smart grid

SMART GRIDS: A PARADIGM SHIFT ON ENERGY GENERATION AND DISTRIBUTION WITH THE EMERGENCE OF A NEW ENERGY MANAGEMENT BUSINESS MODEL

JESUS ALVARO CARDENAS International Business

APPROVED:

Leopoldo A. Gemoets, D.Sc., Chair Jose H. Ablanedo-Rosas, Ph.D Kallol K. Bagchi, Ph.D.

Robert J. Sarfi. Ph.D.

Bess Sirmon-Taylor, Ph.D. Interim Dean of the Graduate School

Copyright ©

by Jesus A. Cardenas 2014

SMART GRIDS: A PARADIGM SHIFT ON ENERGY GENERATION AND DISTRIBUTION WITH THE EMERGENCE OF A NEW ENERGY MANAGEMENT BUSINESS MODEL by

JESUS ALVARO CARDENAS, BSEE, MBA, MSIE

DISSERTATION

Presented to the Faculty of the Graduate School of The University of Texas at El Paso in Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY

International Business THE UNIVERSITY OF TEXAS AT EL PASO May 2014

UMI Number: 3623383

All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion.

UMI 3623383 Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code

ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, MI 48106 - 1346

ABSTRACT An energy and environmental crisis will emerge throughout the world if we continue with our current practices of generation and distribution of electricity. A possible solution to this problem is based on the Smart grid concept, which is heavily influenced by Information and Communication Technology (ICT). Although the electricity industry is mostly regulated, there are global models used as roadmaps for Smart Grids’ implementation focusing on technologies and the basic generation-distribution-transmission model. This project aims to further enhance a business model for a future global deployment. It takes into consideration the many factors interacting in this energy provision process, based on the diffusion of technologies and literature surveys on the available documents in the Internet as well as peer-reviewed publications. Tariffs and regulations, distributed energy generation, integration of service providers, consumers becoming producers, self-healing devices, and many other elements are shifting this industry into a major change towards liberalization and deregulation of this sector, which has been heavily protected by the government due to the importance of electricity for consumers. We propose an Energy Management Business Model composed by four basic elements: Supply Chain, Information and Communication Technology (ICT), Stakeholders Response, and the resulting Green Efficient Energy (GEE). We support the developed model with an exhaustive literature survey, diffusion analysis of the different technologies under the umbrella of Smart Grids (SG), and two surveys: one administered to peers and professionals, and another for experts in the field, based on the Smart Grid Carnegie Melon Maturity Model (CMU SEI SGMM). The contribution of this model is a simple path to follow for entities that want to achieve environmental friendly energy with the involvement of technology and all stakeholders.

iv

TABLE OF CONTENTS ABSTRACT ............................................................................................................................. iv  TABLE OF CONTENTS...........................................................................................................v  LIST OF TABLES .....................................................................................................................x  LIST OF FIGURES ............................................................................................................... xiii  CHAPTER 1: INTRODUCTION ..............................................................................................1  1.1  Background Information ..........................................................................................1  1.2  The Birth of a New Model .......................................................................................4  1.3  The Most Important Elements .................................................................................5  1.4  The Enhanced Energy Management Business Model .............................................6  1.5  Research Question And Objectives .........................................................................8  1.6  Contributions of This Dissertation ...........................................................................9  CHAPTER 2: COMPREHENSIVE RESEARCH IN SMART GRID ....................................10  2.1  Background Information ........................................................................................10  2.2  Theoretical Background .........................................................................................12  2.2.1  Smart Grid Concept Defined ........................................................................12  2.2.2  SWOT Analysis ............................................................................................13  2.3  Methodology For Taxonomy Research .................................................................17  2.3.1  Google Scholar Research Results .................................................................18  2.3.2  Preliminary Conclusions of Google Research ..............................................24  2.4  Literature Survey Using ISI Web of Science .........................................................25  2.4.1  Hypotheses H2 ..............................................................................................26  2.4.2  Research Purposes ........................................................................................28  2.4.3  Research Methodology .................................................................................32  2.4.3.1  Classification by Research Categories ...........................................33  2.4.3.2  Classification by Research Focus ...................................................35  2.4.3.3  Classification by Data Collection Method .....................................36  2.4.3.4  Classification by Data Analysis Technique ...................................38  2.4.3.5  Classification by Discipline ...........................................................39  2.4.3.6  Taxonomy of Paper’s Purpose .......................................................41  v

2.4.3.7  Smart Grid Technologies ...............................................................44  2.4.3.8  Originating Countries .....................................................................45  2.4.4  Further Analysis and Prognosis ....................................................................52  2.4.4.1  ICT Related Papers.........................................................................57  2.4.4.2  Physical Infrastructure Related Papers ...........................................58  2.4.4.3  Economics Related Papers .............................................................59  2.4.4.4  Environmental Related Papers .......................................................61  2.4.5  Hypotheses H2 Results .................................................................................62  2.5  Data Oriented Analysis ..........................................................................................63  2.5.1  Word Mining .................................................................................................63  2.5.2  Bass’ Diffusion Model ..................................................................................67  2.5.3  Author-oriented analysis ...............................................................................70  2.6  Gap Analysis of Literature and Investments ..........................................................75  2.6.1  Background Information ...............................................................................75  2.6.2  Hypotheses H3 ..............................................................................................76  2.6.3  US Government ............................................................................................76  2.6.4  Private Sector ................................................................................................81  2.6.5  Academic Sector ...........................................................................................82  2.6.6  Methodology .................................................................................................83  2.6.7  Results ...........................................................................................................85  2.6.8  Gap’s Conclusions ........................................................................................87  2.7  Conclusions and Graphical View...........................................................................87  2.7.1  Conclusions ...................................................................................................87  2.7.2  Graphical View .............................................................................................91  CHAPTER 3: DIFFUSION OF TECHNOLOGIES AND RISKS ..........................................93  3.1  Generation and Consumption Information ............................................................93  3.1.1  Hypotheses H4 ..............................................................................................94  3.1.2  Private Investments on Energy .....................................................................95  3.1.3  Sources of Consumption and Generation Statistics ......................................97  3.1.4  Global Trends on Energy Use and Availability ..........................................102  3.1.5  Global Trends on Generation and Sources .................................................106  3.1.6  Global Sources of Generation Capacity versus Production ........................109  vi

3.1.7  Diffusion of Renewable Generation ...........................................................114  3.1.7.1  Wind Generated Electricity in the US ..........................................115  3.1.7.2  Solar Generated Electricity in the US ..........................................117  3.1.8  Global Prognosis .........................................................................................118  3.2  Advanced Metering Infrastructure Background ..................................................123  3.2.1  Hypotheses H5 ............................................................................................125  3.2.2  Implementation Progress ............................................................................126  3.2.3  Bass’ Diffusion Model for AMI/AMR .......................................................127  3.2.3.1  Geographical Clusters of States ...................................................127  3.2.3.2  Utility Company Ownership ........................................................135  3.2.3.3  Urban Concentration Analysis .....................................................137  3.2.4  Hypotheses H5’s Results ............................................................................139  3.2.5  Findings and Prognosis for AMI.................................................................139  3.3  Electric Vehicles Background Information .........................................................139  3.3.1  Implementation Progress ............................................................................140  3.3.2  Hypotheses H6 ............................................................................................141  3.3.3  Deployment of Electric Vehicles in the US ................................................141  3.3.4  Findings and Prognosis for Electric Vehicles .............................................149  3.4  Cyber Risks Background Information .................................................................150  3.4.1  Security Breaches Measures and Research Focus ......................................152  3.4.2  Types of Cybersecurity Risks .....................................................................153  3.4.3  Hypotheses H7 ............................................................................................154  3.4.4  Methodology ...............................................................................................154  3.4.5  Hypotheses Results .....................................................................................162  3.4.6  Conclusions .................................................................................................162  3.5  Adding All Diffusions Together ..........................................................................163  3.5.1  Distributed Generation Diffusion ...............................................................163  3.5.2  Smart Meters Diffusion...............................................................................164  3.5.3  Electric Vehicles Diffusion .........................................................................165  3.5.4  Cyber Attacks Diffusion .............................................................................166  3.5.5  All the Prior Diffusions Together ...............................................................167  3.5.6  Conclusions .................................................................................................168  vii

CHAPTER 4: ROLE OF CONSUMERS IN THE NEW BUSINESS MODEL ...................169  4.1  Background Information ......................................................................................169  4.2  Literature Review.................................................................................................170  4.3  Model Development.............................................................................................174  4.3.1  Background Information .............................................................................174  4.3.2  United Kingdom Road Map ........................................................................175  4.3.3  German E-energy Road Map ......................................................................176  4.3.4  United States Road Map for Smart Grids ...................................................178  4.3.5  China Strong Smart Grid ............................................................................179  4.3.6  Masdar: The Sustainable City .....................................................................180  4.3.7  Texas Smart Grid Investment Model ..........................................................182  4.3.8  Ontario Smart Grid Model ..........................................................................183  4.3.9  Developing our Own Model .......................................................................184  4.4  First Survey ..........................................................................................................187  4.4.1  Survey for Smart Energy Perception ..........................................................187  4.4.2  Survey Methodology...................................................................................189  4.4.3  Data Analysis ..............................................................................................190  4.4.3.1  Gender of Respondents ................................................................190  4.4.3.2  Education Level............................................................................191  4.4.3.3  Household Income........................................................................191  4.4.3.4  What is important for the consumer? ...........................................192  4.4.3.5  What is everyone’s role? ..............................................................193  4.4.3.6  Importance to Society...................................................................196  4.4.3.7  Important for the individual .........................................................197  4.4.3.8  Interviewees’ Own Definition ......................................................197  4.4.3.9  Cost of Energy ..............................................................................198  4.4.3.10  Location ......................................................................................198  4.4.4  PLS-SEM Model for the First Survey ........................................................199  4.4.5  Conclusions and Next Steps........................................................................203  4.5  Proposed Business Model Development and Validation .....................................204  4.5.1  New Business Model First Draft.................................................................205  4.5.2  Questions Development, Grouping and Analysis .......................................205  4.5.2.1  Development of Questions ...........................................................205  viii

4.5.2.2  Survey Preparation .......................................................................206  4.5.3  Hypotheses H8 ............................................................................................207  4.5.4  PLS Model for the Second Survey..............................................................209  4.5.5  Responders’ Statistic Analysis....................................................................211  4.5.6  PLS Model Results .....................................................................................216  4.5.7  Hypotheses H8’s Results ............................................................................220  4.5.8  Conclusions and Next Steps........................................................................221  4.6  Cost of Smart Energy ...........................................................................................221  4.6.1  Hypotheses H9 ............................................................................................223  4.6.2  Electricity Consumption .............................................................................225  4.6.3  Generation and Costs by Sources ...............................................................227  4.6.4 Deregulation Status .....................................................................................230  4.6.5  Competition by State...................................................................................231  4.6.6  Urban Concentration ...................................................................................232  4.6.7  Methodology ...............................................................................................233  4.6.8  Hypotheses H9 Results ...............................................................................238  4.6.9  Conclusions .................................................................................................238  CHAPTER 5: CONCLUSIONS AND FUTURE RESEARCH ............................................241  5.1  Global Position Of Smart Grid Distribution Literature .......................................241  5.2  Diffusion Models for ICT Enhanced Technologies .............................................241  5.3  Enhanced Business Model ...................................................................................243  5.4  General Conclusions ............................................................................................244  5.5  Next Steps and Further Research .........................................................................244  REFERENCES ......................................................................................................................246  APPENDIX ............................................................................................................................337  VITA ......................................................................................................................................344 

ix

LIST OF TABLES Table 2.1: Results of Hypotheses H1 ............................................................................................ 25  Table 2.2: General classification of this paper.............................................................................. 34  Table 2.3: Disciplines and Descriptions (Chicco, 2010) .............................................................. 39  Table 2.4: Row & Column Percentages for Research Classification versus Focus...................... 47  Table 2.5: Row & Column Percentages for Data Collection versus Paper’s Focus ..................... 48  Table 2.6: Row & Column Percentages for Data Analysis vs. Paper’s Focus ............................. 48  Table 2.7: Row & Column Percentages for Conference papers’ SG technologies vs. Purpose ... 49  Table 2.8: Row & Column Percentage for Journal Papers’ SG technologies vs. Purpose ........... 50  Table 2.9: Row & Column Percentages for Country vs. Paper’s Purpose ................................... 51  Table 2.10: Row & Column Percentages for Country vs. SG Technology .................................. 52  Table 2.11: Regression statistics for relationship of Journal and Conferences’ papers................ 54  Table 2.12: Category mixes related to SGD techs ........................................................................ 56  Table 2.13: Hypotheses H2 results ............................................................................................... 63  Table 2.14: Top 25 words on number of mentions in the past 5 years ......................................... 64  Table 2.15: Top 20 SGD Technologies mentions in the past 6 years ........................................... 65  Table 2.16: Bass’ Diffusion Model results for SGD technologies ............................................... 69  Table 2.17: Distribution of top 20 countries of origin of SGD papers’ authors ........................... 71  Table 2.18: Distribution of papers by number of authors ............................................................. 71  Table 2.19: Comparison of Journal and Conference papers by country of authors ...................... 72  Table 2.20: Distribution of authors by Country and writing order ............................................... 73  Table 2.21: Distribution of top 20 most prolific writers of SGD papers ...................................... 74  Table 2.22: Type of publication of most prolific authors (co-authors)......................................... 74  Table 2.23: DoE distribution on selected categories .................................................................... 84  Table 2.24: EPRI distribution on selected categories ................................................................... 85  Table 2.25: Academic Research distribution by categories .......................................................... 85  Table 2.26: Comparisons between DoE, EPRI and the Literature Survey ................................... 86  Table 2.27: Gap from the average of DoE and EPRI versus Literature survey ............................ 86  Table 2.28: Correlation results for comparison groups ................................................................ 86  Table 2.29: Hypotheses H3 results ............................................................................................... 87  Table 3.1: Consumption and generation regression lines’ statistics ............................................. 98  Table 3.2: Sensitivity analysis for the crossing-point year ........................................................... 99  Table 3.3: Consumption’s sensitivity analyses ........................................................................... 101  Table 3.4: Generation sources sensitivity analysis ..................................................................... 102  Table 3.5: Regression statistics for Generation and Imports ...................................................... 103  Table 3.6: Regression lines for Energy use and availability ....................................................... 105  Table 3.7: Sensitivity analysis for crossing line year for use and available energy ................... 106  Table 3.8: Capacity factors ......................................................................................................... 108  Table 3.9: Hydro capacity (KWh) .............................................................................................. 109  Table 3.10: Hydro production (KWh) ........................................................................................ 110  Table 3.11: Nuclear Capacity (KWh) ......................................................................................... 110  Table 3.12: Nuclear Production (KWh) ...................................................................................... 111  Table 3.13: Solar Capacity (KWh) ............................................................................................. 111  Table 3.14: Solar Production (KWh) .......................................................................................... 112  Table 3.15: Thermal Capacity (KWh) ........................................................................................ 112  x

Table 3.16: Thermal Production (KWh) ..................................................................................... 113  Table 3.17: Wind Capacity (KWh) ............................................................................................. 113  Table 3.18: Wind Production (KWh).......................................................................................... 114  Table 3.19: Evolution of Sources of Electricity Generation in the US ....................................... 114  Table 3.20: Diffusion Indexes for Wind Electricity Generation in the US by States ................. 115  Table 3.21: t-test for top 10 States using Wind generated Electricity ........................................ 116  Table 3.22: Diffusion Indexes for Solar Electricity Generation in the US by States ................. 118  Table 3.23: t-test for top 10 States using Solar generated Electricity ......................................... 118  Table 3.24: Hypotheses H4 Results ............................................................................................ 123  Table 3.25: Regional Clustering for the US................................................................................ 124  Table 3.26: Ownership of the utility Company .......................................................................... 125  Table 3.27: Urban concentration indexes for the US .................................................................. 127  Table 3.28: Cluster Totals for Diffusion ..................................................................................... 129  Table 3.29: t-test Results for Clusters ......................................................................................... 129  Table 3.30: Average statistics by States’ Clusters ...................................................................... 130  Table 3.31: Standard Deviation statistics by States’ Clusters..................................................... 130  Table 3.32: Standard Deviation t-test for States’ Clusters .......................................................... 130  Table 3.33: Diffusion Results by States’ Division...................................................................... 131  Table 3.34: t-test Results for Divisions....................................................................................... 132  Table 3.35: Average statistics by States’ Clusters ...................................................................... 132  Table 3.36: Standard Deviation statistics by States’ Divisions .................................................. 133  Table 3.37: Statistics from the Individual States ........................................................................ 133  Table 3.38: Standard Deviation t-test for States’ Divisions ....................................................... 134  Table 3.39: Top 10 States Diffusion Statistics............................................................................ 134  Table 3.40: Diffusion Statistics by Company Ownership .......................................................... 135  Table 3.41: Diffusion Statistics for Ownership Categories ........................................................ 136  Table 3.42: Diffusion Statistics for Urban Concentration .......................................................... 138  Table 3.43: Hypotheses H5’s Results ......................................................................................... 139  Table 3.44: Regression results for gasoline price vs PHEV sales .............................................. 142  Table 3.45: Average tax credits vs. PHEVs sold in the US ........................................................ 144  Table 3.46: Regression results for PHEV price vs sales ............................................................. 146  Table 3.47: Registered PHEVs vs. charging stations per state ................................................... 147  Table 3.48: PEV Sales in 2013, Battery Size and Miles Run in One Charge ............................. 148  Table 3.49: Hypotheses H6 Results ............................................................................................ 149  Table 3.50: Categories of Security Breaches (Source: Privacy Rights Clearinghouse) ............. 155  Table 3.51: Breaches’ Victims Categories.................................................................................. 156  Table 3.52: Number of Breaches versus Type of Attack ............................................................ 158  Table 3.53: Number of Victims by Type of Attack versus Breach ............................................ 158  Table 3.54: Number of Breaches and Victims per State ............................................................. 159  Table 3.55: Detail of Victims per Type of Breach...................................................................... 160  Table 3.56: Types of Breaches versus Victimized Areas ........................................................... 161  Table 3.57: Breached Victims versus Victimized Sector ........................................................... 161  Table 3.58: Hypotheses H7’s Results ......................................................................................... 162  Table 3.59: Diffusion Results for Homes with Solar/Wind Energy ........................................... 163  Table 3.60: Diffusion Comparison of SG Technologies ............................................................ 167  Table 4.1: Elements of smart use of energy ................................................................................ 174  xi

