oxides (gases)

CML Institute of Environmental Sciences, University of ODP Ozone Depletion Potential Leiden CO Carbon Monoxide PAH Polycyclic Aromatic Hydrocarbons

CO 2 Carbon Dioxide PC Personal Computer

DA Double Armour (cable) PFE Power Feed Equipment

DC Direct Current pkm Person Kilometre

EAF Electric Arc Furnace PM Particle Matter

EB Exabytes PO 4 Phosphate

EP Eutrophication Potential POCP Photochemical Ozone Creation Potential eq. Equivalent RO Residual Oil

EU European Union SA Single Armour (cable)

IP Internet Protocol SAT -1 South Atlantic (Cable No:) 1

ISO International Organisation for Standardisation Sb Antimony (metal) kbps Kilobits per second SDH Synchronous Digital Hierarchy kg Kilogram SLTE Submarine Line Terminal Equipment km Kilometres SO 2 Sulphur Dioxide (gas)

KTH Kungliga Tekniska högskolan SOx Sulphur Dioxides (gases) kV Kilovolts t tonne kW Kilowatt TAT -14 Trans -Atlantic Telephone (Cable No:)14 kWh Kilowatt hour TETP Terrestrial Ecotoxicity Potential

LCA Life Cycle Assessment TLA Terminal Line Amplifier

LCI Life Cycle Inventory UK United Kingdom

LCIA Life Cycle Impact Assessment US United States (of America)

LME Line Monitoring Equipment V Volt

LW Lightweight (cable) VOCs Volatile Organic Compounds

LWP Lightweight Protected (cable) WDM Wave Division Multiplexing m Metres Wh Watt hour

MAETP Marine Aquatic Ecotoxicity Potential WTE Wave Terminating Equipment

12 . List of Tables A life cycle assessment o f fibre optic submarine cable systems

12. List of Tables

Table 1: Comparison between satellite and submarine cable communication (Adapted from Barattino and Koopalethes, 2007; NEC, 2008)...... 15 Table 2: Additional categories used for the life cycle impact analysis...... 21 Table 3: Distribution of electricity generation processes (Adapted from 3GLCA, 2002; Bergelin, 2008) ...... 28 Table 4: Selected engine emission factors (g/kWh) for Residual Oils (Cooper and Gustafsson, 2004)...... 30 Table 5: Average Cable Ship engine load characteristics by operational mode. (Cooper and Gustafsson, 2004) ...... 31 Table 6: Ratio of cable types in the generic system...... 33 Table 7: Estimated weights of terminal components...... 36 Table 8: Engine Power: Average Research Ship (calculated from GMSL, 2009; confidential source)...... 37 Table 9: Daily Fuel Consumption: Average Research Ship (calculated from confidential source)...... 38 Table 10: Average Research Ship: typical survey mission normalised to 1,000km of cable (calculated from confidential source). .. 38 Table 11: Engine Power: Average Cable Ship (calculated from GMSL, 2009)...... 41 Table 12: Daily Fuel Consumption: Average Cable Ship (calculated from GMSL, 2009)...... 41 Table 13: Average Cable Ship: typical installation mission normalised to 1,000km of cable (calculated from confidential source). 41 Table 14: Average Cable Ship: Estimated operation mode per cable repair...... 43 Table 15: Average Cable Ship: Annual repair mission normalised to 1,000km of cable (confidential source)...... 43 Table 16: Energy consumption – steel recycling process (Jones, 2009; BlueScope, 2009; Corus, 2007; World Steel, 2008)...... 45 Table 17: Energy consumption – aluminium recycling process (Green, 2007; Schmitz et al , 2006)...... 46 Table 18: LCI result summary for submarine cable system ...... 48 Table 19: CML impact categories used for the life cycle impact analysis (Adapted from Guinée et al, 2004)...... 51 Table 20: World normalisation factors, person equivalents per year (Adapted from Sleeswijk et al, 2008)...... 54 Table 21: Calculation summary for the generic cable system (Adapted from Ruddy, 2006)...... 55 Table 22: Summary of capacity calculation for the generic cable system (Adapted from Ruddy, 2006)...... 56 Table 23: Summary of results – per 10,000Gb·km...... 59 Table 24: Comparison of results – Climate change: Example 1 (Adapted from 3GLCA, 2002; European Commission, 2007; Jonsson, 2009b)...... 60 Table 25: Comparison of results – Climate change: Example 2 - Single person, 2 day meeting. (Adapted from 3GLCA, 2002; Jonsson, 2009b)...... 60 Table 26: Description of sensitivity analysis (uncertainties and data gaps) scenarios ...... 87 Table 27: Description of sensitivity analysis (methodological choices) scenarios ...... 89 Table 28: Normalisation results per person per year for 10,000km of cable...... 92

13 . List of Figures A life cycle assessment o f fibre optic submarine cable systems