Table 4.2: Statistics from Question # 4 ....................................................................................... 192  Table 4.3: Row and Columns Percentages of Bucket Assignments ........................................... 194  Table 4.4: ANOVA analyses for Bucket’s Ranks ...................................................................... 195  Table 4.5: Statistics from Society Responsibility’s Question ..................................................... 196  Table 4.6: Word mining of individual inputs.............................................................................. 198  Table 4.7: Residence States of the Survey Respondents ............................................................ 199  Table 4.8: Elements, Questions and Definitions ......................................................................... 200  Table 4.9: Loading and Cross-loading ........................................................................................ 201  Table 4.10: Latent Variables Coefficients .................................................................................. 201  Table 4.11: Correlations among I vs. Square root of AVE......................................................... 202  Table 4.12: P-Values for Model’s Correlations .......................................................................... 202  Table 4.13: SGMM Survey’s Questions ..................................................................................... 206  Table 4.14: Our Survey’s Questions ........................................................................................... 206  Table 4.15: Elements and Questions for the PLS Analysis ........................................................ 210  Table 4.16: Means and t-tests for Questions not used in the Model ........................................... 211  Table 4.17: Gender Differences on SG Technology (Means & Std. Dev.) ................................ 212  Table 4.18: Occupation Driven Differences on SG Technology (Means & Std. Dev.) ............. 212  Table 4.19: Year of Experience Driven Differences on SG Technology (Means & Std. Dev.) . 213  Table 4.20: Education Driven Differences on SG Technology (Means & Std. Dev.) ................ 214  Table 4.21: Location Driven Differences on SG Technology (Means & Std. Dev.) .................. 215  Table 4.22: Survey’s Sample Analysis ....................................................................................... 215  Table 4.23: Model Fit Results..................................................................................................... 216  Table 4.24: Loading and Cross-loading Results ......................................................................... 217  Table 4.25: Model’s Path Coefficients ....................................................................................... 218  Table 4.26: Model’s Path Coefficients p-values ......................................................................... 218  Table 4.27: Model’s Correlations of I vs. Square root of AVEs ................................................ 219  Table 4.28: Model’s Correlations p-values................................................................................. 219  Table 4.29: Model’s Latent Variable coefficients ...................................................................... 220  Table 4.30: Hypotheses H8’s Results ......................................................................................... 220  Table 4.31: Sources of Electricity Generation by Census Division............................................ 227  Table 4.32: Sources of Electricity Generation per State ............................................................. 228  Table 4.33: Generation Costs for Sources of Electricity ............................................................ 229  Table 4.34: States with Electricity Deregulation Status ............................................................. 231  Table 4.35: Competition by State and Ownership of Utility Companies ................................... 232  Table 4.36: Competition compared to Deregulation Status ........................................................ 233  Table 4.37: Cost per Energy generation Mix by State ................................................................ 234  Table 4.38: 27 Regulated Entities Regression Results with R2=0.794 ....................................... 234  Table 4.39: Regulated formerly Deregulated Entities Regression Results ................................. 235  Table 4.40: 16 Deregulated Entities Regression Results with R2=0.765.................................... 235  Table 4.41: Hypotheses H9’s Results ......................................................................................... 238  Table 4.42: Percentages of Tariff vs. Regulation Status ............................................................. 239 

xii

LIST OF FIGURES Figure 2.1: SWOT Analysis .......................................................................................................... 15  Figure 2.2: Log trend line of SG articles ...................................................................................... 18  Figure 2.3: Selected articles by categories .................................................................................... 19  Figure 2.4: Growth of SG by discipline ........................................................................................ 20  Figure 2.5: Time series for SG technologies (Cardenas et al., 2011) ........................................... 20  Figure 2.6: SG technologies per country ...................................................................................... 21  Figure 2.7: SG technologies by state ............................................................................................ 23  Figure 2.8: Normalized articles’ mentions of technology per state ............................................. 23  Figure 2.9: Distribution of Research Papers ................................................................................. 34  Figure 2.10: Primary purpose of papers ........................................................................................ 36  Figure 2.11: Analyzed papers by research purpose ...................................................................... 36  Figure 2.12: Research by collection approach .............................................................................. 37  Figure 2.13: Papers by data collection categories & time ............................................................ 37  Figure 2.14: Data analysis techniques used .................................................................................. 38  Figure 2.15: Data analysis techniques evolution in time .............................................................. 39  Figure 2.16: Distribution of papers by category ........................................................................... 40  Figure 2.17: Number of papers by type in time ............................................................................ 40  Figure 2.18: Categories of papers by topics.................................................................................. 42  Figure 2.19: Categories by time .................................................................................................... 43  Figure 2.20: Distribution of SGD technologies ............................................................................ 44  Figure 2.21: Trend of SGD technologies in time .......................................................................... 45  Figure 2.22: Distribution of papers by country............................................................................. 46  Figure 2.23: Trends of Papers by First Author’s Country and Year ............................................. 46  Figure 2.24: Conference papers by continent and year................................................................. 52  Figure 2.25: Journal papers by continent and year ....................................................................... 53  Figure 2.26: Conference papers’ topics by continent and Chicco’s categories ............................ 54  Figure 2.27: Journal papers’ topics by continent and Chicco’s categories ................................... 55  Figure 2.28: Category mixes versus time ..................................................................................... 56  Figure 2.29: ICT Papers by year & country .................................................................................. 57  Figure 2.30: ICT Papers by SG Technology and Country ............................................................ 57  Figure 2.31: Physical Infrastructure Papers by Country and Year ............................................... 58  Figure 2.32: Physical Infrastructure Papers by SG Technology and Country .............................. 59  Figure 2.33: Economics Related Papers by Country and Year ..................................................... 60  Figure 2.34: Economics related papers by SG Technology and Country ..................................... 60  Figure 2.35: Environmental Related Papers by Country and Year............................................... 61  Figure 2.36: Environmental Related Papers by SG Technology and Country ............................. 62  Figure 2.37: Text Mining Cloud ................................................................................................... 70  Figure 2.38: Distribution of papers by number of authors ........................................................... 72  Figure 2.39: DOE recovery awards .............................................................................................. 78  Figure 2.40: EPRI funded projects................................................................................................ 81  Figure 2.41: Academics distribution of Literature ........................................................................ 83  Figure 2.42: Number of Papers per Technology in the Preliminary Model Graph ...................... 92  Figure 3.1: Private investment in Energy’s sector around the world ............................................ 96  Figure 3.2: Evolution of consumption and generation of electricity ............................................ 98  xiii

Figure 3.3: Projected lines of consumption and generation until 2035 ........................................ 99  Figure 3.4: Evolution of the crossing point from the sensitivity analysis .................................. 100  Figure 3.5: Other uses of electricity to consider (source: UN data) ........................................... 103  Figure 3.6: Importations of electricity and current trend ............................................................ 104  Figure 3.7: Trend of global use and availability of electricity (source: UN data) ...................... 104  Figure 3.8: Projected lines of use and availability of energy (source: UN data) ........................ 105  Figure 3.9: Growth of electricity capacity assuming 6870 hours per year (source: UN data) ... 107  Figure 3.10: Sources of production of electricity (source: UN data) .......................................... 107  Figure 3.11: Capacity percentage -Production vs. installed capacity (source: UN data) ........... 108  Figure 3.12: Diffusion of Wind Generation by State .................................................................. 116  Figure 3.13: Diffusion of Solar Generation by State .................................................................. 117  Figure 3.14: Energy generation using coal (source: World Bank database)............................... 119  Figure 3.15: Electricity generated using hydro power (source: World Bank database) ............. 119  Figure 3.16: Electricity generated with natural gas (source: World Bank database).................. 120  Figure 3.17: Electricity generated using nuclear plants (source: World Bank database) ........... 120  Figure 3.18: Electricity generated burning oil (source: World Bank database) ......................... 121  Figure 3.19: Global energy generation sources (source: World Bank database) ........................ 121  Figure 3.20: Worldwide top electricity generation producers .................................................... 122  Figure 3.21: Diffusion Speeds of AMI for different states clusters ............................................ 128  Figure 3.22: Diffusion Curves by Cluster of States by Division ................................................ 131  Figure 3.23: Diffusion curves for the Top Ten States ................................................................ 135  Figure 3.24: Diffusion Speeds of AMI for different states clusters ............................................ 136  Figure 3.25: Diffusion curves by Utility Company Ownership .................................................. 137  Figure 3.26: Diffusion Speeds for AMI depending on Urban Concentration............................. 138  Figure 3.27: Electric vehicles produced in the U.S. ................................................................... 142  Figure 3.28: Average Gasoline Prices (Source: eia.gov) ............................................................ 143  Figure 3.29: Hybrid vehicles breakdown by brand ..................................................................... 144  Figure 3.30: Number of PHEVs sold versus average price ........................................................ 145  Figure 3.31: Plug-in Electric vehicles breakdown in the U.S. .................................................... 147  Figure 3.32: Types of Cyber Attacks based on Chen et al. (2012) ............................................. 153  Figure 3.33: Number of Breaches per Year ................................................................................ 156  Figure 3.34: Number of Breaches Victims per Year .................................................................. 156  Figure 3.35: Number of Breaches per Categories and Groups of Victims ................................. 157  Figure 3.36: Number of Breached Victims by Group................................................................. 157  Figure 3.37: Diffusion of Wind and Solar Generated Electricity by Households ...................... 164  Figure 3.38: Smart Meters Diffusion by Census Division and Total US ................................... 165  Figure 3.39: Electric Vehicles Diffusion in the US .................................................................... 166  Figure 3.40: Cyber Attacks Diffusion in the US......................................................................... 167  Figure 3.41: Smart grid Technologies Diffusion Comparison.................................................... 168  Figure 4.1: What is the smart grid? (Source: http://www.smartgrid.gov/the_smart_grid) ......... 170  Figure 4.2: Environmental Performance Index (Esty et al., 2006) ............................................. 173  Figure 4.3: ENSG Road Map for UK Smart Grid Deployment .................................................. 175  Figure 4.4: The European Union’s Smart Grid vision (source: VDE, 2010) ............................. 177  Figure 4.5: NIST Smart Grid Framework 1.0 ............................................................................. 178  Figure 4.6: Comparison of China and US/EU Smart Grids (source: Jiandong, 2011) ............... 179  Figure 4.7: Masdar: The Sustainable City (source: http://masdarcity.ae/en/) ............................ 180  xiv

Figure 4.8 Masdar City Energy (source: http://masdarcity.ae/en/) ............................................. 181  Figure 4.9: Texas Smart Grid Investment Model (Source: SGRC) ............................................ 182  Figure 4.10: Ontario Smart Grid (source: http://ieso-public.sharepoint.com) ............................ 183  Figure 4.11: Enhanced Model by Blocks related to PDCA Cycle .............................................. 186  Figure 4.12: Proposed Enhanced Model ..................................................................................... 187  Figure 4.13: Comparison of 2010 Census results and survey respondents................................. 190  Figure 4.14: Comparison of Educational Level of the Respondents vs. 2010 Census ............... 191  Figure 4.15: Household income comparison of 2010 census and survey respondents ............... 192  Figure 4.16: Survey smart grid focus’ importance evaluation .................................................... 193  Figure 4.17: Who is responsible for smart grid’s technologies? ................................................ 194  Figure 4.18: What is important for the society? ......................................................................... 196  Figure 4.19: Personal importance responses ............................................................................... 197  Figure 4.20: Range on cost of energy responses......................................................................... 198  Figure 4.21: Smart Use of Energy Survey Model....................................................................... 200  Figure 4.22: First PLS Model with Results ................................................................................ 203  Figure 4.23: Detailed Enhanced Energy Management Business Model..................................... 205  Figure 4.24: Second PLS Model with Results ............................................................................ 219  Figure 4.25: Total Electricity Consumption by Census Division ............................................... 225  Figure 4.26: Total Electricity Revenue by Census Division....................................................... 226  Figure 4.27: Cost of Electricity per Census Division ................................................................. 226  Figure 4.28: Regression Predicted Values for the Regulated Entities ........................................ 236  Figure 4.29: Regression Predicted Values for the Formerly Deregulated Entities ..................... 237  Figure 4.30: Regression Predicted Values for the Deregulated Entities ..................................... 237  Figure 5.1: Diffusion of Smart Grid Distribution Technologies in the US ................................ 242 

xv

CHAPTER 1: INTRODUCTION 1.1 BACKGROUND INFORMATION In the past one hundred years, countries around the world have been investing in monolithic transmission, distribution and generation infrastructures to support growing electricity needs usually referenced in the sequence of the value chain, generation, transmission and distribution. Unfortunately, these infrastructures have not been robust enough, as there is an estimated global loss of energy ranging from 10 to 52%, mostly due to distribution losses and thefts—which depend on the advancement of the available technology and implemented theft controls in every country (Najjar et al., 2012). These losses in the US are smaller than many other countries, for instance in 1995 they were 7.2%, with 40% of the losses coming from transformers and 60% from the lines (Hong & Burke, 2010). Considering some of the concerns that have an environmental impact, the transportation and energy generation sectors are the top contributors of CO2 to the atmosphere, with 20% and 40% of the emissions respectively (Lo & Ansari, 2012). These sectors mostly depend on burning hydrocarbon based fuels that create emissions which then affect the environment. There is an old initiative toward the generation of energy using renewable resources but it has not reached the right price yet, but at the same time, consumers are buying more devices for modern world needs, namely consumer electronics, which only increases the demand of energy and, as a result of this process, more energy needs to be generated, hence more harm to our environment. By the year 2030, the consumption of electricity throughout the world is expected to increase 76% (Ramchurn et al., 2012). In order to satisfy this required electricity demand, the actual generation processes need to increase, thus increasing the emission of CO2 and SOx to the atmosphere. To overcome this challenge, many countries are writing directives and goals to 1

reduce contamination in the short term. Battaglini et al. (2009) refers specifically to the many environmental protection requirements set forth by the European Community. To address these environmental concerns, there is an ongoing integration and growth of new cleaner sources of energy such as wind, photovoltaic, natural gas, nuclear, and others. Renewable resources have been growing as well, including nuclear generation which has reached 6% of the world’s total produced energy. However, following the incident provoked by the tsunami in the Fukushima nuclear plant in Japan on March 11, 2011, the reaction of the Federal Republic of Germany was to immediately close 8 nuclear sites and schedule the closure of the 9 remaining plants in 10 years (Römer et al., 2012). The rest of the world is also taking precautions to reduce or even eliminate nuclear energy generation, unless a breakthrough technology is discovered. With the task of reducing global contamination, scientists and engineers face the challenge of reducing contamination and making better use of the currently generated electricity via reduction of losses inherent to the distribution and transmission processes of the traditional grid. One possible solution to this problem is the Smart Grid, which is based on recent technologies: The Smart grid is characterized by: the efficient distribution of energy with the inclusion of state of computer power, the use of renewable resources to generate electricity, participation of the consumers in the process by generating and/or conserving energy, low latency feedback to consumers and utility companies about real-time consumption via smart meters to be able to take advantage of smart rates, the use of electric vehicles’ batteries to store and distribute energy at homes, and distributed energy resources, among others. All these Smart Grid elements are being developed and even implemented in some countries, while developing countries are waiting to see the results before following those steps.

2

The United States seems to be the leader in this effort with the support of the Electric Power Research Institute (EPRI), National Institute of Standards and Technology (NIST), Department of Energy (DoE), National Rural Electric Cooperative Association (NRECA), Edison Electric Institute (EEI), and the American Public Power Association (APPA) (Lo & Ansari, 2012). Other national governments are promoting efforts for the implementation of Smart Grids in the near future. For instance, Korea launched the K-grid project in 2002 (Son & Chung, 2009), India created the Indian Smart Grid Task Force (Mukhopadhyay et al., 2012), and China formed the Strong and Smart Grid (SSGC) (Uslar et al., 2012). There are important efforts in the promotion of Smart Grids in Europe, where one of the biggest concerns seems to be the implementation of advanced meters and “green” energy. The US, being leader in developing smart grids appropriated $4.5 billion of the American Recovery and Reinvestment Act (ARRA) to the Department of Energy (DOE) and the Office of Electricity Delivery and Energy Reliability (OE) for deployment of programs such as Smart Grid Investment Grant (SGIG) and the Smart Grid Demonstration (SGDP) Program The Electric Power Research Institute (EPRI) has also been working on this effort with the IntelliGrid Program which focuses on standards, interoperability and cyber security. When we consider the global amount of monies invested on Smart Grid research, the majority comes from the United States, with 31% as reported by Bloomberg New Energy Finance (BNEF). Although the US is the leader, analysts predict that China will overtake this leadership position because the Smart Grid program launched by the Obama administration comes to an end in the year 2015. At the same time, China is continuously growing, with an investment of $3.2 billion compared to the US that has already spent $4.3 billion in 2012.

3

The Smart Grid brings a two-way communication and flow of energy, instead of the traditional one-way flow from traditional electricity (Ramchurn et al., 2012) and information systems (Fadlullah et al., 2012). The US government has invested $3.4 billion dollars in grants for the investigation of Smart Grids (Güngör et al., 2011). As ICT has been evolving from wire-line to wireless media, there are important proposals about the concept of “ZigBee smart energy” with wireless communications to remotely control devices. The utility company or the consumer will be able to remotely turn appliances, or other devices, –on or off depending on their needs and based on the cost of energy or present environmental conditions. With all these technologies under the umbrella of Smart Grids, we chose to focus only on energy distribution for the literature survey in chapter 2. Distribution is a current fast-growing area and very visible to consumers, utility companies, and governments who are trying to involve the general public in this discussion. If distribution is enhanced, the expected result is energy conservation to avoid unnecessary investments in new large generating plants by reducing energy consumption. 1.2 THE BIRTH OF A NEW MODEL Through this dissertation we developed the advent of a new decentralized Energy Management Business Model that is changing the role of the consumer, utilities and government with the emergence of new and advanced technology. The elements of the model are analyzed and measured in regards to their impacts on the implementation of ICT in the energy management systems. Prior efforts have tried to focus on consumers’ participation via smart houses (Tanaka et al., 2012) interconnected with control systems that include distributed generators and storage. In order to understand the growth of renewable resources generation, and 4

in a way the diffusion of distributed generation (DG), we research and analyze all sources of electricity generation in chapter 3. Another component of the model also includes the vehicle-to-grid (V2G) as a source of energy, where the battery pack is charged while being connected to the grid, and later, the battery will become a major source of energy (Khayyam et al., 2012; Erol-Kantarci & Mouftah, 2011). Chapter 5 presents the diffusion of electric vehicles for both Plug-In Hybrid (PHVE) and Plug-in Electric Vehicles (PEV). Although EVs are not considered commonly as sources of energy, their role in the future seems to be critical to achieve the environmental goals because one of the major sources of contamination are the emissions of the fuel vehicles on the road. A factor that needs to be carefully considered is the timing and places for charging batteries of these cars, or the suburbs will collapse at the evening hours when most people go home and plug the car to the grid. There are many studies using Particle Swarm Optimization (PSO) as a strategy for the battery charging process during the night (Celli et al., 2012; Sousa et al., 2012; Silva et al., 2012) Other scholars focus on the outcome of their own models, environmental protection being one of the most mentioned, (Lo & Ansari, 2012), energy conservation (Feng & Yuexia, 2011), and social participation via consumers who explore energy resources possibilities (Aliprantis et al., 2010). Based on the prior models, we developed our own energy management model that includes some of the above mentioned elements that are further analyzed in the following chapters. 1.3 THE MOST IMPORTANT ELEMENTS Based on the literature review and analysis detailed in chapter 2, we consider information and Communication technology (ICT) as the main element of the model because it is the driver 5

of the model. Without contemporary advances in technology, it would be very difficult to look for the levels of efficiency that we can achieve nowadays. For instance, we can have a blackout and the backup source can be activated without our human senses perceiving it; also, informing consumers about the instantaneous price of electricity throughout the day could not be possible without computerized networks reaching every home and maintaining two-way communication. This communication has a potential issue, namely cybersecurity because with all meters connected to huge networks, it would be very difficult to develop robust enough systems to prevent attacks, and what is more critical and vulnerable for the energy sector is the false data injection, as presented in Chapter 3. Although these types of attacks are not frequent, the potential capability of detection is low, increasing the vulnerability of all energy systems, which now require a computerized system to monitor the requests and, compared to normal consumption patterns, they can detect these types of breaches. From the different technologies under the umbrella of Smart Grids (SG), we specifically focus on the most visible element for SG integration, that is, the Advanced Metering Infrastructure (AMI), or smart meters as they are called in Europe. Their diffusion can be representative of the SG progress, so we analyzed them using the Bass Diffusion model in Chapter 3. As a new business model was proposed by Lehr (2013) with focus on the utility companies, the analysis is conducted by region, division, ownership, and urban concentration of the population to better understand the trends and develop our prognosis. 1.4 THE ENHANCED ENERGY MANAGEMENT BUSINESS MODEL This dissertation contributes to the International Business literature by proposing an enhancement to some of the Energy Management Business Models that present the relationships 6

among the selected elements to determine if they are strongly influenced among themselves. The extensive literature survey extends the academic knowledge on the areas where literature is focusing and those areas that are being neglected at this time. The Model is presented in detail in Chapter 4 of this dissertation and is supported by two models that are similar and grounded in theory. We have conducted also two surveys that backup the model, as both the structural equation modeling (SEM) and Partial Least Squares (PSL) models support our proposal. In general, our model seems to cover the key requirements for a Smart Grid model, and it is supported by other research and our primary data. The enhanced model that we propose in this dissertation introduces how distributed generation and possibly the electric vehicle will provide electricity that will be stored in distributed storage to use this energy when necessary, not as generated. Once the energy is stored, it can be automatically distributed with ICT devices communicating through networks to the consumers that require it. Consumers in general will have the ability of responding to the tariff information, so they can increase or decrease consumption based on the real-time cost of this service. By reducing peaks and optimizing consumption, we expect efficient and environmental friendly energy. There are some new models but they are unilateral, with the utility company making all decisions, and consumers just accepting them. The enhanced model foresees decentralized generation where consumers are important actors in the process. Another figure that might come forward very strongly is the service providers who will do the distribution of the available stored energy in a free market. The consumers will also have an important role on the demand response area, as they will decide when to use energy and when not.

7

The involvement of all stakeholders in the process will certainly help for better results. Utility companies have traditionally made most decisions along with the regulating bodies, while the consumer perception and opportunities have not been explored yet. Allowing two-way communication with the consumer opens up a horizon of possibilities to make the distribution process more efficient. To validate the proposed enhanced model we used the literature survey to show how these elements are the most important ones for published researches at this time. The relationships among the elements were modeled and validated with a survey conducted to specialists in the field who show that the storage element is the only one that they do not see as important because we are in the process of designing better batteries, and this is the Achilles heel. We know that at this time there is no good option as of yet. The practical proof of the model is not part of the dissertation, because it would have to be implemented and we do not have the resources. This is a conceptual enhanced model. 1.5 RESEARCH QUESTION AND OBJECTIVES With the proposed model we aim to answer the following problem statement: How can we, as a global society, prepare for the imminent paradigm shift towards distributed generation and distribution automation with ICT and other technologies, which introduce a new business model where consumers might also be producers, whereby millions of connections can make our systems vulnerable, and the economics seem unfeasible? In order to answer this question, we divide the dissertation into three major blocks to achieve the following objectives: 

Understand the global position of Smart Grid based on peer-reviewed publications



Study diffusion models and risks for technologies enhanced with ICT

8



Analyze and design a business model with consumers being also producers of energy.

1.6 CONTRIBUTIONS OF THIS DISSERTATION This dissertation brings forth contributions to the business administration, information systems and energy areas. The Smart Grid is a topic that recently appeared and along with the Information and Communication Technology (ICT) is provoking major changes in the distribution of electricity throughout the world. Being such a new topic, there are not many studies focusing on this area and this dissertation contributes with studies never done before. The major contributions in this dissertation are: 

A Fad or fashion study for peer-reviewed literature, from Internet & Web of Science about Smart Grid, applied to this area for the first time.



An exhaustive literature survey was conducted on 966 papers about Smart Grid to classify them into six categories set by Chicco and the different technologies.



A model was developed for Green Efficient Energy based on global Roadmaps.



The perspectives of government, practitioners and academics on regards to Smart Grids were compared with a novel simple classification method.



Diffusion curves for Solar, Wind, Electric Vehicles and Cyber breaches for the first time.



Partial least Squares (PLS) models were developed to support the proposed model.



First evaluation of Smart Grid by 184 professionals not involved in the specific field, but responding as consumers of electricity.



A snapshot of the opinion of 32 specialists in the field about the elements in the model

 A proposed energy management model validated with a survey using questions from the Carnegie Melon Smart Grid Maturity model for utilities.

9

CHAPTER 2: COMPREHENSIVE RESEARCH IN SMART GRID 2.1 BACKGROUND INFORMATION It is projected that oil and gas reserves will be depleted by 2060-2065 (Klimenko et al., 2008). Needless to say, in the face of dwindling carbon based fuel reserves and fears associated with energy independence, there is considerable attention being paid to energy conservation. While there are numerous initiatives to conserve energy, one of the more promising approaches of achieving energy efficiency is a suite of intelligent technologies held under the umbrella of a Smart Grid. The Smart Grid relies on intelligent systems to make real-time decisions that can save energy without inconveniencing the consumer. Making the smart grid successful will require a creative and multidisciplinary approach from areas such as power and systems engineering, security, business intelligence, social networking, mathematical research, and others. (He, 2010) In 1940, 10% of the energy consumption in the US was used to generate electricity; in 2003 it was 40% (US DoE, 2003), and in 2012 the level still remains around 40%, according to the DoE website (http://www.eia.gov). The largest man-made contributing factor that harms the environment is the energy production processes that use CO2 emissions (Jiang et al., 2009). Among this and other factors, we are facing global warming, which is expected to increase 5 degrees in global temperature by the end of the 21st century, which has not occurred in 70 million years (Klimenko et al., 2009). For the past 25 years the construction of transmission facilities in the United States has decreased as energy demand increased, resulting in grid congestion. To prevent this situation, the Department of Energy (DoE) is working on the implementation of smart grids, following

10

President Bush’s signing of the project “Grid 2030” in 2003 (US DoE, 2003). Smart grids seem to be the future for energy conservation, as they are expected to save 10% of the energy used in the US by focusing on providing the required amount at the right time. The smart grid concept also includes other technologies, including: 

Demand Response (DR) to manage consumption responding to supply conditions



Electric Vehicle (EV)/ Plug-in hybrid electric vehicle (PHEV)



Distribution Automation (DA)  intelligent control over electrical power grid distribution level



Community Energy Storage (CES) presenting an alternative to store energy at suburbs



Advanced Metering Infrastructure (AMI) systems that collect and analyze energy consumption data. The nomenclature of smart meters was included in this category as it is the name used at Europe.



Distributed Storage (DS) as a smart way to reserve the available energy



Distributed Generation (DG) looking for a better way to generate decentralized electricity.

Although Smart Grids are very popular, there are still some unanswered questions about future uses, as its strengths and weaknesses are not recognized because their anticipated benefits have not been fully received yet. The purpose of this chapter is twofold: 

Clearly define what is expected from the use of smart grids; this can be accomplished by conducting a SWOT analysis using the available information; and



Investigate whether the multiple technologies under the umbrella of the smart grid concept are being accepted and promoted worldwide—in what jurisdictions and under what study areas. 11

The first part of the chapter will research definitions and develop a SWOT model. The second part of the paper uses a bibliographic analysis to identify mentions of smart grids in worldwide literature, applying the Management Fashion Theory (Abrahamson, 1991).

2.2 THEORETICAL BACKGROUND 2.2.1

Smart Grid Concept Defined

The largest machine with multiple interconnections in the world is the US power grid, which includes over 9,200 generators and 300,000 miles of transmission lines. The US power grid generates more than 1,000,000 megawatts (He, 2010). Modernizing and further developing this huge machine will require a clear focus on the goals of the project. To better understand SG, we define the concept as “an approach to modernize electrical distribution that would transform the way that a utility interacted with its customers in order to provide a higher level of service and reliability, put the customer in control of their energy costs, and to achieve energy conservation and sustainability goals” (Sarfi et al., 2010, p.200). After conducting thorough research, we can define Smart grids as efficient ways to conserve energy and prevent waste; they ought to be accommodating, as the future of energy might not be based on hydrocarbons but other sources of energy. Motivating users to do energy required activities at the proper time can be accomplished with smart grids. SG’s shall be quality focused to do things correctly every time. The opportunistic concept of the vision means that it will take advantage of any opportunity that might arise and integrate it as a plug and play. The resilient requirement of the model is critical, as the smart grid shall be prepared to resist any cyber-attack. Green is the name given to any environmental activity and the vision of a smart grid shall be environmentally compliant. And finally, the model has to be intelligent, which is 12

indeed the toughest requirement, as it is expected that the grid shall have enough information and programming that it would react smartly to any behavior (He, 2010) The architecture of a smart grid is very important but its decision making mechanisms are equally critical for they: • Shall be flexible to accommodate needs of different utilities, • Shall extend to the ever changing requirements, • Shall be open to interoperate with other different providers, • Shall handle and degrade faulty conditions such as noisy data (Davidson et al., 2010). Less essential, but certainly desirable, are extended capabilities such as utilizing all available data within a utility to influence operation, and allowing utilities themselves to select the level of automation for a given situation or scenario. The areas of application of smart grids include: smart meters integration, demand management, smart integration of generated energy, administration of storage and renewable resources, using systems that continuously provide and use data from an energy network (Davidson et al., 2010) 2.2.2 

SWOT Analysis

Strengths:

Self-healing systems are desirable to prevent dependence on human intervention at critical moments; by providing the systems with enough data, they can make smart decisions at the right moment: artificial intelligence (AI). With the tremendous growth of digital technologies, providing information faster and with fewer 13

errors in communication, the smart grid will utilize a digital platform (Jiang et al., 2009) Demand and load management are critical parts of the concept, as they helps to optimize delivery and consumption by reducing customer demands at peak hours (Liu, 2010) The smart grid shall not have a central vulnerable system that could deactivate the whole network using decentralized control schemes (Jiang et al., 2009) One of SG’s most important features is that it can be customized to specific needs/wants (Jiang et al., 2009) The future of hydrocarbon resources is looking weaker, Therefore, some generations have to consider the integration of intermittent renewable resources, such as wind, solar, etc. (Liu, 2009) Smart sub-stations ought to be autonomous and have enough information and data to optimize their continuous operation (Jiang et al., 2009) Another important feature of SG is that, due to the system transparency, we are able to see what is happening at all times in real-time (Liu, 2009) 

Weaknesses:

Cyber security anticipates compromises of adjacent systems. This has been a major concern area addressed by IT under SG (Overman & Sackman, 2010) A smart grid contains so many sensors and devices that it increases the system complexity for maintenance and repairs (Overman & Sackman, 2010) There could be failures in communications link, sensor and/or actuator, unplanned control center system failure, and nonexistent, late, or improper commands by untrained and/or distracted control room personnel (Overman & Sackman, 2010) As more modern and state of the art devices are integrated into the SG, there are possible compatibility issues (Bull, 2010)

14



Opportunities:

Cyber security controls will become more critical in future systems (Overman & Sackman, 2010) Balancing demand and generation using SGD can achieve optimal flow (Davidson et al, 2010) An information security active defense model will not only protect but also defend the system from attacks and unexpected responses (Zhang, 2010) SG can have decentralized storage areas to achieve the desired system balance (Slootweg, 2009) 

Threats:

Communication channels in the future may be more dedicated (Jiang et al, 2009), creating a need for dedicated conduits for SG, affecting cost and reliability of the system. Due to the complexity of grid, it might not be easy to provide technical support from a single source (Bull, 2010) As web applications are preferred targets of hackers (Bull, 2010), the SG might be attacked until its vulnerabilities are found. New regulations may have an impact on the grid as well (US DoE, 2003)

Figure 2.1: SWOT Analysis 15

2.2.3

Hypotheses H1 In this global environment that we live in, we are seeing that some concepts related to

conservation and better utilization of energy are in vogue. We expect that worldwide literature should also reflect this emphasis on Smart Grid Distribution (SGD), as this is one of the goals of this dissertation to achieve. Therefore, we present our first hypothesis which claims that there shall be an upward trend in scholarly literature about this subject. H1a: Smart Grids research shows growth of published papers in global literature. There is a possibility of SG becoming just a fad. To understand if it is indeed a fad or if it is becoming a fashion or a concept that will persist for a long time we will use the methodology used by Ponzi and Koening to analyze the lifecycle and diffusion of new concepts. Specifically, we will utilize bibliometric techniques (Ponzi & Koening, 2002). The lifecycle of a fad is shown as a quickly increasing concept in popularity that peaks and disappears very quickly, while a fashion grows more slowly, matures and stays atop for a while and begins to come down slowly (Abrahamson, 1991). It is our expectation that SG has not reached the maturity to begin declining, so we propose our second hypothesis: H1b: Smart Grid Distribution has not reached the maturity stage of its global lifecycle. As presented in the paper’s introduction, there are many different technologies that are correlated to the main concept of SG. Entities are pushing technologies according to their strategic plans and needs; therefore, if they have a different technological need, their definition and interpretation of SG is going to vary as well. Thus our third hypothesis proposes that: H1c: There is a different emphasis on Smart Grid Distribution technologies’ literature based on the number of articles published by regulatory jurisdiction. 16

As we conduct bibliometric studies, we shall propose a null hypothesis that the number of articles mentioning every technology is going to be equal for all countries and states. So our final hypothesis claims that: H1d: All technologies under the SG umbrella are equally mentioned in the literature 2.3 METHODOLOGY FOR TAXONOMY RESEARCH We counted the number of articles containing the words “smart grid”, eliminating those papers with citations only to ensure the selected articles are referring to the concept and are not only references. The research was conducted using Google Scholar for all papers published from January 1st. 2001 until December 31st. 2010. Once the information was retrieved, we developed charts with the annual counts of articles over this period on a yearly basis to see if the shape of the chart shows a fad, a fashion or a growing concept and so support hypotheses 1a and 1b. In order to correlate the main areas of study of SG, we only considered the following disciplines: engineering, business, physics, chemistry, environmental and social sciences. It is our expectation that the engineering, physics and chemistry study areas represent the technical aspect behind SG, while the business, environmental and social sciences provide the planning aspect—as they refer to goals, strategies, and non-technical implementation plans. Technologies under the concept of SG are also separated to support hypotheses 1c and 1d, because these technologies will show if there is a different perception in regards to SG at the different jurisdictions. We used only some of the technologies, DA, DR, EV/PHEV, AMI along with smart meters, DS, etc. We will finally probe countries and even states to correlate technology to every jurisdiction’s perception.

17

2.3.1

Google Scholar Research Results

Researching Google Scholar for worldwide articles containing SG and the above mentioned technologies, we were able to find 5,125 articles written in the ten year period. The trend chart shows exponential growth that can be better seen using a log scale— as shown in Figure 2.2. The number of articles written about “smart grids” has been increasing exponentially, and there are no signs of decrement. Based upon these findings we can support both hypothesis 1a and 1b, as we see a rapid growth of worldwide articles without showing signs of stagnation or decrement that may signal the “smart grid” concept being a fad. The trend is still going up or it is about to reach its maturity level before stabilizing and then going down. To better understand the study areas that are promoting this growth, we categorized of articles based on their area of specialization.

Figure 2.2: Log trend line of SG articles The area with the highest number of mentions is engineering, as the know-how about technical achievement of SG is being developed. Far in second place is the business area which accounts only for 20% of the engineering articles, as shown in Figure 2.3 18

Figure 2.3: Selected articles by categories To have a better view of the growth, and determine if the technologies are beginning to slow down and become stagnant, we calculated the percentage increments on a yearly basis per category. The chart shows that the social sciences, environmental and business areas are growing at slower rates than the other sciences. This might be a sign of them getting closer to reaching maturity. In other words, we might say that the foundation for SG in the social, environmental and business areas has been set and the maturity phase will begin with smaller growth and even decrease in the upcoming years. This can be seen in Figure 2.4, as social and environmental sciences are decreasing in growth. The previous charts support hypothesis 1a and 1b in that the overall trend in worldwide literature is still growing. The SG concept has not reached maturity, as there are no signs of stagnation or decrements in growth. In the business, environmental and social sciences, the trend seems to be slowing down and even reaching maturity, but this is still to be shown. On the other hand, engineering, physics and chemistry disciplines are still growing rapidly in literature about the concept and details on how to build and improve SG.

19

Figure 2.4: Growth of SG by discipline

Figure 2.5: Time series for SG technologies (Cardenas et al., 2011) Focusing on the technologies under the SG umbrella, we did the time study shown in Figure 2.5. This study introduces distribution automation as the first technology that was mentioned in SG literature, while CES is the last technology appearing in the papers. All

20

technologies are showing important increments in the past years, and none of them are giving signs of slowing down yet. The lines are showing exponential growth in most instances, such that we can conclude, looking at Figure 2.5, that Demand Response (DR) is the term more commonly associated with SG in the reviewed literature. This is not peculiar, as the smartness concept is linked to responding intelligently to the demand and supply energy levels. The 2nd technology in number of mentions was AMI, better known as smart metering, which has been highly accelerated in the past 2 years.

Figure 2.6: SG technologies per country We asked ourselves if the SG concept, technologies and trends are the same all around the world; so we took literature and data mined the definition concepts from Figure 2.5 and related them to the countries that are implementing it. The results show that the United States is number one in SG mentions in literature. The second place is almost a tie between the United Kingdom and China. After the general interest on demand response, the UK focuses on smart 21

metering while China’s main interest revolves around electric vehicles. Germany and Canada are close in fourth and fifth place with smart metering as their key interest after DR. Many other important points can be noted by looking at Figure 2.6. We support hypothesis 1c and reject hypothesis 1d because we can see that every country has a different perception of what SG is, and not all the concepts are mentioned equally. Because the country with the most articles related to SG is the United States, we further researched and subdivided the papers into the different technologies under the SG umbrella per state. There have been some successful implementations of SG technologies at some states, so data mining by states we found some interesting facts shown in Figure 2.7. California and Pennsylvania are the leaders in regards to technical papers about smart grid implementation, followed closely by Texas and Illinois. It is interesting to note that New Jersey is paying more attention to smart metering than demand response as all other states do. In the US, we can also support hypothesis 1c for every state’s perception in regard to SG, which are different and their levels are not equal—thus rejecting hypothesis 1d. The technologies mentioned for the US and the world follow the same path: demand response is number one at the selected countries and even states in America. Second and third place are also the same for smart metering and electric vehicles. For a better idea on the perceptions at every state, we normalized the results and calculated the percentage of written articles per state and technology to confirm the interest of every state in regard to the SG technologies. The results are shown in Figure 2.8. For demand response, the average percentage of mentions was 36%, with California and New Jersey being the least interested in this area with 23 and 22% of articles—although their interest was biased toward smart metering. Smart metering and electric vehicles are tied in second place in regards to mentions in the literature for the selected states. Electric vehicles are 22

receiving an important push from Colorado while smart meters have spread support throughout the states. On distribution generation Texas and Florida are the least interested states while New Jersey is the one showing the most interest.