13. List of Figures

Figure 1: Report structure ...... 3 Figure 2: LCA stages (ISO 14040:2006, p8) ...... 4 Figure 3: World map of submarine cables (Alcatel, 2009) ...... 7 Figure 4: Development of submarine cables (Hilt, 2009) ...... 8 Figure 5: Unrepeated "festoon" system (adapted from Alcatel, 2009) ...... 9 Figure 6: Branched system architecture (adapted from Alcatel, 2009) ...... 10 Figure 7: Ring system architecture (adapted from Alcatel, 2009) ...... 10 Figure 8: Components of a submarine cable system (Letellier, 2004) ...... 11 Figure 9: Types of submarine cable: Double Armoured (DA), Single Armoured (SA), Lightweight Protected (LWP) and Lightweight (LW) (Beaufils, 2000)...... 12 Figure 10: Architecture of the SLTE (Breverman et al , 2007) ...... 13 Figure 11: Life Cycle stages of a submarine cable (Adapted from USEPA, 2006)...... 17 Figure 12. Graphical representation in GaBi, built on a modular system (PE & LBP,2009)...... 22 Figure 13: Basic structure of the life cycle of a submarine cable system...... 25 Figure 14: Total weight verses weight of steel per 1,000 metres of cable...... 32 Figure 15: Distribution of all other raw materials (excluding steel) by cable type...... 33 Figure 16: Distribution of raw materials in the terminal station...... 35 Figure 17: Diagram of Submarine Line Terminal Equipment (SLTE) (Adapted from Markow, 2009)...... 36 Figure 18: Summary of results – per 10,000Gb·km...... 58 Figure 19: Primary Energy vs. Electricity consumption - per 10,000Gb·km...... 61 Figure 20: Primary Energy vs. Heavy Fuel Oil consumption – per 10,000Gb·km...... 62 Figure 21: Abiotic resource depletion potential - per 10,000Gb·km...... 63 Figure 22: Acidification potential - per 10,000Gb·km...... 64 Figure 23: Ecotoxicity potential - per 10,000Gb·km...... 65 Figure 24: Climate change potential - per 10,000Gb·km...... 66 Figure 25: Photochemical ozone creation potential - per 10,000Gb·km...... 67 Figure 26: Stratospheric ozone depletion potential - per 10,000Gb·km...... 68 Figure 27: Eutrophication potential - per 10,000Gb·km...... 69 Figure 28: Human toxicity potential - per 10,000Gb·km...... 70 Figure 29: Impact distribution for Raw Material sub-models...... 72 Figure 30: Primary Energy verses Resource Depletion for selected raw material sub-models...... 73 Figure 31: Primary Energy verses Climate Change for selected raw material sub-models...... 74 Figure 32: Impact distribution for Design & Manufacturing sub-models...... 75 Figure 33: Primary Energy verses Electricity for selected manufacturing sub-models...... 76 Figure 34: Primary Energy verses Climate Change for selected manufacturing sub-models...... 77 Figure 35: Impact distribution for the Installation sub-models...... 78 Figure 36: Primary Energy verses Climate Change for selected installation sub-models...... 79 Figure 37: Primary Energy verses Acidification for selected installation sub-models...... 79 Figure 38: Impact distribution for the Use & Maintenance sub-models...... 80 Figure 39: Primary Energy verses Climate Change for selected use & maintenance sub-models...... 81 Figure 40: Primary Energy verses Acidification for selected use & maintenance sub-models...... 81 Figure 41: Electricity verses Heavy Fuel Oil for the use & maintenance sub-models...... 82 Figure 42: Primary Energy verses Eutrophication the use & maintenance sub-models...... 82 Figure 43: Primary Energy verses Marine Aquatic Ecotoxicity Potential for the use & maintenance sub-models...... 83 Figure 44: Impact distribution for the end-of-life sub-models...... 84 Figure 45: Primary Energy verses Climate Change for selected end-of-life sub-models...... 85 Figure 46: Primary Energy verses Acidification for selected end-of-life sub-models...... 85 Figure 47: Electricity verses heavy fuel oil for the end-of-life sub-models...... 86 Figure 48: Results of sensitivity analysis (uncertainties and data gaps)...... 88 Figure 49: Results of sensitivity analysis (methodological choices) ...... 90

100

14 . References A life cycle assessment o f fibre optic submarine cable systems

14. References

14.1. Public References

ABB, 2009a. Environmental Product Declaration: Converter Module DCS 400. Online at www.abb.com (accessed on 2009-05-25). ABB, 2009b. Environmental Product Declaration: Synchronact 5, Synchronizing and Paralleling Equipment and System. Online at www.abb.com (accessed on 2009-05-25). ABB, 2009c. ABB DC drive: DCS800-A, 18 A to 9800/19600 A, Enclosed converters. Online at www.abb.com (accessed on 2009-05-25). ABB, 2009d. Environmental Product Declaration for AC-generator type AMG 0900 5125 kVA power. Online at www.abb.com (accessed on 2009-05-25). Akiba, S. and S. Yamamoto, 1998. WDM Undersea Cable Network Technology for 100 Gb/s and Beyond. Optical Fiber Technology. Vol. 4, pp.19-33 (1998). Article No. OF970236. Alcatel, 1998. Submarine Power Feed Equipment (PFE). Product specification. Online at http://www1.alcatel- lucent.com (accessed on 2009-05-25). Alcatel, 2009. How Does It Work? Network architecture. Online at http://www1.alcatel- lucent.com/submarine/how/index.htm (accessed on 2009-08-18). Amano, K. and Y. Iwamoto, 1990. Optical Fiber Submarine Cable Systems. Journal of Lightwave Technology , Vol.8, No. 4, pp.595-609, April 1990. André, M. and N. Brochier, 2007. Overview of Submarine System Upgrades. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, USA. Arnaiz, S., B. in ’t Groen, C. Rivera, S. Sander, G. Caldeira, N. Vaz, I. Arroyo, S. HeBe and J. Horstink, 2008. Optical Fibre Cable Recycling at the L-FIRE European Project. Electronics Goes Green 2008+ , Merging Technology and Sustainable Development. Stuttgart: Fraunhofer IRB Verlag. ISBN 978-3-8167-7668-0. Ayres, R.U., 1995. Life cycle analysis: A critique. Journal of Resources, Conservation and Recycling , Vol.14, p199-223. Barattino, W. and N. Koopalethes, 2007. The emergence of affordable broadband services for remote locations using sfoc technology. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, USA. Bare, J.C. and T.P. Gloria, 2006. Critical Analysis of the Mathematical Relationships and Comprehensiveness of Life Cycle Assessment Approaches. Environmental Science and Technology, Vol.40, No.4, pp.1104- 1113.Baumann, H. and A.-M. Tillman, 2004. The Hitch Hiker’s Guide to LCA. Studentlitteratur: Lund, Sweden. ISBN: 91-44-02364-2. Beaufils, J.M., 2000. How Do Submarine Networks Web the World? Optical Fiber Technology. Vol.6, pp.15-32. Bergelin, F., 2008. Life cycle assessment of a mobile phone, a model on manufacturing, using and recycling. Master thesis, Ericsson Research and Uppsala Universitet. ISSN: 1401-5773, UPTEC Q08 014. Betts, P., 2009. Personal communication between 2009-03 and 2009-08 with Mr Betts of Global Crossing Ltd, London, England. www.globalcrossing.com. BlueScope, 2009. Energy use in BlueScope Steel’s steelmaking. Online at www.bluescopesteel.com (accessed on 2009-07-02). BRE, 2005. Green Guide to Specification, BRE Materials Industry Briefing Note 3a: Characterisation. Building Research Establishment Ltd. Online at www.bre.co.uk (accessed on 2009-07-17). Breverman, C., G. Valvo and B. Jander, 2007. A new generation of submarine line terminal equipment. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, USA. CARB, 2008. Emissions Estimation Methodology for Ocean-Going Vessels. Planning and Technical Support Division, California Air Resources Board (CARB). Chalmers, 2009. SPINE LCI dataset. Production of paint, thinner and enamel mainly for surface treatment of steel. Chalmers University, Göteborg, Sweden. Online at http://www.cpm.chalmers.se/CPMDatabase/Scripts/sheet.asp?ActId=1997-07-10828#Flow%20Data (accessed on 2009-07-17) Ciroth, A. and H. Becker, 2006. Validation – The Missing Link in Life Cycle Assessment. Towards pragmatic LCAs. International Journal of LCA , Issue 11, Vol. 5, pp.295 – 297, 2006. doi: http://dx.doi.org/10.1065/lca2006.09.271. Cisco, 2008. Approaching the Zettabyte Era. A White Paper by Cisco Systems, Inc. Cooper, D. and T. Gustafsson, 2004. Methodology for calculating emissions from ships: 1. Update of emissions factors. Report series for SMED and MSMED&SLU. Norrköping: SMHI Swedish Metrological and Hydrological Institute. Corbett, J.J., 2004. Verification of ship emission estimates with monitoring measurements to improve inventory and modelling, final report. Prepared for the California Air Resources Board (CARB) and the California Environmental Protection Agency (CEPA). 23 November 2004.

101

14 . References A life cycle assessment o f fibre optic submarine cable systems

Corbett, J.J. and P. Fischbeck, 1997. Emissions from Ships. Science , New Series, Vol.278.No.5339 (Oct.31.1997), pp.823-824. Corbett, J.J. and H.W. Koehler, 2003. Updated emissions from ocean shipping. Journal of Geophysical Research , Vol. 108, No. D20, 4650, doi:10.1029/2003JD003751. Corbett, J. and J. Winebrake, 2007. Sustainable Goods Movement, Environmental Implications of Trucks, Trains, Ships, and Planes. EM Magazine, Air & Waste Management Association, November 2007, pp.8-12. Corning, 2008. Fiber 101, Selected Concepts, May 2008. Corning Incorporated. Online at www.corning.com/opticalfiber/ (accessed on 2009-04-11). Corus, 2007. Sustainable Steel: Key environmental messages for steel & metal packaging. Online at www.corusgroup.com (accessed on 2009-07-06). CPNI, 2006. An overview of the use of submarine cable technology by UK PLC. Report for the Centre for the Protection of National Infrastructure (CPNI), United Kingdom, March 2006. Davis, J.R., 2001. Copper and copper alloys . ASM International. ISBN 0871707268. Dodbiba G., K. Takahashi, J. Sadaki and T. Fujita, 2008. The recycling of plastic wastes from discarded TV sets: comparing energy recovery with mechanical recycling in the context of life cycle assessment. Journal of Cleaner Production, Vol.16 (2008), pp.458-470.EAA, 2007. Aluminium Recycling in LCA. July 2007. European Aluminium Association. EEA, 1997. Life Cycle Assessment (LCA), A guide to approached, experiences and information sources. Environmental Issues Series no.6. European Environment Agency (EEA). Ellingsen, H., A.M. Fet and S. Aanondsen, 2002. Tool for Environmental Efficient Ship Design. ENSUS 2002, Newcastle, UK, December 16.2002. European Commission, 2002. Quantification of emissions from ships associated with ship movements between ports in the European Community. Final Report, July 2002. Prepared by Entec UK Limited. European Commission, 2007. Impact Assessment. Commission Staff Working Document. Accompanying document to the Proposal from the Commission to the European Parliament and Council for the regulation to reduce CO 2 emissions from passenger cars. Commission of the European Communities. {SEC(2007)1724}. Eyring, V., J.J. Corbett, D.S. Lee and J.J. Winebrake, 2007. Brief summary of the impact of ship emissions on atmospheric composition, climate, and human health. Document submitted to the Health and Environment sub-group of the International Maritime Organization on 6 November 2007. Fet, A.M., J. Embelmsvåg and J. Johannese, 1996. Environmental Impacts and Activity Based Costing during operation of a Platform Supply Vessel. A report for Farstad Shipping ASA, Ålesund, Norway. Fet, A.M., Michelsen and T. Johnsen, 2000. Environmental performance of transportation - a comparative study. Report nr 3/2000. Prepared for the Research Council of Norway, by Norwegian University of Science and Technology (NTNU). Fet, A.M., 2003. Sustainability reporting in shipping. Journal of Marine Design and Operations , No.B5 2003, pp.11-24. Fet, A.M., 2009. ISO 14000 as a Strategic Tool for Shipping and Shipbuilding. Online at www.ntnu.no/iot (accessed on 2009-03-24). Fullenbaum, M., 2004. State of the art unrepeatered cable: how advanced fibres and packaging have paved the way to advanced systems. Optical Networks Division, Alcatel Submarine Networks. Paper presented at the submarine cable conference “SubOptic 2004”, Monaco. Giurco, D., M. Stewart, T. Suljada and J. Petrie, 2006. Copper recycling alternatives: An environmental analysis. Department of Chemical Engineering, University of Sydney, NSW, 2006, Australia. 5th Annual Environmental Engineering Research Event, 20–23 October 2006, Noosa, QLD. GMSL, 2009. Fleet. Global Marine Systems Ltd, United Kingdom. Online at www.globalmarinesytems.co.uk (accessed on 2009-03-23).Gnauck, A.H. and A.R. Chraplyvy, 2008. High-Capacity Optical Transmission Systems. Journal of Lightwave Technology , Vol.26, No.9, pp.1032-1045 (2008). Graedel, T.E., D. Vanbeers, M. Bertram, K. Fuse, R.B. Gordon, A. Gritsinin, A. Kapur, R.J. Klee, R.J. Lifset, L.Memon, H. Rechberger, S. Spatari, and D. Vexler, 2004. Multilevel Cycle of Anthropogenic Copper. Environmental Science & Technology , Vol.38, No.4, pp.1242-1252. Green, J.A.S., 2007. Aluminium Recycling and Processing for Energy Conservation and Sustainability . ASM International. ISBN 0871708590. Guggemos, A.A. and A. Horvath, 2005. Comparison of Environmental Effects of Steel- and Concrete-Framed Buildings. Journal of infrastructure Systems , June 2005, pp93-101. Guinée, J.B. (final ed), M. Gorrée, R. Heijungs, G. Huppes, R. Kleijn, A. de Koning, L. van Oers, A. W. Sleeswijk, S. Suh, H. Udo de Haes, H. de Bruijn, R. van Duin, M.A.J. Huijbregts, E. Lindeijer, A.A.H. Roorda, B.L. van der Ven and B.P. Weidema, 2004. Handbook on Life Cycle Assessment. Operational Guide to the ISO Standards . Dordrecht: Kluwer Academic Publishers. Harasawa, S., Sumitani and K. Ohta, 2008. Reliability Technology for Submarine Repeaters. Fujitsu Science Technology Journal , Vol.44, No.2, pp.148-155 (2008). Herron, J.K., J. Porto, J. De la Cruz, S. Drew and J. Grant. 2007. The metamorphosis of marine maintenance providers: meeting the expanding needs of cable system owners. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, Maryland, USA.