Figure 2.7: SG technologies by state

Figure 2.8: Normalized articles’ mentions of technology per state

23

2.3.2

Preliminary Conclusions of Google Research

Based on the research conducted using Google Scholar, in the period from 2001 to 2010, we can conclude that the SG concept has been growing rapidly around the world. The concept has not reached the expected maturity level, but because the bibliometric study shows that SG has experienced growth in the past but is beginning to slow down now, we doubt that it is a fad. Until the number of SG papers stabilizes or decreases in number, we might be able to conclude if it is a fashion, as we expect the number of articles to reduce slowly until it goes to a minimal level; a fad would grow quickly, reach its peak, and then diminish rapidly. Thus we do not expect SG to be a fad (Cardenas et al., 2011). Dissecting the SG concept into the different technologies related to the SG concept, we are able to see that the individual components are also growing and none of them has reached the maturity stage yet; although the growth has taken over 7 years, so we do not expect them to be fads either. With the business, social science and environmental disciplines showing increases of around 100% in the past year, we can infer that these study areas seem to be slowing down compared to the engineering, physics and chemistry areas with yearly growths from 271 to 715%. Worldwide, we found that demand response is the concept more closely related to SG followed very closely by smart meters and electric vehicles. This is surprising given that distribution was expected to be more related than electric vehicles, showing a global perspective of energy conservation, enhanced distribution, optimized generation, and intelligent consumption breaking away from the current energy schemes towards a greener environment and smarter systems driven by business intelligence.

24

Table 2.1: Results of Hypotheses H1

2.4 LITERATURE SURVEY USING ISI WEB OF SCIENCE In order to use a more rigorous method for paper publication, we analyze and count the peer-reviewed articles published containing the words: “Smart Grid” & “Distribution”. Two different publishing sources were used: conference and journal papers. This research was conducted using the ISI Web of Science for serious academic peer-reviewed papers published from January 1st. 2008 until December 31st. 2013. It is important to emphasize that the first mention of SG happened in 2008, for that reason we are analyzing up to 2013, to include more than five years of information, an earlier version of this literature review was published by the Journal of Cleaner Production (Cardenas et al., 2014). In a period of one year, from the time when the original paper was written to the time of the conclusion of this dissertation, over 150 papers were added to ISI listing. Therefore, we expect the same to happen in 2013, thus we are not worried about the lower number of conference papers in the past year.

25

2.4.1

Hypotheses H2

While the field holds many publications, in this section we are going to focus exclusively on peer-reviewed papers. We expect that the behavior of academic reviewed articles is going to be the same as the more general Internet listing site. Thus, we develop the following hypothesis: H2a: Peer-reviewed conference and journal papers about SG are still showing growth. Smart Grid literature has been mostly related to engineering, specifically focusing on its technical aspects. It is our expectation that most serious literature will be devoted to developing theories for its Smart Grid implementation. Because SG is such a new concept, new technologies and new operating philosophies have to be developed to achieve the expected results (Klein et al. 2011). Thus, through the peer-review literature, we will analyze if the technology has matured enough to be at the theory building, testing or implementation stage. H2b: Smart Grid Distribution literature is more focused on theory building and testing than empirical implementation. If the diffusion of Smart Grid Distribution literature is reaching at least the early majority, there is going to be enough examples from case or field studies to be analyzed in the articles and develop prognoses for future implementations. As technology evolves, the need for accurate operating information becomes paramount (Bank, 2012.) Due to the newness of SGD, it is our expectation that there are more case studies than sources of data at this time: H2c: The majority of peer-reviewed papers collect data from case studies rather than other research sources. In the early 70’s Akao and others developed the Quality Function Deployment (Chan and Wu, 2002) This technique presents a sequence for planning the product or service all the way to process control; based on this theory, we expect strategy (planning) be the first part of the 26

implementation process, while control will be the last part. Because SGD is a new technology, we expect most papers to be related to strategy instead of process controls, or quality focus. The next hypothesis proposes that: H2d: Strategic papers have a leadership role over quality focus on efficiency and control. With the “green revolution” push in the new millennium and an increase on energy demand, the new energy has to be generated from renewable wind, solar and tidal resources (Ramchurn et al., 2012). If renewable energy is correlated to the SG concept, we expect that the generation of energy using renewable resources will be correlated to the number of papers written on SGD. The name for the renewable generation of energy is coined under distributed generation (DG) or distributed energy resources (DER). Although the necessary investment is high at first, the benefits are being studied throughout the world, such that we can expect that: H2e: Distributed Generation is steadily growing in importance throughout the world. The United States has been the leader on some technologies in the past. Recently, the US has been the creator of programs such as Intelli-Grid, Gridwise, Modern Grid Initiative, and Smart 2030, which have been propelled by the stimulus plan (ARRA). With this investment in Smart Grids, we expect the US to be the leader in the generation of literature focusing on planning and strategies for the future. H2f: The US is the leader in peer-reviewed SGD literature Considering the country of origin of the writers, it is our expectation that most papers for both conferences and journals would be from the US, but what about the economic blocks? Is America a leader in peer-reviewed conference and journal papers when compared against the European or Asian blocks? Because America is very much represented by the US and Canada,

27

we expect that the leadership position might be challenged by the European Union or the Tigers of Asia, but America may still be the leader. H2g: America is the leader on SG conference papers in the world H2h: America is the leader on SG journal papers in the world Conference papers are preliminary journal papers in nature, so we expect that the number of conference papers will predict the number of journal papers in that subject matter. Because the number of peer-reviewed conference papers is much higher than journal papers, that relationship might be linear: H2i: Journal papers follow a linear regression to the number of conference papers in the SG subject. 2.4.2

Research Purposes

Following the method used by Cardenas et al. (2014) and Gupta et al. (2006) the selected published papers are also classified into the three categories of research: theory building, theory verifying, and theory application. Theory building is going to include the published papers that present new theories or formulate existing ones; theory verifying include the research that prove previously presented theory, while theory application are practical papers that present how the technologies are implemented in the field. In the category of theory-building authors develop new relationships, algorithms, or hypotheses to be confirmed in future research papers (Kleinberg et al., 2009). Although this literature survey is written from the social science perspective of Information and Decision Sciences, these papers are required to be conceptual and also contain empirical data to support the author(s)’ hypotheses. For example, Tom Jauch (2009) proposes the coordination of Volt/VAR Management systems and equipment with an innovative adaptive technique to 28

automatically adjust to changing loads, circuit changes, reverse power applications, and circuit switching and reconstruction. Sanz et al (2009) also researched the introduction of electronic circuitry in the control of distribution to optimize the process. More recently, Dukpa and Venkatesh (2010) proposed a new model for distribution systems using fuzzy charts. For theory-verification, the authors conduct tests to prove those hypotheses previously proposed in earlier research (Saleem et al., 2009; Vokony and Dan, 2009). For examples, Tenti et al. (2010) introduced the token ring approach to minimize the lost energy where a voltage control algorithm sets a progressive reduction of the voltage at the other end down to zero. Moreover, Li et al. (2008) examined the system configuration and control methods at the typical radial distribution network, where locations of DEs are affecting the grouping of the buses; therefore, they affect the reduced network structure and the appropriate range of the controller parameter. The most commonly observed type of empirical research for Smart Grid Distribution is the application of existing theories, models, or frameworks. For example, Bushby (2009) focused on the Building Automation and Control Networks (BACnet) to provide a standard, network visible interface to configure and manage facility load shedding operations. Son et al. (2009) applied the principles of Smart Grid in Korea under the name K-grid established there since 2004; and Kirkham (2009) presented research using four different innovative methods to measure current efforts to address the issues of transduction and isolation from the circuits measured. Smart Grid Distribution (SGD) is a recent development, the focus in the past years was on energy transmission, and hence there are not many available sources for data collection. However, many of the papers on SGD refer to simulations and pilot runs before implementation. These simulations and pilot runs could help prevent catastrophic events if the proposed models

29

are not validated or even tested beforehand due to the size of the endeavor. Due to the fact that empirical studies are designed to focus on the effectiveness of proposals for new theories (e.g., algorithms to forecast electric vehicles charging process (Soares, J. et al. 2012)), it is not surprising that 60.5% of the reviewed papers are empirical studies, with a quarter (25.0%) being conceptual, and only 13.8% focusing on developing new models. In this dissertation we classify the data-collection approaches for empirical research in Smart Grid’s Distribution into these categories: case study, field research, laboratory research, archival research, and surveys (Gupta et al 2006). The SGD implementation process is in the early stages; therefore, we were only able to identify 314 case studies in the analyzed papers. Also, there were utilities and other companies’ data collection efforts over short periods of time. The scope of case studies varies, ranging from radial distribution networks (Calderaro et al. 2011) to a proposed roadway microgrid to be deployed at Lincoln, Nebraska (Qiao et al. 2011), and a super-grid implementation at Sardinia, Italy (Purvins et al. 2011). A field research study (18.3%) is similar to a case study, but it does not require many visits to the project being analyzed; only one or two visits would be enough for the data collection purpose. In other words, we only need a short period of time to gather the necessary information needed to make conclusions (Gong & Guzmán, 2012; Son & Chung 2009). Some of the examples of field research studies include Carlson et al.’s (2009) review of the application of cutout type reclosers, and the work of Lindsey (2010) on the application of line sensors in the field. The laboratory research category (27.2%) includes simulations and experiments that are done under controlled conditions to discover new theories. Because of the newness of the Smart

30

Grid technology, many of the published papers are about experimentation to simulate the possible conditions of circuitry, controls, and processes beforehand (Mohammed et al. 2010; Lo and Chen 2011). Examples of laboratory studies include an evaluation of the accuracy for an islanding impedance detection system conducted by Fort et al. (2010) and the development of a Smart Grid test bed using emerging technologies such as multi-agent systems (Belkacemi et al., 2011). Archival research (21.2%) refers to the use of secondary data obtained from sources other than direct data collection. Examples of this research include the use of publicly available data from public or companies’ sources. This category also includes general information or comments about Smart Grids that have been published and are available for reference. Because prior experiences on the Smart Grid technologies have not been documented, some papers only present comments about public information, while others use these sources of information for further researches (Liu 2010; Bilgin & Güngör, 2012). Two examples of archival researches are the worldwide annual electrical power generation, reported from Siemens databases in the paper written by Bosselmann (2010), and the information about the McAlpine substation at Charlotte, NC as presented by Miller et al. (2012) When there is no information available due to cost, time, or other constraints, there is the possibility of conducting a survey to gather information and come up with conclusions. In this particular area, there were only seven of these cases, one example being Pakonen et al. (2012) who surveyed 18 Finnish distribution network operators (DNOs) about smart meters. In this section, we discuss the analysis techniques mentioned most in the published literature. From all of the analyzed papers, 215 of them contain concepts but do not have any data analysis (e.g., Galli et al 2011; Heydt 2010). In most of these cases, the papers present

31

challenges or opportunities detected by the authors and suggest possible alternatives but do not proceed further; therefore, the next research steps should include data collection. Simulation is the technique that is used the most in the literature, which is not surprising due to the newness of the Smart Grids in distribution and the scope of the implementation challenge (e.g., Gudi et al 2012; Strasser et al 2011). The next categories are time series (e.g., Acampora et al. 2011; Giri et al 2012), descriptive statistics (e.g., Cheema et al. 2010; Ochoa et al 2011), and mathematical modeling (e.g., Andreotti et al 2012; Mohagheghi et al 2009). Smart Grid Distribution literature is evolving because the analysis technique that is mostly used is the simulation of what is going to happen when this technology is implemented. Simulation has the purpose of providing confidence for implementation, and a successful simulation shall help the scholars identify opportunities, and enhance the process before investing in a major implementation. Charging electric vehicles’ batteries at certain times (Han et al. 2012), using wind power for distributed generation at homes (Wang et al. 2010), and strategies for demand response (Fuller et al. 2011) show the ample field of simulation approaches discussed there. 2.4.3

Research Methodology

For this literature survey, we queried Reuter’s ISI Web of Science using Smart Grid and Distribution as selected topics with no restrictions other than requiring the papers to be peerreviewed. The result was a list of 966 papers in a 6 year period from 2008 to 2013; that is, there was no mention of Smart Grid distribution prior to 2008 as catalogued by the ISI Web of Science. As we reviewed these papers, 688 of them were proceedings from conferences and society meetings, while 278 were journal papers. The trend of journal papers on this topic has been climbing steadily, while in regards to conference papers the last two years seem to be on the 32

same level. In the year 2013 we are seeing as many conference papers as in 2012, but the number of conference papers included in the listing is smaller than 2012. So, it is our expectation that once all conferences are added, the number of papers will be much higher. Looking at all listed papers thus far, the total number of papers increased from 151 to 324 in the period 2011-2012, and then to 345 in 2013 (even though the information is not complete). Based on these numbers, we can support hypothesis H2a that there is still growth on peerreviewed papers on SGD. Probing deeper into all selected papers, we classified them according to the categories shown in Table 2.2, and the next sections describe our findings. 2.4.3.1

Classification by Research Categories

This survey classifies the articles as empirical research if they contain any real data. Papers outside this category were classified as “modeling and analytical methodologies” based on mathematical functions and/or simulated datasets, “conceptual and general category” (including papers without any data and based on comments about the topic), and the final category is called “survey and review” which consists of papers reviewing the existing literature. Out of the 966 reviewed papers, a total of 584 fell into the empirical research category and the remaining were classified as modeling, conceptual, survey or education-related published papers. Figure 2.9 illustrates the distribution of papers in these categories where empirical research is the majority of all papers. This is expected because the different technologies under the umbrella of Smart Grid have practical applications and are being analyzed before they are implemented at macro levels. Because there are important differences among peer reviewed papers depending on the publication outlet, another important delineation is by use of their

33

publication source. Some articles are published at conference proceedings or general meetings of engineering societies, while the others are published for academic and scientific journals. Table 2.2: General classification of this paper Classification Research Classification Research Focus Data Collection Data Analysis Types of Category Taxonomy of Paper’s Purpose Smart Grid Categories Originating Countries

Components of the classification Empirical Research, Modeling & Analytical Methodologies, Conceptual and General, Survey and Review, and SG Education Theory building, theory testing, and applications Case study, surveys, field, laboratory and archival researches. Descriptive statistics, neural networks, mathematical modeling, regression, Cronbach’s alpha, game theory, DEA, time series, simulation, non-linear and linear programming, or non-data analysis Physical (P), Regulatory (R), Environmental (E), Social (S), Economic (F), Information and Communication technology ICT (I) Strategy, Quality Management, Supply Chain Management, Environmental Issues, Service Design, Process Design, Scheduling, Planning, and Blackout Demand Response, Automated Meter Reading, Distribution Automation, Distributed Generation, Electric Vehicles, Energy storage, Cybersecurity, Vol/VAR Analysis, User friendly enhancements, Efficiency, and Self-healing United States, China, Canada, Germany, Italy, United Kingdom, Japan and others

Figure 2.9: Distribution of Research Papers

34

2.4.3.2

Classification by Research Focus

Separating the reviewed papers based on the purpose of the research, we classified them into three categories: theory building, theory verifying, and applications. Theory building is second in order with a 14.1% contribution, theory testing is third with only 9.2%, and application papers are identified as the major purpose of research, accounting for 76.7% of all analyzed papers. Although there are many research papers focusing on theory building, the percentage is low compared to the overwhelming amount of application papers. Figure 2.11 shows an increase of papers on theory building until 2010, when the trend reached a peak and then proceeded to descend to 40%. However, theory-building papers did come back up in 2012 to a higher level than the one in 2010. Theory testing is consistently trending upwards while application papers are showing a positive trend, although the overall number of published articles in 2013 is going to still grow. Application papers focus on the practical implementation of technologies or ideas; therefore we are seeing a large number of application papers as practitioners are analyzing SG for implementation. Authors seem to be focusing on the process of implementing those practical applications around SG, based on field and lab results without proper theory supported, but with a strong pragmatic support. Based on the numbers shown on both Figures 2.10 and 2.11, we can reject hypothesis H2b, as we can see that the number of papers focusing on applications are now higher than the theory based ones. Our prognosis is that the applications’ papers will continue growing for a longer period of time, while the theory related ones will still climb until the reach the peak and begin descending.

35

Figure 2.10: Primary purpose of papers

Figure 2.11: Analyzed papers by research purpose 2.4.3.3

Classification by Data Collection Method

Empirical research papers can be classified by the data-collection method used. Almost one third of the papers used case studies (32.5%); Archival research accounts for 21.2% of all papers and laboratory research also accounts for 27.2%. It is not surprising to see case studies as the majority of data collection efforts because SGD technologies are recent and being fine-tuned in the field by simulating and quantifying their impact to a specific selected case. Later this case will be compared to others under different characteristics to validate the hypothesis and shape

36

theories. That is, case studies represent a grounded reality. Figure 2.13 shows that trends are positive in most cases. Field and lab research are at the highest level in the period of time analyzed, showing increased interest in collecting data at these locations as real life validations are needed. Considering all these facts, we support hypothesis H2c, as we see that case studies are the primary source of data collection, but as this data is available in time, the use of archival information will increase.

Figure 2.12: Research by collection approach

Figure 2.13: Papers by data collection categories & time

37

2.4.3.4

Classification by Data Analysis Technique

Papers can also be classified based on the numerical analysis techniques used. Although some of the published papers have no data analysis, we identified 12 techniques used in the analyzed papers. Some papers use more than one technique, but for analytical purposes we chose the one technique which is the focus of the paper. Figure 2.14 weights the relative importance of the techniques used when measured as a percentage of the total incidences (966).

Figure 2.14: Data analysis techniques used Simulation is the analysis technique used most frequently in the analyzed papers with a 35% contribution and continuing on an upward trend. We expected to find such a large number of simulation papers because 45.8% of the case studies studying applications use simulation as the main tool and 17% of the simulations use the Monte Carlo method. Papers without data (22%) are also showing an upward trend as new techniques, opportunities and challenges are constantly discovered and discussed before conducting any data analysis.

38

Figure 2.15: Data analysis techniques evolution in time 2.4.3.5

Classification by Discipline

To separate the categories of papers, we used the six layers presented by Giannfranco Chicco (2010) identified on the basis of interaction between elements in the Smart Grid papers. It is important to emphasize that there are going to be overlaps as two or more layers can be addressed in a single paper. Table 2.3: Disciplines and Descriptions (Chicco, 2010) Category Physical (P) Regulatory (R) Environmental (E) Social (S) Economics (F) Information (I)

Description of Category Network interconnections used for the distribution of energy Applying standards, rules and other types of incentives or penalties to moderate the activity in the market, control prices and/or protect consumers Activities to have an impact on the defense of human life and protection of the planet Customers’ willingness to do something by investing on devices, changing consumption habits, participating in the utility companies’ programs, etc. Those activities that have an impact on the economy of the consumer or the company Includes efforts to integrate ICT in the distribution of energy via management, communications, and control

39

The Smart Grid literature has been evolving from an early focus on transmission and generation with unilateral flow of communications and electricity to bidirectional communications whereby electricity flows from the generator to the consumption location and from the consumption location distributed generator to the grid. Using the definitions in Table 2.3, we identified the papers with the six categories using the first letter as, identifier, with the exception of economics changed to financial (F) to avoid repeating (E) which was already being used for environmental papers. All categories show upward trends to date.