102

14 . References A life cycle assessment o f fibre optic submarine cable systems

Hilt, J., 2009. The Benefits of Submarine Cables over Satellite Networks. Presentation by Hibernia Atlantic. Online at www.hiberniaatlantic.com (accessed on 2009-03-24). Hunkeler, D. and G. Rebitzer, 2005. The Future of Life Cycle Assessment. International Journal of LCA , Issue 10, Vol. 5, pp.305–308, 2006. doi: http://dx.doi.org/10.1065/lca2005.09.001. Igarashi, Y., I. Daigo, Y. Matsuno and Y. Adachi, 2007. Dynamic Materials Flow Analysis for Stainless Steels in Japan – Reductions Potential of CO2 Emissions by Promoting Closed Loop Recycling of Stainless Steel. ISIJ International , Vol.47, No.5, pp.758-763 (2007). Ishida, K., K. Shimizu, S. Mitani, J. Abe, T. Tokura and T. Mizuochi, 2007. RZ-DQPSK transponder for 40 gbps submarine cable systems. Mitsubishi Electric Corporation, Japan. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, Maryland, USA. ISO 14040:2006. Environmental Management – Life cycle assessment – Principles and framework (SS-EN ISO 14040:2006). Swedish Standards Institute (SIS förlag AB): Stockholm, Sweden. ISO 14044:2006. Environmental Management – Life cycle assessment – Requirements and guidelines (SS-EN ISO 14044:2006). Swedish Standards Institute (SIS förlag AB): Stockholm, Sweden. Jones, A.T., 2009. Electric Arc Furnace Steelmaking. American Iron and Steel Institute. Online at www.steel.org (accessed on 2009-07-06). Jönsson, Å., T. Björklund, A.M. Tillman, 1998. LCA Case Studies: LCA of Concrete and Steel Building Frames. International Journal of LCA , Issue 3, Vol.4, pp216-224, 1998. Kidorf, H., 2006. Submarine systems supply market rebounds. Lightwave Magazine, 1 st November 2006. Online at www.pioneerconsulting.com (accessed on 2009-07-15). Kordahi, M., S. Shapiro and G. Lucas, 2007. Trends in submarine cable systems. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, USA. Lack, D.A, J.J. Corbett, T. Onasch, B. Lerner, P. Massoli, P.K. Quinn, T.S. Bates, D.S. Covert, D. Coffman, B. Sierau, S. Herndon, J. Allan, T. Baynard, E. Lovejoy, A.R. Ravishankara and E. Williams, 2009. Particulate emissions from commercial shipping: Chemical, physical, and optical properties. Journal Of Geophysical Research , Vol. 114, D00F04, pp1-16, doi:10.1029/2008JD011300, 2009. Lecroart, A., N. Shaheen and P. Shawyer, 2007. Cabled science observatories solutions: bringing power and broadband communication to the ocean depths. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, Maryland, USA. Letellier, V., 2004. Submarine systems, from laboratory to seabed. Optics & Photonics News . Optical Society of America. February 2004. Lifset, R.J., R.B. Gordon, T.E. Graedel, S. Spatari, and M. Bertram, 2002. Where Has All the Copper Gone: The Stocks and Flows Project, Part 1. JOM , Vol.54, No.10, pp.21-26. Lingg, B. and S. Villiger, 2002. Energy and Cost Assessment of a High Speed Ferry. Master thesis, Royal Institute of Technology (KTH), Stockholm and Swiss Federal Institute of Technology (ETH) Zürich. Louw, A., 2009. Personal communication between 2009-03-10 and 2009-07-13 with Mr Louw of Mertech Marine, South Africa. www.mertechmaine.co.za.Markow, 2009. Summary of Undersea Fiber Optic Network Technology and Systems. Terremark Inc. Presentation. McGrath, M., 2009. Meeting with Mr McGrath of Telecom New Zealand at Cable Terminal Station, Takapuna, New Zealand on 2009-02-27. www.telecom.co.nz. Merret, J. and R. Laude, 2007. Re-use and re-routing of retiring systems. Alcatel Submarine Networks UK Ltd, London, UK. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, Maryland, USA. Mertech, 2009. Company Information. Mertech Marine. Online at www.mertechmarine.co.za (accessed on 2009-07- 13). NEC, 2008. NEC’s Submarine Cable System. Presentation by Masamichi Imai, December 5, 2008, Broadband Network Operations Unit, NEC Corporation. Online at www.c- hotline.net/docs/html/NECC8446/dl/necc081205e_1.pdf (accessed on 2009-03-24). Oikawa, H., J. Yoshimura and H. Watanabe, 2006. High-Performance Submarine Line terminal Equipment for Next- Generation Optical Submarine Cable System: FLASHWAVE S680. Fujitsu Science Technology Journal, Vol.42, No.4, pp.469-475 (2006). PE & LBP, 2008. GaBi Manual, Introduction to GaBi 4. Software-System and Databases for Life Cycle Engineering. PE International and the Chair of Building Physics, University of Stuttgart. PE & LBP, 2009. GaBi software database plan. Taken from example database. Poole, R., 2009. Personal communication on 2009-04-23 with Mr Poole of Earth Sciences and Surveying (EGS). www.egssurvey.com. Pre, 2000. The Eco-indicator 99. A damage oriented method for Life Cycle Impact Assessment. Manual for Designers, 17 April 2000, Second edition. PRé Consultants B.V. Ridder, J., 2007. Aspects of submarine cable retirement. Deutstche Telecom AG Competence Centre Submarine Cables. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, Maryland, USA.

103

14 . References A life cycle assessment o f fibre optic submarine cable systems

Ruddy, M., 2006. An Overview of International Submarine Cable Markets. Presentation by Terabit Consulting for the Executive Telecoms Briefings, Boston University, December 12, 2006. Online at www.terabitconsulting.com (accessed on 2009-07-15). Salomone, R., F. Mondello, F. Lanuzza and G. Micali, 2005. ENVIRONMENTAL ASSESSMENT: An Eco-balance of a Recycling Plant for Spent Lead–Acid Batteries. Environmental Management Vol.35, No.2, pp.206–219. DOI: 10.1007/s00267-003-0099-x. Schmitz, C., J. Domagala and P. Haag, 2006. Handbook of aluminium recycling: fundamentals, mechanical preparation, metallurgical processing, plant design . Vulkan-Verlag GmbH. Sleeswijk, A., L. van Oers, J. Guinée, J. Struijs, M. Huijbregts, 2008. Normalisation in product life cycle assessment: An LCA of the global and European economic systems in the year 2000. Science Of The Total Environment, Vol.390, pp.227–240 (2008). Doi:10.1016/j.scitotenv.2007.09.040. Spatari, S., M. Betz, H. Florin, M. Baitz and M. Faltenbacher, 2001. Using GaBi 3 to Perform Life Cycle Engineering. International Journal of LCA , Issue 6, Vol.2, pp.81-84, 2001. Spatari, S., M. Bertram, R.B. Gordon, K. Henderson and T.E. Graedel, 2005. Twentieth century copper stocks and flows in North America: A dynamic analysis. Ecological Economics, Vol.54 (2005), pp.37–51. Steel Recycling Institute, 2007. The Inherent Recycled Content of Today’s Steel. Report by the Steel Recycling Institute. Online at www.recycle-steel.org (accessed on 2009-07-02). Städje, J., 2009. Sunet först med 40 Gbps under Atlanten. Report for web magazine Teknik360 . Online at http://t360.idg.se/2.8229/1.216279/sunet-forst-med-40-gbps-under-atlanten (accessed on 2009-03-06). Suyama, M., M. Nagayama, H. Watanebe, H. Fujiwara and C. Anderson, 1999. WDM Optical Submarine Network Systems. Fujitsu Science Technology Journal , Vol.35, No.1, pp.34-45 (1999). Telegeography, 2009. Global bandwidth forecast Service: Methodology. Telegeography Research, PriMetrica Inc: Washington, D.C., USA. Todd, K., 2009. Personal communication on 2009-04-08 with Mr Todd of Global Marine Systems Ltd, Chelmsford, England. www.globalmarinesystems.co.uk Trischitta, P.R., A.J.C. Medina and R.C. Remedi, 1997. The Pan American Cable System. IEEE Communications Magazine, December 1997. UN, 2009. World Population Prospects: The 2008 Revision, Population Database. Population Division of the Department of Economic and Social Affairs of the United Nations. Online at http://esa.un.org/unpp/p2k0data.asp (accessed on 2009-08-19). USEPA, 2006. Life Cycle Assessment: Principles and Practice. Report prepared by Scientific Applications International Corporation (SAIC), for National Risk Management Research Laboratory, US Environmental Protection Agency, EPA/600/R-06/060, May 2006. Veverka, D., 2009. Personal communication between 2009-03 and 2009-09 with Mr Veverka of Southern Cross Cables Ltd, New Zealand. www.southerncrosscables.com. Viking and Österberg, 2004. Thesis work in Regional Planning. Department of Urban Studies, The Royal Institute of Technology (KTH), Stockholm, Sweden. Willey C., P.J. Footman Williams, P. Rego, J. de la Cruz, 2007. Re-installation of recovered submarine cables; case histories of success. Tyco Telecommunications (US) Inc. Paper presented at the submarine cable conference “SubOptic 2007”, Baltimore, Maryland, USA. Williams, E. and Y. Sasaki, 2003. Energy Analysis of End-of-life Options for Personal Computers: Resell, Upgrade, Recycle. World Steel, 2008. Fact Sheet: Energy, Steel and Energy. World Steel Association. Online at www.worldsteel.org (accessed on 2009-07-06).