Figure 2.16: Distribution of papers by category

Figure 2.17: Number of papers by type in time

40

An important finding on this chart is the increase of economics related papers in the past three years. These papers tended to focus on efficiency because the energy generation and conversion process are of low efficiency (Dovì et al., 2009). This financial analysis has not peaked yet, and is rapidly growing to overtake second place from physical interconnection papers. The papers show an increased awareness on the financial impact of SG with the appearance of a new business model based on the elements of distributed generation, electric vehicles and automated metering that are inspiring new customer behaviors. The role of the consumers and producers merging together as prosumers will bring added complexity to the commercial relationship between utilities and entities generating energy for self-consumption because they can also sell their excess capacity to the utility company. In the same way that communication and energy now are flowing bi-directionally instead of unilaterally, energy costs will also have tariffs from and for the utility company, creating a new situation that has not been addressed recently. This new model is an effect of the upward trend of economics papers related to SG because it will require participation of many stakeholders who were previously passively involved in this process. The social and cultural changes from this new business model have to be analyzed in detail before companies fail to involve consumers and implement SG with their strong opposition. On the other hand, if companies successfully involve the stakeholders in the enhancement of the generation, distribution and consumption processes, the result could bring forth a new quality of life for consumers in general. Chapter 4 presents this enhanced model that has been developed within this dissertation. 2.4.3.6

Taxonomy of Paper’s Purpose

By analyzing the general purpose of the published paper, we use several categories to sum up a given project’s intention. The categories include: Strategic, Quality Management, 41

Supply Chain Management, Environmental Issues, Service Design, Process Design, Scheduling, Planning, and Blackouts. Strategic papers will present clearly defined strategies or guidelines for future implementation. Quality Management focuses on optimization and performance. Supply Chain Management centers on the continuous supply of inputs. Environmental Issues papers are concerned with reducing environmental impacts (“greener” planet). Service and Process Designs are very similar because energy provision is a service, but the focus on the generation, transmission and distribution process will be considered as process design, while service design is related to the activities around the process to aid and enhance the customer perception; Scheduling is related to Planning in time; while the Planning category is used for those activities as developed beforehand. Finally, Blackouts papers concentrate on avoiding power outages. The results are presented in Figure 2.18 and their trends are presented in Figure 2.19. The three most important topics in the reviewed literature are Strategy (31%), quality management (29%), and supply chain (14%). It is not surprising to see strategy as the leading category because the alternatives for proper implementation are still being developed in some of the technologies.

Figure 2.18: Categories of papers by topics The value in quality management is increasing due to the growing importance of using resources efficiently. The efforts to optimize energy distribution are aiming to save some of the 42

typical loss of generated energy due to energy transportation losses (Görbe et al., 2012), which in the US account for 5-12% (Lauby, 2010) and 15 to 20% in India (Sinha et al. 2011). The third important category is the supply chain, which describes how energy users have changed from only being consumers to now becoming “prosumers” who generate and consume energy (Zhabelova & Vyatkin 2012).

Figure 2.19: Categories by time Figure 2.19 shows how Quality Management is overtaking strategy in 2013 in total number of papers, which confirms the increasing notion of a SG definition as a way to look for efficiency (Wissner, 2011). Strategy is still strong because some areas of SG have not been fully developed, such as electric vehicles. EVs bring new and expensive interactions including driving range, recharging time (Silvester et al. 2012) and vehicle charging, which still needs to be considered before massive implementation can be accomplished (Shuaib, 2012). Based on the numbers and trends, we reject hypothesis H2d because we see that strategies have been passed by quality.

43

2.4.3.7

Smart Grid Technologies

There are several technologies under the umbrella of Smart Grid Distribution (SGD); the most typical ones are Advanced Metering Infrastructure (AMI), Demand Response (DR), Distribution Automation (DA), Electric Vehicles (EV), Distributed Generation (DG), System Efficiency Improvement, Self-Healing, Cybersecurity, and Distributed Storage. Looking at the 966 papers, we classified them into 11 categories. Figures 2.20 and 2.21 show distribution and trends. Only efficiency and Volt/VAR analysis show reduction of papers beyond the 2010 level, all others show positive trends as techniques are becoming more popular and important to our society. The three most important technologies based on the number of papers are: Distributed Generation (23%), Distribution Automation (18%), and System Efficiency (16%). These three areas show the current trends of SGD. Distributed Generation is a growing area mostly in Europe, where there is a major effort to move to self-generation using renewable resources that could reduce 2-4% of the energy losses in Europe (Lasseter, 2011); based on this growth we support hypothesis H2e.

Figure 2.20: Distribution of SGD technologies

44

There is also an important effort to automate the distribution process with the use of ICT devices. The automation process will allow quicker and better responses on energy consumption that will result in efficient use of energy, thus reducing the negative impact on environment (Koroneos at al. 2012). Additional methods to make energy distribution more efficient vary from replacing transformers with new more efficient ones (Grigoras et al. 2010) to the identification of the most probable configuration of the wind to locate mills (Chertkov et al. 2011).

Figure 2.21: Trend of SGD technologies in time 2.4.3.8

Originating Countries

Figure 2.22 shows the countries where the papers originated, i.e. the home of the first authors, the country referred in the paper, or the sponsor country of the publication. More than one quarter of the papers are from the US, which generated 27% of all literature, followed by China with 13%, Canada with 6%, Italy with 5% and India with 4%. Figure 2.23 shows the continuous increase of papers coming from other countries not listed in the top 5. The diverse composition of this category shows a worldwide interest in the area. The US is still maintaining leadership while other countries are making smaller contributions. These figures support 45

hypothesis H2f, which states that the US is the world leader in peer-reviewed SGD literature generation. Something important to emphasize is the growth of literature from other countries.

Figure 2.22: Distribution of papers by country Even though the top 5 represent 55% of all papers analyzed, the number of contributions from other countries is increasing, as there is growing worldwide attention on this subject. Some countries are specializing in certain technologies. For instance, Portugal is writing more papers about electric vehicles than most countries, with a particular focus on particle swarm for vehicle to grid scheduling (e.g. Soares et al. 2012, Sousa et al. 2012).

Figure 2.23: Trends of Papers by First Author’s Country and Year 46

It is interesting to note that as many as 81 countries have participated in the generation of literature. We see a technological wave throughout the world to meet the upcoming challenges and to be prepared for implementation. Environmental protection and energy sustainability are two of the many subjects addressed globally and their importance is critical. Analyzing matrices by doing cross-tabulation between columns and rows, we have identified some facts that are important for this area of knowledge. Table 2.4 shows that applications have the major contribution for the published papers with a 77% contribution, followed by theory building in the modeling and analytics area. We are seeing more application papers of Smart Grid Distribution because this area is being optimized with the aid of technologies already available and previously used for transmission (e.g., SCADA, D-SCADA) or in the areas of communication (e.g. wireless technologies). The empirical analysis is the largest share of the pie with 60%, followed by the conceptual papers with less than half of the empirical contribution. 92% of the conceptual papers are focused on practical applications. Table 2.4: Row & Column Percentages for Research Classification versus Focus Application Empirical Research Conceptual & General Modeling & Analytical Survey SG Education Total

58% 30% 11% 1% 0% 100%

Theory Theory Theory Theory Total Application Total building Testing building Testing 68% 72% 60% 73% 16% 11% 100% 10% 6% 25% 92% 6% 2% 100% 21% 22% 14% 63% 22% 15% 100% 0% 0% 1% 100% 0% 0% 100% 0% 0% 0% 100% 0% 0% 100% 100% 100% 100% 77% 14% 9% 100%

Table 2.5 shows that case studies is the data collection method used by the largest group of papers with 33% of all analyzed articles. Archival research is used mostly for applications (89%) because secondary data sources are used to represent the way things are in real life, not to build or test theory. For theory building and testing, authors are writing around 34% of the 47

papers using case studies, followed by laboratory research with around 29 and 37% of the papers. These numbers show that using implementations in the field and case studies represent 51% of all papers confirming the expected focus on practical applications for future implementation validation. Table 2.5: Row & Column Percentages for Data Collection versus Paper’s Focus Application Case Study Lab Research Archival Research Field Research Surveys Total

32% 26% 25% 17% 1% 100%

Theory Theory Theory Theory Total Application Total building Testing building Testing 35% 33% 33% 75% 15% 9% 100% 29% 37% 27% 72% 15% 13% 100% 11% 9% 21% 89% 7% 4% 100% 24% 21% 18% 71% 19% 11% 100% 0% 0% 1% 100% 0% 0% 100% 100% 100% 100% 77% 14% 9% 100%

Table 2.6: Row & Column Percentages for Data Analysis vs. Paper’s Focus

Application Simulation No data analysis Time Series Descriptive Model Game Theory Neural Networks Genetic Algorithm Linear Regression Non-linear Crombach's Alpha DEA Total

34% 25% 21% 13% 4% 1% 1% 1% 1% 0% 0% 0% 0% 100%

Theory Theory Theory Theory Total Application Total building Testing building Testing 40% 45% 36% 73% 16% 12% 100% 15% 11% 22% 86% 9% 5% 100% 19% 17% 21% 80% 13% 8% 100% 13% 15% 13% 75% 15% 10% 100% 3% 6% 4% 75% 11% 14% 100% 4% 2% 1% 36% 45% 18% 100% 1% 0% 1% 90% 10% 0% 100% 1% 2% 1% 56% 22% 22% 100% 3% 0% 1% 56% 44% 0% 100% 0% 1% 0% 67% 0% 33% 100% 1% 0% 0% 67% 33% 0% 100% 1% 0% 0% 0% 100% 0% 100% 0% 1% 0% 0% 0% 100% 100% 100% 100% 100% 77% 14% 9% 100%

Table 2.6 shows that simulation is the technique used most often with 36% of the total papers, mainly because the implementation of automation and new developments in this area in 48

the field need prior validation to prevent catastrophic failures. The creation of algorithms to simulate consumer patterns was also used to build theory, while some other papers simulated current or expected conditions to test already developed theory. Simulations represent a larger share than theory testing (45%), theory building (40%) and applications (34%). Table 2.7: Row & Column Percentages for Conference papers’ SG technologies vs. Purpose Strategy Distributed Generation Distribution Automation Efficiency Distribution Network Electric Vehicles AMR/ AMI Cybersecurity Demand Response Self-healing Distributed Storage Vol/VAR User friendly Total

Supply Supply Process Service Quality Process Service Quality Chain Planning Others Total Strategy Chain Planning Others Total Mgt Dsgn Dsgn Mgt Dsgn Dsgn Mgt Mgt

15%

18%

70%

9%

0%

19%

28% 23%

21%

24%

39%

3%

0%

5%

8% 100%

25%

18%

3%

31%

11%

26%

9% 19%

41%

28%

2%

13%

4%

8%

3% 100%

19%

20%

3%

9%

14%

26%

7% 16%

37%

38%

3%

5%

6%

10%

3% 100%

10%

11%

0%

7%

7%

0%

5%

8%

40%

42%

0%

8%

6%

0%

4% 100%

4%

6%

9%

4%

7%

7%

28%

7%

16%

27%

16%

4%

6%

6%

24% 100%

3% 6%

11% 6%

0% 0%

4% 4%

23% 20%

5% 7%

0% 0%

6% 6%

16% 33%

51% 31%

0% 0%

5% 5%

23% 23%

5% 8%

0% 100% 0% 100%

7%

4%

0%

15%

7%

5%

5%

6%

39%

21%

0%

21%

8%

5%

5% 100%

6%

1%

2%

5%

7%

0%

16%

4%

40%

10%

7%

10%

10%

0%

23% 100%

3%

1%

9%

4%

2%

2%

2%

3%

30%

13%

35%

9%

4%

4%

4% 100%

2% 0% 2% 0% 0% 0% 100% 100% 100%

13% 50% 31%

27% 50% 30%

13% 0% 13%

33% 0% 8%

7% 0% 6%

7% 0% 6%

0% 100% 0% 100% 6% 100%

1% 2% 2% 9% 2% 0% 0% 0% 0% 0% 100% 100% 100% 100% 100%

Tables 2.7 and 2.8 show the relationship between the Smart Grid technology and the focus of the conference and journal papers. Table 2.7 shows the relationship for conference publications only. As shown in Table 2.7, strategy is the most published paper focus followed closely by quality management and supply chain. Strategies for distribution automation (25%) and efficiency (19%) represent 44% of all published strategies. This is due to the new definition of the Smart Grid as the automation of distribution with the inclusion of ICT for optimized results. The focus of conference papers for distributed generation is most importantly related to supply chain (70%), as this concept of home generation of own energy is growing rapidly in some countries. Planning papers are focusing on achieving efficiency (26%) and distribution 49

automation (26%), but the approach is beyond strategy and moving into the “how to” mode. The process design papers focus mainly in distribution automation (31%), demand response (15%) and efficiency (9%) because these are the practical papers— “the nuts and bolts” of the Smart Grid implementation. Table 2.8: Row & Column Percentage for Journal Papers’ SG technologies vs. Purpose Strategy Distributed Generation Efficiency Distribution Automation Cybersecurity Distribution Network Electric Vehicles Demand Response Self-healing Distributed Storage AMR/ AMI Vol/VAR User friendly Total

Supply Supply Quality Process Service Quality Process Service Chain Planning Others Total Strategy Chain Planning Others Total Mgt Dsgn Dsgn Mgt Dsgn Dsgn Mgt Mgt

10%

20%

68%

9%

0%

22%

30% 22%

15%

26%

48%

3%

0%

3%

5% 100%

21%

22%

0%

4%

15%

33%

10% 16%

42%

40%

0%

2%

7%

7%

2% 100%

21%

14%

5%

22%

10%

22%

10% 15%

45%

26%

5%

12%

5%

5%

2% 100%

12%

9%

0%

0%

20%

11%

0%

8%

48%

30%

0%

0%

17%

4%

0% 100%

11%

11%

0%

9%

10%

0%

0%

8%

43%

39%

0%

9%

9%

0%

0% 100%

3%

5%

11%

9%

25%

11%

10%

8%

14%

19%

24%

10%

24%

5%

5% 100%

8%

4%

0%

17%

0%

0%

20%

6%

44%

19%

0%

25%

0%

0%

13% 100%

3%

6%

0%

9%

5%

0%

20%

5%

23%

38%

0%

15%

8%

0%

15% 100%

4%

1%

16%

0%

0%

0%

0%

4%

33%

8%

58%

0%

0%

0%

0% 100%

0% 0% 4% 0% 0% 3% 0% 0% 0% 100% 100% 100%

33% 25% 0% 33%

33% 38% 0% 29%

0% 0% 0% 16%

17% 25% 100% 8%

17% 13% 0% 7%

0% 0% 0% 3%

0% 0% 0% 4%

4% 5% 0% 9% 10% 2% 4% 0% 9% 5% 0% 0% 0% 4% 0% 100% 100% 100% 100% 100%

100% 100% 100% 100%

Table 2.8 shows the relationship between the Smart Grid technology and the focus of journal papers. As shown in Table 2.8, the order of the conference papers’ contribution is changed for journals, because quality is the dominant focus here, followed closely by strategy and supply chain. Quality related papers focus mostly on efficiency (22%) and distributed generation (20%). SGD Strategies at journals have the same influence of distribution automation and efficiency (21%) and they are 42% of all published strategies. The focus of journal papers for distributed generation is most importantly related to supply chain (68%). Planning papers are focused on achieving efficiency (33%) and distribution automation (22%). Service design papers focus mainly on electric vehicles (25%), efficiency (15%), distribution automation (10%), and 50

advanced metering (10%). Journal papers show technologies reaching maturity because conference papers emphasize strategy while journals are paying more attention to Quality. Table 2.9: Row & Column Percentages for Country vs. Paper’s Purpose Strategy USA China Canada Italy India Korea UK Germany Brazil Spain Japan Iran Others Total

28% 12% 5% 6% 6% 4% 3% 3% 4% 1% 3% 3% 24% 100%

Supply Quality Process Service Planning Others Chain Mgt Dsgn Dsgn Mgt 23% 28% 31% 23% 33% 28% 14% 8% 9% 17% 16% 19% 7% 7% 3% 6% 2% 6% 5% 5% 4% 5% 2% 6% 4% 4% 4% 3% 2% 4% 4% 6% 4% 2% 2% 0% 5% 4% 5% 3% 4% 0% 4% 3% 1% 6% 8% 4% 4% 0% 6% 0% 4% 4% 4% 4% 1% 3% 0% 2% 1% 5% 0% 3% 0% 6% 1% 2% 3% 2% 2% 2% 24% 24% 29% 27% 25% 21% 100% 100% 100% 100% 100% 100%

Total Strategy 27% 13% 6% 5% 4% 4% 4% 3% 3% 2% 2% 2% 25% 100%

33% 29% 28% 36% 41% 34% 24% 24% 35% 13% 33% 40% 31% 31%

Supply Quality Process Service Planning Others Total Chain Mgt Dsgn Dsgn Mgt 25% 14% 9% 6% 7% 6% 100% 33% 8% 6% 9% 7% 8% 100% 37% 17% 4% 7% 2% 6% 100% 30% 14% 6% 6% 2% 6% 100% 27% 12% 7% 5% 2% 5% 100% 29% 23% 9% 3% 3% 0% 100% 38% 15% 12% 6% 6% 0% 100% 30% 12% 3% 12% 12% 6% 100% 35% 0% 16% 0% 6% 6% 100% 50% 21% 4% 8% 0% 4% 100% 17% 29% 0% 8% 0% 13% 100% 20% 15% 10% 5% 5% 5% 100% 29% 13% 10% 7% 5% 5% 100% 29% 14% 8% 7% 5% 5% 100%

Table 2.9 shows the relationship between the focus of the papers and the country where the document was generated. As the US is the leader in strategy, quality, and supply chain, we support hypothesis H2f, stating that the US is the leader in strategic papers. As shown in Table 2.9, strategy is the most published paper focus with 28% followed closely by supply chain (28%), and then quality (23%). Although the US is the leader in all categories, it is important to emphasize the contribution of China in planning (16%), service (17%), and quality (14%), while Canada has a balanced contribution on most topics. Table 2.10 shows the relationship between the SG technologies and the countries publishing articles. The leadership of the US in the Cybersecurity (44%), Demand Response (37%), and the top 3 (25%) shows the interest on developing these technologies while the rest of the world is following the example. Germany is writing more papers on efficiency (33%) than the rest of the countries, which shows an interest in migrating to efficient ways of generating energy— as they have a plan to go green without nuclear power. The UK is an important 51

contributor for electric vehicles, with 9% of all papers on this subject. In general, we see that the US is leading over all SGD technologies with China as a distant second. Based on this analysis we can conclude that the US is and will be the leader of all clean energy efforts that come to place in the next 20 years. Table 2.10: Row & Column Percentages for Country vs. SG Technology USA China Canada Italy India Korea UK Germany Brazil Spain Japan Iran Others Total

DG

DA

Eff.