14.2. Internal Ericsson References

3GLCA, 2002. Life cycle assessment of a 3G system (Final report). Document No: 9/0363-FCP1032560. Berggren, 2009. Personal communication on 2009-04-21&22 with Mr Berggren of Ericsson Network Technologies, Hudiksvall, Sweden. Characteristics Specs, 2009. System Characteristics; EDA 1200 4.2. Document No: 155 02-FGC 101 0316/3 Uen. Malmodin, 2009. Personal communication with Mr Malmodin of Ericsson Research, Kista, Sweden. MECH, 2001. Mechanics manufacturing processes. Document No: 5/1551-FDP 103 2560 Uen. Jonsson, 2009a. M. Sc. Thesis: LCA on Submarine Opto-Cable. Document No: EAB-09:002628. Jonsson, 2009b. Telepresence - Comparing Bit and Air Travel. Document No: EAB-08:082380. Norlund, 2009. Personal communication on 2009-04-23 with Mr Norlund of Ericsson Network Technologies, Hudiksvall, Sweden. PBA, 2001. RBS PBA Assembly and RBS Final Assembly. Document No: 7/1551-FCP 103 2560 Uen. Pb-Battery, 2001. Lead acid battery, manufacturing process. Document No: 8/1551-FCP 103 2560 Uen Jonsson, 2009. Telepresence - Comparing Bit and Air Travel. Document No: EAB-08:082380.

104

15 . Appendices A life cycle assessment o f fibre optic submarine cable systems

15. Appendices

15.1. Appendix A – Calculation of the Generic Cable System Adapted from Ruddy, 2006

Cable capacity - Year end 2006 Lit Ready Capacity for Lit Design Lit Capacity used for Service Route Capacity Capacity, as % of Study Cable Name Date Length km (Gbps) Est. (Gbps) design Landings (Gbps) Transatlantic Columbus-2 1994 12188 2 2 100% 6 CANTAT-3 1994 7500 5 5 100% 6 TAT-12/TAT-13 1995 12553 30 30 100% 5 Atlantic Crossing-1 (AC-1) 1998 14000 140 140 100% 4 Columbus-3 1999 10000 20 40 50% 5 Yellow (Level -3) / Atlantic Crossing-2 (AC-2) 2000 6960 320 1280 25% 2 320 Hibernia Atlantic 2001 11700 220 1920 11% 4 220 FLAG Atlantic-1 (FA-1) 2001 12800 530 2400 22% 4 530 TAT-14 2001 15000 640 640 100% 7 640 VSNL Transatlantic (Tyco) 2001 12500 480 2560 19% 2 480 Apollo 2003 13000 320 3200 10% 4 320 Total: 11 Systems, 22% lit 128201 2707 12217 22.16% 49 Transpacific TPC-5 1995 22560 20 20 100% 6 Pacific Crossing-1 (PC-1) 1999 13076 180 640 28% 4 China-US Cable Network 2000 30800 80 80 100% 9 Japan-US Cable Network 2001 21000 400 640 62% 6 400 VSNL Transpacific (Tyco) 2002 24100 640 7680 8% 5 640 Total: 5 systems, 14,6% lit 111536 1320 9060 14.57% 30 South Asia Sea-Me-We-2 1994 18000 1 1 100% 15 FLAG (Fiberoptic Link Around the Globe) 1997 27763 10 20 50% 17 Sea-Me-We-3 1999 39000 58 80 72% 39 i2i (ISCN) 2002 3200 160 8400 2% 2 160 SAT -3/WASC/SAFE (South Atlantic-3/West Africa 2002 27850 30 120 25% 17 Tata Indicom Chennai - Singapore (TICSCS) 2004 3100 320 3175 10% 2 320 Sea-Me-We-4 2006 20000 160 1280 13% 17 160 Falcon 2006 10300 90 2560 4% 16 Total: 8 systems, 5,3% lit 149213 829 15636 5.30% 125

Systems Length Lit Design Lit vs Design Landings Total 24 388950 4856 36913 13.16% 204 Lit Capacity Lit Design Lit Capacity used for Route Capacity Capacity. as % of Study Length km (Gbps) Est. (Gbps) design Landings (Gbps) Average System 16206 202 1538 13.16% 8.5 381 Normalised to 10000km of cable 10000 5.2 381

Notes: Lit capacity: current installed capacity Design capacity: Limits to upgrading process Bandwidth usage: Actual usage of cable

105

15 . Appendices A life cycle assessment o f fibre optic submarine cable systems

15.2. Appendix B – Data Sources

No. Process Nation Data Source Data Provider Year 1 Acrylonitrile-butadiene-styrene (ABS) in municipal waste incinerator RER ELCD/PE-GaBi PE-GaBi 2006 2005 2 Acrylonitrile-butadiene-styrene granulate (ABS) RER ELCD/PlasticsEurope PE-GaBi 2006 2005 3 Aluminium die-cast part DE PE PE-GaBi 2006 2005 4 Aluminum ingot RER BUWAL PE-GaBi 2006 1996 5 Argon (gaseous) DE PE PE-GaBi 2006 2005 6 Bitumen at refinery EU-15 PE PE-GaBi 2006 2003 7 Brass DE PE PE-GaBi 2006 2002 8 Cable waste in municipal waste incinerator RER PE 2005

9 Cast iron part (sand casting) DE PE PE-GaBi 2006 2005 10 Commercial waste in municipal waste incinerator RER PE PE-GaBi 2006 2005 11 Copper mix (99,999% from electrolysis) DE PE PE-GaBi 2006 2002 12 Corrugated cardboard CH BUWAL PE-GaBi 2006 1996 13 Diesel at refinery EU-15 ELCD/PE-GaBi PE-GaBi 2006 2003 Chalmers 14 Enamel paint and thinner for steel 1997 University 15 Epoxy resin RER PlasticsEurope PE-GaBi 2006 2005 16 Fuel oil heavy at refinery EU-15 ELCD/PE-GaBi 2003

17 Fuel oil heavy at refinery US PE 2003

18 Fuel oil light at refinery EU-15 ELCD/PE-GaBi 2003

19 Glass (white; packaging) CH BUWAL PE-GaBi 2006 1996 20 Glass fibres DE PE PE-GaBi 2006 2005 21 Hexamethylenediamine (HMDA; via Adipic acid) DE PE PE-GaBi 2006 2005 Landfill (Commercial waste for municipal disposal; AT, DE, IT, LU, NL, SE, 22 RER PE PE-GaBi 2006 2005 CH) 23 Lead (99,995%) DE PE PE-GaBi 2006 2000 24 Limestone hydrate (Ca(OH)2) DE PE PE-GaBi 2006 2000 25 Natural gas mix ELCD/PE-GaBi 2002

26 Nitrogen (gaseous) DE PE PE-GaBi 2006 2005 27 Oxygen (gaseous) DE PE PE-GaBi 2006 2005 28 Paper / Cardboard in municipal waste incinerator RER ELCD/PE-GaBi PE-GaBi 2006 2005 29 Paper woody uncoated CH BUWAL PE-GaBi 2006 1996 30 Polycarbonate granulate (PC) RER ELCD/PlasticsEurope PE-GaBi 2006 2005 31 Polyester resin (UP; unsaturated) PE 2005

32 Polyethylene (PE) in municipal waste incinerator RER ELCD/PE-GaBi PE-GaBi 2006 2005 33 Polyethylene high density granulate (PE-HD) RER ELCD/PlasticsEurope PE-GaBi 2006 2005 34 Polypropylene film (extended) (PP) RER PlasticsEurope PE-GaBi 2006 2005 Populated printed wiring board (before RoHS), in municipal waste 35 RER PE PE-GaBi 2006 2005 incinerator 36 Propene (propylene) RER PlasticsEurope PE-GaBi 2006 2005 37 Silica sand (flour) DE PE PE-GaBi 2006 2005 38 Silicon mix (99%) DE PE PE-GaBi 2006 2000 39 Sodium hydroxide (100%; caustic soda) RER ELCD/PlasticsEurope PE-GaBi 2006 2005

106

15 . Appendices A life cycle assessment o f fibre optic submarine cable systems

No. Process Nation Data Source Data Provider Year

40 Stainless steel cold roll DE PE PE-GaBi 2006 2004 41 Steel billet DE PE PE-GaBi 2006 2004 42 Steel cast part alloyed DE PE PE-GaBi 2006 2005 43 Steel cold rolled DE PE PE-GaBi 2006 2004 44 Steel sheet 1.5mm el. zinc plated (0.01mm; 1s) DE PE PE-GaBi 2006 2004 45 Styrene RER PlasticsEurope PE-GaBi 2006 2005 46 Styrene-butadiene rubber mix (SBR) DE PE PE-GaBi 2006 2005 Cooper and 47 subsea.Average CABLE ship MANOEUVRING (24hrs) 2004 Gustafsson Cooper and 48 subsea.Average CABLE ship PORT (24hr) 2004 Gustafsson Cooper and 49 subsea.Average CABLE ship TRANSIT (24hrs) 2004 Gustafsson Cooper and 50 subsea.Average RESEARCH ship MANOEUVRING (24hrs) 2004 Gustafsson Cooper and 51 subsea.Average RESEARCH ship PORT (24hrs) 2004 Gustafsson 52 Sulphuric acid (96%) RER PE PE-GaBi 2006 2005 53 Thermal energy from gas BUWAL PE-GaBi 2006 1996

54 Thermal energy from heavy fuel oil EU-25 ELCD/PE-GaBi PE-GaBi 2006 2002 55 Thermal energy from natural gas EU-25 ELCD/PE-GaBi PE-GaBi 2006 2002 56 Tin plate DE BUWAL PE-GaBi 2006 1996 Truck -trailer > 34 - 40 t total cap./ 27 t payload / Euro 3 (Truck fleet, long - 57 GLO PE PE-GaBi 2006 2005 dist.) 58 Truck-trailer up to 28 t total cap. / 12,4 t payload / Euro 4 GLO PE PE-GaBi 2006 2005 59 x Air travel Bergelin 2008