DN

EV

CS

25% 11% 7% 7% 3% 5% 1% 1% 3% 4% 4% 2% 27% 100%

25% 14% 4% 6% 6% 3% 5% 1% 5% 2% 2% 2% 24% 100%

25% 16% 5% 3% 5% 5% 2% 7% 1% 2% 3% 1% 26% 100%

15% 13% 8% 8% 5% 1% 5% 0% 9% 4% 0% 0% 31% 100%

20% 19% 6% 4% 3% 3% 9% 6% 3% 0% 3% 3% 23% 100%

44% 15% 5% 2% 3% 8% 3% 6% 0% 2% 0% 2% 11% 100%

2.4.4

AMI

DR Others Total DG DA Eff. DN EV CS AMI DR Others Total

24% 37% 36% 11% 9% 7% 5% 6% 6% 2% 4% 6% 5% 4% 4% 0% 2% 3% 4% 4% 4% 5% 4% 4% 5% 0% 2% 9% 0% 0% 0% 4% 3% 0% 6% 5% 29% 22% 22% 100% 100% 100%

27% 13% 6% 5% 4% 4% 4% 3% 3% 2% 2% 2% 25% 100%

21% 20% 28% 30% 17% 29% 9% 9% 23% 33% 38% 20% 25% 23%

17% 20% 13% 22% 24% 17% 24% 6% 26% 17% 13% 20% 18% 18%

15% 20% 13% 10% 17% 20% 9% 33% 6% 13% 21% 5% 17% 16%

4% 8% 11% 12% 10% 3% 12% 0% 23% 13% 0% 0% 10% 8%

5% 11% 7% 6% 5% 6% 18% 12% 6% 0% 8% 10% 7% 7%

Further Analysis and Prognosis

Figure 2.24: Conference papers by continent and year 52

10% 7% 6% 2% 5% 14% 6% 12% 0% 4% 0% 5% 3% 6%

5% 5% 6% 2% 7% 0% 6% 9% 10% 21% 0% 0% 7% 6%

8% 4% 6% 4% 5% 3% 6% 6% 0% 0% 8% 15% 5% 6%

14% 6% 11% 12% 10% 9% 12% 12% 6% 0% 13% 25% 10% 11%

100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100%

To understand the impact of neighboring countries as blocks, we are consolidating the number of publications into continents, in order to assess the influence and spread of these technologies across borders. We are using the country of the first author only, classified into the main 5 continents: Africa, America, Asia, Europe and Oceania.

Figure 2.25: Journal papers by continent and year Both Figures 2.24 and 2.25 show the analyzed papers grouped by publishing outlet and the continent of the paper’s first writer. America is the leader in conference papers that present strategies and technologies earlier, but Europe has been very close to taking leadership in conference papers’ production. With these facts we cannot support hypothesis H2g, which states that America is the leader in conference papers if we compare continents instead of countries. It is important to note that considering continents, America is represented mostly by the US, Canada and an emerging presence of Brazil, while Europe includes Germany, France, UK and other countries. In the journal papers categories, which present more solid and proven proposals, Asia has taken the leadership role. Europe was the leader of journal publications until 2012 when Asia 53

took the leadership position. Therefore we cannot support hypothesis H2h because the US is not a leader in this area. We expect journals to be influenced by the number of conference papers. We thus ran a regression study for the Chicco’s categories vs. 5 continents. We discovered a linear regression with an R2 of 0.927 which is deemed adequate for this regression that is significant to 0.9%, supporting our hypothesis H2i. Table 2.11: Regression statistics for relationship of Journal and Conferences’ papers

Figure 2.26: Conference papers’ topics by continent and Chicco’s categories Figures 2.26 and 2.27 show the distribution of Chicco’s categories on the different continents separated by conference or journal papers. Journal publications are led by Europe in all categories while Asia is passing America in the Physical interconnection category. 54

Figure 2.27: Journal papers’ topics by continent and Chicco’s categories Figure 2.28 shows that physical interconnections with ICT aid under the category PI, have the highest contribution (21%), followed closely by the same category PI, which also includes financial analysis PFI (16%). In third place is the financial ICT impact FI, with a contribution of 7%. The SF column is completely blank, which means that the combinations of social involvement and financial results have not been used in conjunction, except when including ICT to the equation. In other words, ICT allows consumers to reduce costs, but without ICT it is not feasible to reduce some of these expenses. Adding ICT to social and financial columns, the results show the third highest column. Unexpectedly, the weakest line is the regulatory environmental, which was expected to be an important one due to the needs of renewable energy. The new clean technology should come along with political and their upcoming regulations to ensure sustainable energy development for all society (Klemeš et al., 2012). It is awakening for us the importance of ICT in general SGD instead of regulatory efforts, as this shows that the Clean Tech Energy Revolution is heavily impacted by ICT, more than any other effort to move towards a clean tech nation (Pernick & Wilder, 2007, 2012). 55

Figure 2.28: Category mixes versus time Table 2.12: Category mixes related to SGD techs DG DA Eff DN EV CS AMI DR SH Others Total PI 31 43 31 29 2 18 17 6 18 12 207 PFI 31 30 20 22 2 8 11 9 11 13 157 FI 6 19 18 2 7 4 8 4 1 69 PEI 36 8 4 5 3 1 1 4 1 4 67 PEFI 35 3 3 3 1 2 2 5 2 11 67 I 5 20 9 7 11 2 1 2 57 PESFI 15 9 14 2 2 3 45 PSFI 1 2 5 1 14 3 1 8 1 36 PRI 5 9 5 3 3 4 1 1 31 EFI 13 5 2 2 1 1 3 27

Figure 2.28 shows how most of the highest literature contributors are still growing with the exceptions of the EFI (environmental-financial-ICT) category which shows a small decrease in 2012. The financial impact of ICT (FI) is growing. Papers show a move from physical interconnection analysis to pure ICT impact on the energy distribution process. The combination PFI (physical-financial-ICT) has migrated to either PI or FI, which is a sign of specialization in the papers, as they address either the interconnection physical aspect or the financial one.

56

2.4.4.1

ICT Related Papers

Figure 2.29 shows an upward trend in papers related to ICT, even though Canada and Germany show decreases beyond 2010’s performance. US leadership is holding strong compared to the rest of the countries in the world. The United Kingdom and China are showing growth, although there were adverse conditions in the year 2012.

Figure 2.29: ICT Papers by year & country

Figure 2.30: ICT Papers by SG Technology and Country There are 53 countries writing papers about ICT, and the top seven contribute 69% of all literature (shown in Figure 2.29) The United States is undoubtedly the largest contributor with 28% of all papers. In second place is China with a remote contribution of 9%, but the gap might 57

be shrinking as we see more papers and conferences held in China regarding the use of ICT in Smart Grid distribution. Figure 2.30 shows how DG & AMR are spreading throughout the world. Because there is strong leadership from the US in regards to Volt/ VAR and demand response (DR), we can infer that as these two areas are technical in nature, the world is following the US lead in these fields.

2.4.4.2

Physical Infrastructure Related Papers

Figure 2.31 shows a continuous increase chart on physical interconnection related papers, with the US being the leader in growth after the drop in 2011. Canada, Italy and UK show small growths in physical connection papers while there is also important growth in Brazil and India.

Figure 2.31: Physical Infrastructure Papers by Country and Year Although there are 61 countries contributing to this subject, the selected seven countries contribute 60% of the total papers. The United States is the strong leader (24% of all papers), and China is in second place (12%). The US is publishing twice as much as the next highest publishing country. The US is writing about strategies for new developments, while China is working on improving those developments.

58

Figure 2.32 shows how the physical interconnection of distributed generation has more literature around the world than in the US. This is an area of opportunity for America because there is a strong European push towards this technology that has not been developed with the same emphasis here. Distribution automation and efficiency are important topics that have a strong push from the US, focusing on state of the art and ICT technologies to optimize processes and transferring decision making from people to automatic controls.

Figure 2.32: Physical Infrastructure Papers by SG Technology and Country The importance of DA and DG into the physical infrastructure category shows that they are components of a new SGD supply chain, bringing reductions to the fluctuation of energy while reducing costs (Daim et al. 2012).

2.4.4.3

Economics Related Papers

Figure 2.33 shows a deeper fall compared to the other layers. Economics, however, have to change to adapt to the real world and transform from being a part of the problem to an element of the solution (Spangenberg, 2010). Germany is showing important decreases in economics related papers. The US is the leader in this category with China a distant second. The shown 59

seven countries in the chart represent 65% of all literature, although there are 57 countries writing about this topic. The United States has a contribution of 29% of all papers. Second place is China with a remote contribution of 13%, that is, the US is publishing more than twice as much as the next largest contributing country. It is important to emphasize that green initiatives have high initial capital requirements that have to be recovered soon (Chang et al. 2011).

Figure 2.33: Economics Related Papers by Country and Year

Figure 2.34: Economics related papers by SG Technology and Country Figure 2.34 shows that the main interest for economic analysis in the distributed generation area, as a new business model will come out of this technology implementation. Second in importance is the topic of efficiency, followed by distribution automation. China is 60

looking at DA before DG because of the national interest on automation. In China, home generation is not a priority under actual conditions. The cost of energy will soon become a major topic of discussion because DG is closely related to financing, so the implementation of selfgeneration will require contracts between stake holders that were not involved before. The market suggests the system will soon move into a growth phase of technology diffusion towards a new economic model (Köhler et al. 2012).

2.4.4.4

Environmental Related Papers

As shown in Figure 2.35, Canada, Italy and Germany are showing important decreases on environmental related papers while the US’s leadership is stabilizing and China is quickly growing. Although there are 50 countries that have written papers about environmental issues, the selected seven countries contribute 61% of the papers on this topic. The United States is stabilizing as the leader with a contribution of 24% of all papers. In second place is China with a remote contribution of 13%.

Figure 2.35: Environmental Related Papers by Country and Year According to the UN millennium development goal statistics in 2009 China was the largest polluter with a contribution of 7,687,114 thousand metric tons of CO2 emission in the 61

environment. These pollution levels arise because most of their energy generation depends on coal, and 80% of CO2 emission comes from coal (Feng & Yuexia, 2011.) The US, with 5,299,563 tons of CO2 emission is publishing more than three times as many papers as China. As expected, distributed generation is the technology most related to environmental protection, as it pertains to renewable energy and home generation to reduce and prevent contamination.

Figure 2.36: Environmental Related Papers by SG Technology and Country

2.4.5

Hypotheses H2 Results

Table 2.13 shows the results of all the hypotheses presented in section 2.5.1. From the results it is evident that the US is the leader in this technological breakthrough and that, after being discussed at conferences, SG is growing in importance in journal papers. There aren’t many data secondary sources, so case and field studies are primary sources for data analysis. As blocks, the strength of the European Union and Asia is evident. They are trending up and have already passed America (US and Canada) in literature generation at both journal and conferences in the past year.

62

Table 2.13: Hypotheses H2 results

2.5 DATA ORIENTED ANALYSIS 2.5.1

Word Mining

Continuing our analysis on the 966 papers listed, we conducted a word mining process. Of the 966 papers, only 959 of the papers had abstracts available. Most papers have an abstract available within the ISI database, and these abstracts contain the key words and actions of the researchers. Under these premises, we conducted a word mining analysis on the 959 abstracts to determine what the direction and key areas of research are for this particular technology. 63

The first step was to identify the keywords that are important while being together, for instance, distribution automation. Distribution automation is an element of Smart Grids specifically focusing on automating controls of the distribution process, so we placed a dash to unite both words when together (distribution-automation). Table 2.14: Top 25 words on number of mentions in the past 5 years Conference Papers Journal Papers Words 2008 2009 2010 2011 2012 2013 Total 2009 2010 2011 2012 2013 Total system 36 71 237 218 447 291 1,300 3 52 116 156 244 571 smart-grid 14 50 162 216 357 223 1,022 2 50 96 147 200 495 power 19 45 157 176 346 156 899 13 37 118 119 165 452 distribution 28 48 131 193 337 194 931 2 45 91 106 168 412 paper 13 32 104 128 262 156 695 4 33 56 84 125 302 energy 9 20 97 147 249 136 658 8 31 69 80 132 320 network 6 12 70 118 183 150 539 5 22 46 74 132 279 control 15 34 91 105 144 115 504 6 23 50 74 121 274 grid 7 26 49 87 165 129 463 5 15 39 63 96 218 communication 4 7 56 73 162 100 402 1 11 41 55 103 211 technology 6 24 65 85 178 75 433 2 13 27 38 49 129 data 4 22 45 63 141 71 346 1 12 18 42 97 170 proposed 12 8 31 51 103 71 276 2 8 39 60 99 208 management 10 12 43 47 116 52 280 4 8 35 42 62 151 based 3 5 37 56 102 84 287 11 28 27 77 143 Results 4 11 23 47 101 67 253 10 31 40 72 153 load 7 8 38 55 73 65 246 14 36 41 65 156 voltage 11 18 46 51 86 57 269 1 20 26 46 34 127 new 13 32 33 46 101 50 275 2 11 21 27 38 99 generation 10 23 62 97 43 235 1 10 25 47 52 135 Model 3 10 41 45 83 48 230 1 18 18 34 43 114 electricity 11 12 30 40 73 49 215 4 14 10 40 49 117 distribution-network 2 2 33 42 87 70 236 9 16 26 42 93 power-system 4 24 27 48 80 47 230 1 14 26 18 23 82 transmission 13 17 34 31 82 31 208 1 9 19 22 48 99

Grand Total 1,871 1,517 1,351 1,343 997 978 818 778 681 613 562 516 484 431 430 406 402 396 374 370 344 332 329 312 307

The key concepts that were marked this way are: advanced-metering, cyber-security, demand-response, distributed-generation, and distributed-storage, distribution-automation, distribution-networks, electric-vehicles, energy-efficiency, information-technology, micro-grid, self-healing, smart-grid and vehicle-to-grid. In the second step, some of the words with the same 64

meaning were merged to identify the key words in the literature. The final step was to cluster key words to predict direction on this area. Table 2.14 shows the top words and the number of mentions in general; this list does not include articles, prepositions, auxiliaries and conjunctions, so that we can focus only on words related to the SGD topic. Table 2.15: Top 20 SGD Technologies mentions in the past 6 years Words 2008 2009 smart-grid 14 50 distributed-generation 8 8 distribution-network 2 2 power-system 4 24 electric-vehicle 5 real-time 14 4 demand-response 5 5 power-grid 2 10 smart-meter 7 5 distributed-storage 1 power-quality 2 distribution-automation 6 self-healing 1 vehicle-to-grid 1 information-technology 6 multi-agent 6 micro-grid two-way demand-side cyber-security

Conference Papers Journal Papers 2010 2011 2012 2013 Total 2009 2010 2011 2012 2013 175 221 372 235 1,067 2 50 116 147 201 11 62 99 62 250 3 27 31 42 42 33 42 89 70 238 9 16 26 42 27 48 80 47 230 1 14 26 18 23 11 44 66 50 176 5 23 47 15 17 28 50 30 143 11 12 21 27 19 24 58 39 150 17 2 10 27 17 25 51 24 129 12 13 18 33 29 30 41 28 140 4 2 5 15 23 8 18 37 13 77 6 2 17 13 10 6 25 15 58 3 11 7 5 9 1 31 19 66 1 3 4 3 6 25 8 43 1 1 7 7 13 13 27 1 6 17 1 12 9 3 6 36 1 4 4 3 8 9 12 4 39 7 1 3 5 3 10 21 9 3 3 4 4 11 2 21 1 2 4 6 2 2 8 1 13 4 2 7 1 10 4 2 17 1 2 2

Grand Total Total 516 1,583 145 395 93 331 82 312 90 266 71 214 56 206 76 205 49 189 38 115 26 84 8 74 16 59 25 52 12 48 8 47 15 36 13 34 13 26 5 22

The total number of words contained in the abstracts was 154,706. The connected words that we mentioned in the prior section were counted as well as the times their initials were mentioned. The results are shown in Table 2.15. As the terms “Smart Grid” and “Distribution” were used for filtering out results, it is not surprising that these terms were high on number of mentions. From the technologies under the umbrella of Smart Grid, distributed generation was the one with most mentions followed by distribution networks and electric vehicles. Focusing on the SGD technologies and the TLAs related to them, we added the mentions of the complete 65

word and the TLAs, and then separated these technologies to determine if they show a trend in regards to mentions in the selected literature. Results are shown in Table 2.15, where the continuous growth of some of the technological nomenclatures is evident. To identify the trend of these technologies, we proceeded to mine the words and cluster them by publishing year of the paper. The table uses yellow coloring for base line (first year) and to identify when the number does not change more or less than 10% over time; green shows an increase over the prior year’s performance and red represents a decrease in the number of word usage over 10%. Table 2.15 confirms that there is not enough conference papers listed in ISI because most words are showing reductions compared to 2012. Looking at conferences up until 2012 and journal papers up until 2013, we make the following observations: Smart grids, distributed network, distribution generation, distribution network, power systems, demand response, smart meters, distribution automation, and self-healing grew or were stable in the past year, reinforcing the point that these technologies are considered as SGD tools. On the contrary, electric vehicles, distributed storage, power quality and cybersecurity decreased in mentions in the last year. These results may well infer that the increase of electric vehicles’ mentions comes coupled with the growth of vehicle-to-grid mentions, showing that the concept of the vehicle is evolving from a simple transportation mode to a new generation of energy source (Brown et al. 2010). Micro-grids mentions are reduced but distributed generation mentions are relatively stable, which might infer that the concept of the microgrid is diluted under that name into DG, DN and DA. Information-technology words were also reduced, but distribution automation increased by more than the decrease. Therefore, the overall concept has not faded, but moved from ICT to automatic controls and demand responses.

66

The term energy-efficiency, which is also a part of the modern definition of a smart grid, was considered as a stand-alone concept for a short time and it is now fading. The last concept that was reduced from 2011 was Cybersecurity, which entered into SGD and quickly faded because the need of security measures has always been an integral part of the smart grid concept. 2.5.2

Bass’ Diffusion Model

As noted by Peres, Muller and Mahajan (2010), traditionally, the main thread of diffusion models has been based on the framework developed by Bass. Thus we adopted the Bass model in the present research. We begin this section with the representation of the Bass model for a single series. We adopt the definition of diffusion offered by Peres et al. (2010). They define diffusion as “the process of the market penetration of new products and services, which is driven by social influences. Such influences include all of the interdependencies among consumers that affect various market players with or without their explicit knowledge.” Using the extension of the basic diffusion model (Mahajan et al., 1990) a non-linear solution is implemented. Boswijk and Franses (2005) borrow from financial econometrics a modified stochastic error process for the Bass model. Their approach is designed to capture heteroscedasticity errors and a tendency for the data to revert to the long-term trend. Additionally, this model includes an additional regressor, which leads us to rely on simulated maximum likelihood to estimate the parameters. The first developed Bass model states that the probability that an individual will adopt the innovation — given that the individual has not yet adopted it—is linear with respect to the number of previous adopters. The model parameters p, q, and m can be estimated from the actual adoption data. Here p is the innovation parameter, q is the imitation parameter, and m is the maturity or saturation level. Following the representation as advocated in Boswijk and Franses

67

(2005), we discuss the multi-level model for a panel of diffusion time series of smart meters. Finally, we report the parameter estimation of this last model. The Bass model assumes a population of m potential adopters, where, in the context of citations, we will associate m with the maturity level or in this case the maximum amount of written, published and listed papers. In our context, adopters should be viewed as scholars that write a paper related to Smart Grids that is peer-reviewed and later listed in Reuter’s ISI Web of Science. Maturity can be viewed as the saturation level when the total number of papers has been exhausted. For each adopter, the time to adoption is a random variable with a distribution function F(t) and density f(t), such that the hazard rate equals f(t) = p + q F(t) 1 - F(t)

(1)

where p and q are the parameters that determine the shape of the diffusion process. The cumulative number of adopters at time t, denoted by N(t), is a random variable with mean _ N(t) = E [ N(t) ] = m F(t) (2) where t is measured in continuous time and E denotes the expectation operator. It can be shown that the function n(t) obeys the following differential equation: _ _ _ dN(t) ____ q N(t) [ m – N(t)] n(t) = dt = p[ m – N(t)] + __ (3) m According to Boswijk and Franses (2005), the solution of this differential equation is given by: __

N(t) = m F(t) = m

1 ‐  e 

–(p+q)t

1+(q/p)e

(4)

‐(p+q)t

Note that the parameters p, q and m exercise a non-linear impact on the pattern of N(t) and n(t). Basic characteristics of the diffusion are also non-linearly dependent on p and q. These parameters provide us with speed of diffusion data. According to Van den Bulte (2004) a high p value shows that the diffusion has a quick start but it will go down quickly while a high q value shows that the diffusion is slow at first but will accelerate afterwards. We are going to calculate p, q and m, but our main index is going to be the diffusion speed defined as the ratio of q/p. It is 68

important to notice that if p is almost zero, the ratio will tend to infinite, so we are going to use a natural log of q/p. Table 2.16: Bass’ Diffusion Model results for SGD technologies Category Cum Level Maturity Innovation Imitation Smart Grid 1,583 1,965 0.0070 1.07 Distributed Generation 395 467 0.0045 1.21 Distribution Network 331 435 0.0038 1.14 Power System 313 378 0.0117 0.99 Electric Vehicle 266 287 0.0009 1.67 Real-time 214 307 0.0124 0.84 Demand Response 206 337 0.0097 0.81 Smart Meters 189 271 0.0152 0.79 Distributed Storage 115 126 0.0011 1.61 Distribution Automation 74 115 0.0053 0.97 Self Healing 59 62 0.0001 2.15

Analyzing the available five year trend, we used the Bass’ Diffusion Model to determine if the technology is about to reach maturity or if it is in the early stages. We expect the word count to follow the “S” shaped curve for non-linear regression. The curve is broken into 5 sections: Innovators (5%), early adopters (11%), early majority (34%), late majority (34%) and late adopters (16%). The curve in formula (1) contains 3 variables that are important: p, q and m. The results are shown in Table 2.16, where microgrids and power quality seem to be words that will not be used much more. Smart grid, advanced metering and distributed generation still have some time to go, while real-time is the concept that is exponentially growing and will probably mature in a longer period. Using Wordstat, specialized software for word mining, we introduced all the available abstracts. The resulting Figure 2.37 shows an illustration of the number of mentions, proximities and relationships. By only looking at Figure 2.37, the filtering words: smart, grid and distribution are evident with the number of mentions, as was expected. The other words that are noticeable are system and power, which reinforce the 69

development of a new way of distributing power with the use of systems (ICT). The next concepts are energy, electric, generation, control and network, which relate to the power sector and the use of ICT.

Figure 2.37: Text Mining Cloud

2.5.3

Author-oriented analysis

We used the identified 966 papers written by 2,632 authors (average of 2.72 authors per paper) from 81 different countries (average of 11.9 papers per country). The origin country was selected based on the author’s country. That is, if the author got his/her bachelor degree in China and is currently in the US working on a PhD, the country of origin is still China. When biographical information was not available for authors, the country referred to in the general information was assigned to the authors. Table 2.17 presents the distribution of the top 20 authors’ countries. The United States (US) has the highest number of authors publishing peerreviewed papers with 24% (almost a quarter of all authors, with 808 papers). Authors from China ranked second with 13% (with 435 papers). Italy occupied third place with 6% (199 papers). Brazil is a close fourth place with 5% (with 174 papers). Almost half of all authors (48%) are from these first mentioned countries. The number of authors per paper is presented in Table 2.18. The average number of authors per paper is 2.72 (2,632/966). Ten percent of the papers were 70

written by a single author; 21% of the papers were prepared by two authors, and 26% were presented by three authors; and 21% were written in collaboration of four authors. Table 2.17: Distribution of top 20 countries of origin of SGD papers’ authors Country 2008 2009 2010 2011 2012 2013 Total USA 15 50 89 122 245 287 808 China 6 3 36 63 195 132 435 Italy 1 22 37 58 81 199 Brazil 6 8 33 127 174 Canada 3 12 24 43 77 159 Korea 15 1 28 37 61 142 UK 1 1 10 13 49 59 133 Germany 2 3 18 25 61 7 116 India 2 1 5 17 38 51 114 Japan 3 11 27 47 23 111 Spain 4 8 6 36 40 94 Portugal 8 40 31 79 Australia 15 9 17 23 64 Netherlands 3 11 14 25 53 Iran 1 3 12 6 14 15 51 Belgium 5 14 25 44 Finland 4 4 16 19 43 Malaysia 1 20 21 42 France 1 3 1 3 13 19 40 Denmark 3 3 4 8 22 40

Table 2.18: Distribution of papers by number of authors # of authors Quantity % Single author 96 10% Two authors 197 20% Three authors 254 26% Four authors 204 21% Five authors 110 11% Six authors 48 5% Seven authors 32 3% Eight authors 9 1% Nine authors 8 1% Ten authors 5 1% Eleven authors 1 0% Twelve authors 2 0% 71

Figure 2.38: Distribution of papers by number of authors Table 2.19: Comparison of Journal and Conference papers by country of authors

USA China Italy Brazil Canada Korea UK Germany India Japan Spain Portugal Australia Netherlands Iran Belgium Finland Malaysia France Denmark

2008 15 6

3 1 2 2 4

1

1

Conference Papers 2009 2010 2011 2012 2013 49 82 92 141 209 3 13 47 164 78 13 5 52 36 6 1 33 106 8 21 30 44 13 1 3 31 24 1 6 10 36 29 14 20 56 7 1 4 14 32 42 2 11 21 32 18 4 1 29 17 8 20 27 12 9 13 11 3 6 8 18 3 10 6 10 15 1 11 24 13 19 1 16 15 3 1 9 11 3 3 4 8 18

Total 2009 588 1 311 106 1 146 106 72 2 83 99 3 95 84 1 55 55 45 35 45 36 32 32 25 36

Journal Papers Grand 2010 2011 2012 2013 Total Total 7 30 104 78 220 808 23 16 31 54 124 435 9 32 6 45 93 199 7 21 28 174 4 3 13 33 53 159 25 6 37 70 142 4 3 13 30 50 133 4 5 5 17 116 1 3 6 9 19 114 6 15 5 27 111 4 5 7 23 39 94 20 4 24 79 3 4 12 19 64 5 6 7 18 53 2 4 6 51 4 3 1 8 44 4 4 3 11 43 4 6 10 42 3 4 8 15 40 4 4 40

Analyzing the country of origin and how many first authors a country has, we see that the US has 259 first authors. China is a distant second with 124, but the majority of papers there 72

have at least 3 authors, while in the US, 56 papers have a single author. There are just a few lone rangers in other countries, as the number of papers with one and two authors are almost equal for most cases. The next part of our analysis is to identify the most productive writers, shown in Table 2.21, based on the previously mentioned research criteria. Morais and Vale, from Portugal, are the two most prolific authors although they have written only in the past three years. Their focus is on electric vehicles and the burden they will bring to the electric grid when it is time to charge them. Schneider and Fuller from the US have focused their research on distribution automation, but they have not written any other papers in the past year. Table 2.20: Distribution of authors by Country and writing order Country 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th 11th 12th Total USA 259 229 155 89 43 16 12 4 1 808 China 124 104 89 66 25 12 6 4 4 1 435 Italy 50 41 37 28 17 8 5 4 3 2 2 2 199 Brazil 31 31 30 25 20 17 7 6 4 3 174 Canada 54 53 35 12 3 1 1 159 Korea 35 29 26 19 14 10 6 2 1 142 UK 34 34 28 17 9 6 3 1 1 133 Germany 33 32 26 14 8 2 1 116 India 41 34 28 7 3 1 114 Japan 24 25 19 17 14 7 2 1 1 1 111 Spain 24 22 21 15 9 3 94 Portugal 19 19 18 15 6 1 1 79 Australia 15 20 12 5 5 4 2 1 64 Netherlands 17 13 12 9 1 1 53 Iran 20 15 10 2 2 1 1 51 Belgium 12 9 7 5 4 4 3 44 Finland 10 11 10 8 4 43 Malaysia 9 10 6 8 5 2 2 42 France 9 12 10 7 2 40 Denmark 8 10 10 5 2 1 1 1 1 1 40

73

Table 2.21: Distribution of top 20 most prolific writers of SGD papers Author 2008 2009 2010 2011 2012 2013 Total Morais, H 2 10 5 17 Vale, Z 2 10 5 17 Heydt, GT 4 1 1 2 8 Schneider, KP 2 2 3 7 Senjyu, T 2 2 3 7 Mohagheghi, 1 1 1 1 1 2 7 Fuller, JC 1 2 4 7 Gungor, VC 1 2 2 2 7 Yang, F 1 2 1 2 6 Sousa, T 1 4 1 6 Pilo, F 1 2 3 6 El-Saadany, EF 3 2 1 6 Jauch, ET 3 1 1 1 6 Faria, P 1 2 3 6 Chow, MY 1 1 1 3 6 Wang, P 1 1 3 5 Chassin, D 2 3 5 Yona, A 2 2 1 5 Soares, J 1 4 5 Suryanarayanan, S 1 2 2 5

Table 2.22: Type of publication of most prolific authors (co-authors) Conference Papers Journal Papers Grand Author 2008 2009 2010 2011 2012 2013 Total 2009 2010 2011 2012 2013 Total Total Morais, H 2 5 4 11 5 1 6 17 Vale, Z 2 5 4 11 5 1 6 17 Heydt, GT 4 1 5 1 1 1 3 8 Schneider, KP 2 2 2 6 1 1 7 Senjyu, T 2 1 3 6 1 1 7 Mohagheghi, 1 1 1 1 1 1 6 1 1 7 Fuller, JC 1 2 3 6 1 1 7 Gungor, VC 1 1 1 1 2 2 6 7 Yang, F 1 1 1 1 4 1 1 2 6 Sousa, T 1 2 1 4 2 2 6 Pilo, F 2 1 3 1 2 3 6 El-Saadany, EF 3 1 4 1 1 2 6 Jauch, ET 3 1 1 5 1 1 6 Faria, P 1 1 2 4 1 1 2 6 Chow, MY 1 1 1 3 1 2 3 6 Wang, P 1 1 1 3 2 2 5 Chassin, D 2 2 4 1 1 5 Yona, A 2 1 1 4 1 1 5 Soares, J 1 1 2 3 3 5 Suryanarayanan, S 1 1 2 4 1 1 5

74

Analyzing the contribution in journals, we can see the impact of Güngör from Turkey, who started writing in 2010 and focused on wireless communications; more importantly, his contribution has been consistent and mostly in journal papers. 2.6 GAP ANALYSIS OF LITERATURE AND INVESTMENTS 2.6.1

Background Information

Information and Communication Technology (ICT) has been implemented globally, and has facilitated many processes that in the past could not be accomplished. Along with the development of new ideas, however, inevitably come the weaknesses surrounding them. The old energy scheme of large energy generating sites, transporting electricity through HV wires to substations, is being challenged with the new strategies of distributed generation—that provide a localized energy generation for small communities or even homes. In this section of the dissertation we will compare three different communities: 1. Government agencies that are distributing the monies assigned for the American Recovery and Reinvestment Act (ARRA) for the Department of Energy (DoE) more than 28 billion dollars were used for several projects under this government department. 2. Private sector research represented by the Electric Power Research Institute (EPRI) that although it is partially funded with DoE monies, represents the interests and focus of practitioner’s research. 3. The Academic sector, represented by peer-reviewed articles published and listed by ISI Web of Science.

75

2.6.2

Hypotheses H3

Because ARRA is a governmental program to aid communities, it is our expectation that the largest amount will be invested on measures that will result in an improved image favorable to the government. Based upon this observation, we expect the spread of funds to be a populist measure that, by reaching the consumers, will probably have an impact on the majority’s perception of the government. H3a: ARRA largest share on energy is devoted to homes’ efficiency to be a populist factor. Practitioners and researchers are curious by nature, and are accustomed to confronting failures and continually pursuing success in their areas. Due to recent concerns about nuclear energy, some countries are closing their nuclear sites, despite the fact that researchers are still looking for alternative ways to prevent the presented issues and develop further safer nuclear energy generation processes. It is our expectation that EPRI will still be investing heavily in nuclear energy, as it was their focus since day one. Thus, our next hypothesis is: H3b: Research of nuclear energy is still supported by EPRI Practitioners and academics are not always in the same boat, so we expect that there may be different targets for these groups, but the governmental approach might be closer to the academic world, as they represent the consumers who use the energy. It is our expectation that academically generated literature and governmental fund distribution may be more similar than the practitioner’s research focus. H3c: Academic research and DoE have similar focuses on the energy area 2.6.3

US Government

The US government is partially responsible for the well-being of citizens and providing electricity, a commodity that is critical for all people. The government’s role on the provision of 76

this utility becomes very important. For many years, the government has spent tax monies on building a gigantic electricity infrastructure that has been used consistently with focus on: large generation plants, long high voltage transmission towers and wires and multiple substations where electricity is brought to lower voltages for distribution. This electric infrastructure has been considered the largest machine in the world, as it resembles a single machine with a size that goes from side to side on the continental US. With the evolution of technology, the overall scheme is going to change dramatically, such that all the world’s governments are thinking about their investment in energy for the short term. In the case of the US, the government has been looking for initiatives on energy investment that will aid the future direction. President Obama has promoted energy from renewable sources to reduce the harm to the planet and, thus, the government has supported certain initiatives. On April 13, 2010, the White House presented these initiatives: 

Invest $3.4 billion on a Smart Grid Investment Grant (part of ARRA).



Launch the Advanced Research Projects Agency-Energy (ARPA-E)



Question how the federal government can help to create a "self-sustaining home energy efficiency"



Create new efficiency standards for home appliances,



Develop a new National Fuel Efficiency Policy for cars model years 2012 through 2016



Create measures to increase the production of biofuels



Develop an interagency task force for a federal strategy of carbon capture and storage,



And produce a new Environmental Protection Agency ruling (called the Mandatory Reporting of Greenhouse Gases Rule) requiring the reporting of greenhouse gas emissions by major emitters in the United States. 77

In 2009, the President signed the American Recovery and Reinvestment Act (ARRA), with an approximate cost of $787 billion dollars (which was later revised to 831 billion dollars). This amount was assigned to different categories. $41.96 billion dollars were assigned to the Department of Energy (DoE) that has already awarded funds for $35.89 billion dollars and have already paid $30.57 billion dollars to incentivize the economy and support the plans toward optimized and more efficient energy (http://energy.gov/recovery-act). The Department of Energy (DoE) distributed the budgeted monies in the following way, as shown in the Figure 2.39.

Figure 2.39: DOE recovery awards The wording used to identify the categories is the one used by DoE. The category where most money is devoted is the energy efficiency portion, supporting hypothesis H3a. This is a clear sign of governmental efforts to make efficient use of the currently infrastructure for generation, transportation and distribution of electricity. The second important category is the environmental cleanup. This is a posteriori activity that supports the economy while trying to clean up or reduce the harm already done to the earth. The third category is the modernization of the grid, which for the authors of this paper is the most important part of the investment. Only 78

14% of the monies are assigned to modernization! It is surprising to see that DoE is spending almost the same amount for capturing carbon as it is for modernizing the multibillionaire infrastructure, According to the “Department of Energy: Successes of the Recovery Act,” published in January 2012, DOE invested its funds into the following pillars to ensure America’s long-term competitiveness: 

Increasing energy efficiency;



Restructuring the transportation system;



Doubling renewable generation;



Investing in smart grid infrastructure;



Expanding innovative research;



and Cleaning up our nation’s legacy nuclear waste

Energy efficiency is achieved through the weatherization of homes of more than 650,000 low-income families nationwide (through December 2011). By doing these retrofits, consumers have improved the energy efficiency of their homes, saving money on their energy bills. According to Eisenberg (2010) families have an average savings of $437 per year on their bills due to this program. CO2 has been accepted as a real potential condition with serious consequences for human society. The power and energy industry emit 40% of the global CO2 production, while transportation follows in importance with 24% (Venayagamoorthy, 2009).Therefore, there is an increasing interest in the development and production of electric vehicles. Thanks to this effort, before the ARRA there were only 500 electric vehicles charging stations and the expectation was for 18,000 by the end of 2012. 79

Doubling the renewable generation of energy has been one of the targets for DoE while investing the ARRA funds. These funds were used to finance the largest photovoltaic plant of the world in Agua Caliente, Arizona; the largest wind farm in the world to date is in the state of Oregon and two of the largest solar thermal projects are in the deserts of Arizona and California. More than $4 billion in Recovery Act Smart Grid investments have been devoted to modernizing the electrical grid and building a more stable and secure system that integrates distributed generation. There have been some forums to discuss how Smart Grid technologies can potentially transform the way that consumers interact with the grid. In order to obtain the maximum benefit from it all stakeholders need to understand the process (Gonzalez et al., 2011). For innovative research, the Advanced Research Projects Agency - Energy (ARPA-E) is funding projects that the industry is not likely to undertake because of uncertainty. These projects play a key role to ensure that the US maintains leadership in the next generation of advanced clean energy technologies. Some of the programs include: e-vehicle batteries, carbon capture, advanced fuels, grid scale storage, building efficiency, and power electronics. Carbon capture plays a substantial role in the effort to mitigate carbon emissions, because in the past 20 years almost three quarters of the emissions have come from burning fossil fuels (Saylor, 2012). There are approximately 15 projects that seek to reduce the cost of capturing carbon from coal-fired power plants. The last category is related to the Cold War legacy clean-up of nuclear weapons programs. Although much has been spent on this cleaning effort, the $6 billion investment is expected to reduce more than 40% of the total nuclear waste footprint (372 sq. mi).

80

2.6.4

Private Sector

After the great Northeastern blackout of 1965, the US congress challenged the vulnerability of the energy industry, resulting in the birth of the Electric Power research Institute (EPRI), an independent, non-profit research center. In 1972, the center was created with a representation of 90% of the electricity generated in the country. At this place, specialists can collaborate to help strengthen systems to prevent outages and enhance systems. EPRI is partially funded by DoE, but it also represents the practitioners and scientists interests in the various areas under the jurisdiction of energy. 5% of the 90 funds received by EPRI come from the government, while the rest comes from private investors and companies interested in the nuclear energy sector. EPRI represents a different approach or view to the energy sector, and has invested almost 300 million dollars in research project distribution, as shown in Figure 2.40.

Figure 2.40: EPRI funded projects It is noteworthy that almost half the funds of EPRI are dedicated to researching the Nuclear Program. This figure supports hypothesis H3b. There is a major interest in developing and designing new ways of producing nuclear energy. Some of the most important projects funded by EPRI in the nuclear program are:

81

The Power Delivery and Utilization Program come second in importance. This area represents the transmission, distribution and consumption of electricity that is becoming the center of the Smart Grid concept. Transmission has been smart for many years, but distribution has not been the focus of attention until recently, when smart grids are looking at microgrids, smart homes, advanced metering infrastructure, and others. The third topic of importance for EPRI is the generation program. This program focuses on distributed generation and renewable resources because these subjects are becoming increasingly popular due to the governments’ pressure on reducing emissions and improving efficiency to reverse the environmental damaged already done. 2.6.5

Academic Sector

To include the academic sector perspective in this study, we analyzed peer-reviewed papers that are listed in Reuter’s ISI Web of Science and that contain the Smart Grid Distribution terms. 966 articles were listed for the period from 2008 through 2013. Most of the papers listed in ISI, and related to Smart Grid Distribution (SGD), are published by IEEE, representing 69% of the total papers analyzed. These papers focus on how the different techniques will be implemented; therefore, they focus on practical applications via case studies as well as lab and field research using various statistical techniques. The majority of the papers are related to strategy, quality, and supply chain (71% of all papers). Smart Grid and ICT are closely related because the reality of SGD is accomplished with the aid of new information and communication technologies. It is not surprising, then, that 95% of all analyzed papers refer to ICT in some way. Concerns about infrastructure needs follow in importance because of the current state of the grid, which needs to be revamped or modernized.