60 x Dismantling electronic scrap (for processes) Bergelin 2008

61 x Electricity (China av.) CN Bergelin 2008

62 x Gold production Bergelin 2008

63 x Hazardous waste treatment GLO Bergelin 2008

64 x Kaldo furnace for processes Bergelin 2008

65 x Nickel production Bergelin 2008

66 x Schredding electronic scrap (for processes) Bergelin 2008

67 xH450: Standard components model[2] AG Ericsson 2002

68 xM240: RBS PBA process Ericsson 2002

69 xM330: PCB manufacturing Ericsson 2002

70 xM332: PCB materials[2] Ericsson 2002

71 xM401: Connectors Ericsson 2002

72 xM531: IC chip class-A manufact. part II (Copy) Ericsson 2002

73 xM532: IC chip class-B manufact. part II Ericsson 2002

74 xP012: Electricity (EU av.) v EU Ericsson 2002

75 xP014: Electricity (US av.) v US Ericsson 2002

76 xP015: Electricity (Japan av.) v JA Ericsson 2002

77 xX007: Silver[2] Bergelin 2008

78 Zinc redistilled mix DE PE PE-GaBi 2006 2002

107

15 . Appendices A life cycle assessment o f fibre optic submarine cable systems

15.3. Appendix C – Stakeholder Analysis

Company Link to submarine cables Location

Alcatel System manufacture, installation and maintenance England, France

British Telecom Cable Owner United Kingdom

Corning Incorporated Optical fibre manufacture USA

EGS Cable route survey Hong Kong

Ericsson Cables Cable manufacture Sweden

Fujitsu Repeater manufacture Japan

Global Crossing Cable owner Global offices

Global Marine Systems Ltd Cable installation and maintenance England

ICPC International Cable Protection Committee Global

Mertech Marine Cable recovery South Africa

NEC Cable and repeater manufacture Japan

Southern Cross Cable Network Cable owner Global offices

Sumi Electric Corning Incorporated Japan

Sumitomo Cable manufacture Japan

Telecom New Zealand Cable operator New Zealand

Telegeography Research and monitoring of submarine systems USA

Telia Cable owner Sweden

Tyco System manufacture, installation and maintenance USA

108

15 . Appendices A life cycle assessment of fibre optic submarine cable systems

15.4. Appendix D – Detailed System Flowchart of GaBi software sub-model

109

15 . Appendices A life cycle assessment of fibre optic submarine cable systems

15.5. Appendix E – Questionnaire to Suppliers

Craig Donovan Master’s Thesis Stockholm Sweden

Dear Sir/Madam,

Re: Data collection - Life Cycle Assessment (LCA) of submarine cables

I am a student undertaking my Master of Science at the Royal Institute of Technology (KTH) in Stockholm, Sweden. The field of study is Environmental Engineering and Sustainable Infrastructure . Background to the course can be found at (http://www.kth.se/studies/master/programmes/be/2.1572?l=en).

I have recently commenced my thesis project in conjunction with KTH and Ericsson Research in Stockholm, in the area of Life Cycle Assessment (LCA) of submarine fibre optic cables . The research encompasses assessment of the environmental impact (for example CO2 emissions) of submarine cables from cradle to grave, or, from raw material extraction to decommissioning.

These cables support our voice and data communications in an extremely efficient manner, yet little is known about the impact of these cables from an LCA perspective. We have undertaken some preliminary investigation and a study of this kind does not appear to have been completed. Therefore, I hope that this study can be of benefit to the submarine cable industry as a whole.

The process of building an LCA model requires accounting for the inputs - raw material and energy, and the outputs - emissions and waste, of the studied system. In this case information relating to the production of optical fibre, plastics, steel and other materials used in the production of submarine cables systems.

As a supplier of raw materials to Ericsson Cables, the data you could provide would greatly help my research and facilitate the building of a complete life cycle model.

110

15 . Appendices A life cycle assessment of fibre optic submarine cable systems

Questions to suppliers:

a. Do you have any personnel working with LCA and has an LCA study been undertaken for the product?

b. What are the raw materials that go into making the product?

c. How far are the raw materials transported to your processing factory?

d. By what method are they transported? The approximate gross tonnage of the ship/train/truck?

e. How much energy does it take to produce X amount of final product, ready for supply to Ericsson Cables?

f. What is the source of that energy (For example: electricity from coal)?

g. How much waste is produced in this process and that can be attributed to the product?

h. What percentage of that waste is recycled and how much is sent to landfill?

Much of this data would be contained in annual reports, product specification sheets and process specifications. Confidentiality would be strictly maintained if these resources were made available.

Due to time limitations, it is important that any data be supplied in good time in order to progress the research.

I can be contacted at: [email protected]

Thank you in advance for your support.

Yours faithfully

Craig Donovan

111

15 . Appendices A life cycle assessment of fibre optic submarine cable systems

15.6. Appendix F – Sensitivity Analysis Results

Terminal Raw Repeater Repeater Raw Materials, Cable Lifetime 25 Recycling internal Materials, Manufacturing Terminal EU-25 maintenance Ship emissions years at No Raw transport transport at electronics at Manufacturing and E-o-L at electricity use electricity by ship at to Marine 10000Gb.km recovery at Impact category at 500% 1000% 400% & E-o-L at 0% 0% at 70% production 150% Distillate fuel relative E-o-L

Primary Energy 1,6% 3,8% 1,0% -0,6% -1,5% -17,3% -13,0% 7,5% 0,0% -12,4% 2,6%

Electricity 0,0% 0,0% 1,1% -0,5% -1,4% -26,8% 0,0% 0,0% 0,0% -4,1% -4,1%

Heavy Fuel Oil 0,0% 0,0% 0,0% -0,1% 0,0% 0,0% 0,0% 26,3% 0,0% -22,8% -22,8%

Abiotic Resource 2,1% 5,0% 0,9% -0,6% -1,4% -15,0% -18,8% 9,9% 0,0% -13,9% 6,5% Depletion

Acidification Potential 0,3% 0,8% 0,0% 0,0% -0,1% -1,9% 0,9% 24,8% -46,7% -21,1% -18,9%

Freshwater Aquatic 0,1% 0,1% 0,9% -0,5% -1,2% -21,3% -76,4% 5,2% -16,9% -8,1% -1,2% Ecotoxicity Potential

Terrestrial Ecotoxicity 0,2% 0,5% 0,4% -0,2% -0,7% -8,7% -21,5% 17,8% -59,5% -16,7% -13,2% Potential

Marine Aquatic 0,2% 0,5% 0,6% -0,4% -1,0% -13,9% 65,1% 12,3% -27,1% -13,5% -2,0% Ecotoxicity Potential

Global Warming 1,1% 2,6% 0,6% -0,3% -0,9% -9,0% -7,6% 17,0% 2,1% -16,9% -7,7% Potential

Photochemical Ozone 0,5% 1,3% 0,1% -0,1% -0,2% -1,9% 1,3% 24,5% -42,7% -21,4% -17,1% Creation Potential

Ozone Depletion 1,7% 4,1% -1,1% -15,6% -19,2% 0,0% 5024,6% 8,9% 0,0% -39,4% 126,0% Potential

Eutrophication 0,5% 1,3% 0,1% -0,1% -0,3% -1,8% -3,5% 24,9% -21,9% -21,2% -19,0% Potential

Human Toxicity 0,1% 0,2% 0,2% -0,1% -0,3% -5,0% -13,4% 21,7% -69,3% -19,0% -14,6% Potential

112