82

Figure 2.41: Academics distribution of Literature The top category of the literature is the distribution automation, or enhancements to eliminate or reduce dependency on operators with the use of computer and information and communication systems. In second place, the focus is on system efficiency, stating that by improving process quality we can overcome the need for more generation, as distribution losses are reduced. An interesting fact is that strategic papers are being replaced by quality assurance ones because of the move from concepts and ideas to grounded efforts for improvement.

2.6.6

Methodology

Because the nomenclature used by the three different sectors is different, the first step of the process is to correlate all of them into a single category that accommodates the different terms without losing the focus of activities and projects. The second step is to develop a comparison table for the sectors, while the third step is to statistically test the distributions and determine if they are similar or dissimilar. The Department of Energy, EPRI and the literature survey use different categories, so we will focus on the EPRI nomenclature, which is more

83

general than the other two sources. It is important to define what is going to be included in each category to conduct the gap analysis. As such the following terms are defined as follows: 

Power Delivery and Utilization programs include the optimization of the current infrastructure and process.



Nuclear Program – Includes all efforts to further develop the use of nuclear energy. It does not include the clean-up efforts as they are more related to environmental protection. This category focuses strictly on new forms of use of nuclear energy.



Generation Programs - Focuses specifically on the increase of energy generation via new sites, expansion of the current ones or storage means to be used as forms of generation.



Environmental Program – Includes all efforts to prevent environment damage, as well as clean-up already contaminated areas. Wind and solar energy generation are included in this category as their main purpose is to prevent environmental harm. Correlating the categories selected for this comparison with the financial reports from

DoE, we can see the following correlation: Table 2.23: DoE distribution on selected categories Power Delivery DoE Categories Environment Generation Nuclear & Utilization Total Contribution Energy Efficiency 11 11 39.29% Environment cleanup 5 5 17.86% Modernizing grid 4 4 14.29% Carbon capture 3 3 10.71% Transportation 2 2 7.14% Innovations 2 2 7.14% Renewables 1 1 3.57% Total 10 1 0 17 28 100.00% Contribution 35.71% 3.57% 0.00% 60.71% 100.00%

84

Table 2.24: EPRI distribution on selected categories EPRI Categories Environment Generation Environment Programs 41.95 Generation Programs 51.75 Nuclear Program Power Delivery & Utilization Programs Total 41.95 51.75 Contribution 14.41% 17.77%

Nuclear

Power Delivery & Utilization

136

136 46.70%

Total Contribution 41.95 14.41% 51.75 17.77% 136 46.70%

61.5 61.5 61.5 291.2 21.12% 100.00%

21.12% 100.00%

Because EPRI nomenclature was selected for the gap analysis, there is a direct correlation among the categories, as shown in the next table. The categories of academic research listed by ISI Web of Science are correlated to the 4 categories in the Table 2.25. Table 2.25: Academic Research distribution by categories Academic Research Environment Generation Automated Meter Reading Cybersecurity Demand Response 54 Distributed Generation 219 Distribution Automation Distribution Network Efficiency Electric Vehicles 70 Energy storage 35 Self-healing User friendly Vol/VAR Total 289 89 Contribution 29.9% 9.2%

2.6.7

Nuclear

Power Delivery & Utilization 55 62

173 75 154

0 0.0%

43 3 23 588 60.9%

Total Contribution 55 5.7% 62 6.4% 54 5.6% 219 22.7% 173 17.9% 75 7.8% 154 15.9% 70 7.2% 35 3.6% 43 4.5% 3 0.3% 23 2.4% 966 100.0% 100.0%

Results

Once the categories have been correlated, we present the complete table with all of the data:

85

Table 2.26: Comparisons between DoE, EPRI and the Literature Survey

EPRI DOE Lit Survey

Power Delivery & Utilization Environment Generation Nuclear 14.41% 17.77% 46.70% 21.12% 28.57% 3.57% 0.00% 67.86% 29.92% 9.21% 0.00% 60.87%

Based on this preliminary analysis, we can see a strong correlation between DoE and the literature survey’s focuses. EPRI strongly supports nuclear energy, which is being reduced and even eliminated worldwide, most notably in Europe. Table 2.27 shows the relations between these two organizations versus the published literature. Based on the Table 2.27 comparison, the deltas for DoE vs. Lit survey are much smaller than the deltas for EPRI. To do a statistical comparison of the data, we ran Pearson correlation analyses that yielded the following results: Table 2.27: Gap from the average of DoE and EPRI versus Literature survey

Environment Programs Generation Programs Nuclear Program Power Delivery & Utilization Programs

DOE Vs Lit Survey EPRI Vs Lit Survey Average Delta Average Delta 29.24% 0.67% 22.16% 7.76% 6.39% 2.82% 13.49% 4.28% 0.00% 0.00% 23.35% 23.35% 64.36%

3.49%

40.99%

Table 2.28: Correlation results for comparison groups EPRI DoE Lit_Surv EPRI Pearson Correlation 1 0.537 0.524 Sig. (2-tailed) 0.639 0.649 N 4 3 3 DoE Pearson Correlation 0.537 1 1.000** Sig. (2-tailed) 0.639 0.01 N 3 3 3 Lit_Surv Pearson Correlation 0.524 1.000** 1 Sig. (2-tailed) 0.649 0.01 N 3 3 3 **. Correlation is significant at the 0.01 level (2-tailed).

86

19.88%

Based on the Pearson correlation analyses, we discovered that both DoE and the literature survey do not correlate to EPRI, while the literature survey and DoE correlate significantly at .01 levels, supporting hypotheses H3c.

2.6.8

Gap’s Conclusions

The focus of academic research and the funds assigned to DoE from the Recovery plan follow the same pattern. Utilization takes more than 60% of the efforts, followed by 30% from environmental activities such as distributed generation, electric vehicles, and other tasks, to protect the environment through “green energy” goals. EPRI focuses 40% of its funding on nuclear energy, which is zero for both DoE and the literature survey, because of the bad rap that this energy generation has received after the Fukushima accident provoked by the tsunami in 2011. The nature of EPRI has been closely linked to nuclear energy, as they are the outlets used by the government to further explore this possibility. Table 2.29: Hypotheses H3 results

2.7

CONCLUSIONS AND GRAPHICAL VIEW

2.7.1 Conclusions Our conclusions are based on the analysis of 966 peer reviewed papers listed at ISI Web of Science, from the period ranging from 2008 to 2013 that focused on the Smart Grid and 87

distribution topics. Overall, we observed that there is an increasing trend in the number of published papers over the past six years. Over 60% of the articles were related to empirical research, which is not surprising because Smart Grid distribution efforts require a high level of analysis before its implementation in the field. The increasing trends in papers suggest that the Smart Grid is not a fad because the trend has been growing for over six years. Even though the list of conferences papers for 2013 is not complete, we are projecting a trend upward which can be supported with the journal papers trend, which is not expected to change more in the following months. Journal papers follow a model that is based on the number of conference papers, so we can predict the journal papers based on historical performance and use the regression formula calculated in chapter 5. Empirical vs. conceptual. 604 of the papers related to Smart Grid distribution are published by IEEE. They represent 63% of the total papers analyzed. Therefore we expected that there would be more empirical papers than conceptual ones, because this is mostly an engineering area of study. The amount is overwhelmingly high compared to conceptual presentations because of the growing interest on the application of Smart Grid distribution technologies implemented or to be implemented in the field using already proven practices from other fields. Empirical papers represent 62.5% of all theory development documents because the empirical approach is based on using data to support or develop theories. Case studies vs. archival and laboratory research. The focus on proving the applications on the field can also be seen as more case studies are used than lab or archival research studies, because SGD can hardly be simulated under controlled conditions. Thus field work is necessary. The uniqueness of case studies can be a detriment to the overall implementation, but a conscious decision making process while selecting the cases can allow authors to further project their

88

findings. Applications are the main purpose of the empirical research and it is evident how the number of empirical case studies’ papers focused on applications is 24.5% of all analyzed papers and 43% of all empirical papers. This shows the importance of practical applications in the Smart Grid literature of these times. Methods of analysis. We discovered that the dominant choice of analysis method is simulation, as it appears in 35% of all analyzed papers. As there are so many expectations in the field of Smart Grid distribution, simulating the new technologies and theories is an important tool for success. Many of the papers are still lacking the validation in actual findings, thus making them incomplete in their effort, but the first steps of contemplating a possible scenario is accomplished here. Types of categories. Strategy, quality, and supply chain papers are topics with a positive trend, and they alone represent 74.5% of all papers. Although strategy and supply chain were affected by the reduced number of conferences in 2013, they are still trending up. Quality assurance management is continuously growing. Even with the 2013 drop of conferences, there are still more papers addressing quality assurance than in 2012. The quality focus of the papers is based on a popular new definition of the Smart Grid as an effective way of distributing energy, so efficiency can be related to achieving quality in the technologies. This approach is supported worldwide. Taxonomy of the papers’ purpose. The relationship between Smart Grid and ICT is very strong, as one of the modern definitions of Smart Grids is the use of information and communication technologies to transmit and distribute energy. Thus it is not surprising that 97% of all analyzed papers refer to ICT in some way. Due to the fact that most papers focus on applications, physical interconnection or infrastructure papers account for 72%. One area that

89

seems to be growing in this subject is the economic studies, representing 53% of all papers, because they, together with ICT, reinforce the definition of efficient systems to distribute energy at the optimum cost. Smart Grid categories. The top category is distributed generation, which includes worldwide effort to use renewable resources generation and microgrids. Distribution automation is second, focusing on the enhancements to eliminate or reduce dependency on operators with the use of computer and information and communication systems. In third place we have system efficiency effort, as authors are clearly stating that by improving quality of the current process, we can overcome the need for more generation, as distribution losses are reduced or even eliminated. Countries generating literature. The US is the clear leader in all areas showing the important role that it is taking into the third millennium— leading all global efforts toward clean tech energy. China is a distant second in most categories. We do not foresee them taking a leadership position anytime in the near future. Italy, Brazil and Canada are also showing strength in their generation of literature but still far below the US. China’s approach is to work on improving quality of some technologies. Italy’s approach is growing on the generation of distributed automation papers, and Portugal’s approach, even though they are # 12 on the list, focuses upon having a higher content on PHEV— with most of the papers coming from the same group of authors who are focusing on the charging process of electrical vehicles. It is interesting to see how the strategic papers are being replaced by quality assurance papers, largely because strategic papers are prevalent among distribution automation, efficiency, self-healing, and demand responses because these technologies are being strategized towards implementation. Distributed generation and electric vehicles papers’ focus on supply chain as

90

more efforts are taken to develop new forms of energy generation. These papers also attempt to develop a new business model where stakeholders will be generating and consuming energy. Distribution automation brings forward an opportunity of efficiency by removing decision making elements that make the process variable, and replaces them with automated controls which will react the same way consistently and are based on more available and trustworthy information. As more ICT devices are included in the grids, consumers have access to real-time information to help them shed loads during peak times, helping to conserve energy. 2.7.2

Graphical View

In order to see the identified technologies in a model, we used the Plan-Do-Check-Act (PDCA) cycle (Du et al., 2008). The PLAN section was replaced with the supply chain term, which includes the generation element from EPRI. We also included the Electric Vehicles category because the future of electricity might include vehicles connections to the grid (V2G). The DO element in the cycle includes the Information and Communication Technology (ICT), which is represented mainly by Distribution Automation. The CHECK and ACT sections were merged into a section that we are going to name Stakeholders Participation; here we need to have Distribution Networks to communicate the required data with low latency and broad bandwidth; the other element which is also very important is the participation of all stakeholders, so Demand Response is added to this part of the cycle. The result of the model shall include two major concepts: Efficiency and Environment protection, which we then merge into the Green Energy concept. Placing the results of the literature survey in a graphical manner, we were able to develop the preliminary model chart that shows the number of papers next to the name—they are color coded with dark green for the higher and red for the lowest.

91

Figure 2.42: Number of Papers per Technology in the Preliminary Model Graph With this first look to the model, we see that the most papers are related to supply chain, mostly in regards to the distributed generation topic. Second in importance is Distribution Automation, which is the penetration of ICT into the distribution of energy. In third place is efficiency, which is one of the most targeted efforts for researchers and also for the government funds.

92

CHAPTER 3: DIFFUSION OF TECHNOLOGIES AND RISKS 3.1 GENERATION AND CONSUMPTION INFORMATION In the year 2009, according to the World Bank, 74.13% of the world’s population had access to electricity. The region where this access was most restricted is Sub-Saharan Africa (SSA), where only 32.34% of the population enjoyed this service. Considering Latin America and the Caribbean (LAC) as another poverty pole region, the database reported that 93.42% of LAC inhabitants had access to electricity. There was a significant difference between LAC and SSA, as the situation in LAC is comparatively much better. Using numbers from the 2009 census, we deduced that 38 million people in LAC did not have access to electricity, compared to 562 million people in SSA. Thus there is a need for further expansion of distribution and generation in the LAC, because the consumption of electricity in this region was 1,890 KWh (kilowatt hours) per inhabitant—compared against the world average of consumption at 2,713 KWh. There is still a great deal of effort necessary, but the situation in SSA is much worse, as they only consume 456KWh. Because there are definitively more urgent needs at SAA than electricity, this paper will focus on LAC and the implementation of new technologies for the generation and distribution of electricity. If we focus on the poorest countries, in the year 2000, a little bit over 33% of the population had the chance of accessing electricity; based on these countries’ needs companies were overproducing energy. As a result of this decision, losses in transmission and distribution averaged 25% of the total production, so that those firms involved in the electricity sector were operating mostly at significant losses (Kenny & Søreide, 2008). Another factor affecting the progress in the electricity sector in the LAC was the popular idea, from the 1960’s, that reinforced the state as requiring a monopolistic and preponderant role 93

in everything related to energy provision and distribution. The LAC is now facing serious challenges, as it is creating competitive markets while modernizing old institutional structures and national standards and norms (Herrera, 2001). These changes have helped the penetration of energy generation and distribution in the LAC, with a high percentage of private participation in this sector. According to the World Bank database, the LAC has received 25.3 billion dollars of investments from private companies. This influx of capital into developing countries in the LAC is important for those economies and represents an important push towards technologies that the public sector could not finance—in some cases due to money supply and technological advancement. 3.1.1

Hypotheses H4

Private investors are getting into the energy generation sector and the areas that need more money are the ones related to macro-generation of electricity, such as large thermoelectric or wave generating processes. It is our expectation that most of the private capital is coming into macro-generation plants that government cannot support or fund anymore. Thus our next hypothesis is: H4a: The largest private sector global investment is devoted to the generation of energy Modernization is bringing an ever increasing growth of energy demand while generation depends on growth by adding more macro-generation sources; therefore if there is not enough investment to increase capacity or make more efficient the current sources, demand is going to be higher than generation in the near future. Therefore, we hypothesize that: H4b: The slope of energy consumption is greater than energy generation, thus resulting in an expected breakdown.

94

Energy efficiency improvement measures, distributed generation, and other measures are silently coming in, resulting in a reduction of demand pressure for the future. It is our expectation that by analyzing the past data, we can see that the demand pressure is softening in the future based on generation additions, improvements and efficiency. If this claim is correct, we propose that: H4c: Sensitivity analysis shows that the break-even year is moving forward as efficiency and other measures kick in Environmental groups are putting so much pressure on governments to reduce thermal generation of electricity, as it is one of the processes that are emitting more CO2 into the atmosphere. Even with this pressure, the electricity generation infrastructure and economical alternatives are not presenting a clear replacement alternative in the near future. Therefore, we expect thermal energy to still continue growing in the future; thus we hypothesize that: H4d: Thermal energy generation will be the source with the highest growth slope in the future. The world’s population growth and lowering prices of some devices that are becoming indispensable are forcing an ever increasing growth of household consumption, putting extra pressure on electricity generation. We expect the slope of consumption of homes to be the one with the largest slope of growth as more economies are joining the global ICT growth. Therefore, we proposed the following hypothesis: H4e: Household consumption has the largest slope of all areas 3.1.2

Private Investments on Energy

There has been a concentration of private investment in the energy area of the LAC, compared to the rest of the world, as can be seen in Figure 2.39 from the World Bank database. Analyzing the investments in LAC, it is clearly noticeable that Brazil is by far the country where 95

the most private capital has been invested in energy, mostly toward two important projects: The 3.15 GW mega power plant San Antonio Hydro, with an investment of $7 billion USD, and the 3.3 mega power plant Jirau Hydro, with a private investment around $5.5 billion USD. These plants were financed in 2009, so their completion dropped the private investment in 2010 to 60%. Putting Brazil aside, however, the other countries grew 50% compared to 2009 (Jett, 2011).

Figure 3.1: Private investment in Energy’s sector around the world Approximately 80 percent of the investments in LAC countries were in divestiture projects, that is, governments engaged in privatization and allowed independent power producers (IPP) to enter into the electricity arena of the country (Jamasb, 2006). Because electricity has to be transferred from the generation source to the consumer through transmission and distribution lines, which belong to the utility, IPPs have to pay for the transference of energy from its generating point to their final consumers (Jia & Yokohama, 2003). Unfortunately, privatization in past years has become a source closely related to corruption, as public or state owned companies have traditionally been one of the greater sources of corruption—wherein political parties were financed and “loyal” stakeholders were rewarded with certain jobs (Tanzi, 1998).

96

Figure 3.1 shows the investment buckets and geographical areas where they are used. The investments on generation are well above all others, thus supporting our hypothesis H4a, which states that generation is the main investment for private capital. In the 1990’s, there was an important participation of private investment in the electricity sector of more than 75 developing countries; the approximate total amount of private investments was around $160.7 billion US dollars in 695 projects (Jamasb, 2006). Radulović (2009) concludes that future changes in the energy area will have an important impact on a new economy, an economy of knowledge. With the growth of technology around the world, there is an important trend of energy demand increase that requires additional infrastructure construction, network expansions and robustness, security of supply chain across markets, optimum service to consumers and maximum profits. Unfortunately, there has been recently a reduction of interest on private investment, and one of the main issues for this reduction of private funding of the electricity sector has been the issue of governance. Focusing on the electricity sector in LAC, Jamasb (2006) reported that successful reforms may improve efficiency and offer lower prices with better quality for electricity services. But if we implement flawed reforms and inefficient regulation and competition, these issues can eliminate the benefits of implementing reforms. As energy generation and distribution has been a natural monopoly, it is necessary to regulate it in order to optimize performance toward economic efficiency (Kenny & Søreide, 2008). 3.1.3

Sources of Consumption and Generation Statistics

Based on secondary data, available at the United Nations Databank, we discovered that there is a tremendous consumption growth that is quickly closing the gap of electricity production.

97

Figure 3.2: Evolution of consumption and generation of electricity Using a linear regression, we calculated the y-axis crossing point and the slope of the line for our projections. The slope (B1) is larger for consumption than for generation; therefore, hypothesis H4b is supported. The following statistics back-up our projections for the consumption and generation activities: Table 3.1: Consumption and generation regression lines’ statistics Consumption Generation

B0 6.438 10.577

B1 0.600 0.482

Std. Error 0.025 0.019

Beta t 0.984 23.894 0.985 24.877

Sig. R-square 0.000 0.968 0.000 0.970

If the trends continue the same, by the year 2025, consumption will be larger than production, aggravating the supply problem that already exists. Countries with natural resources could see this as an opportunity to export their electricity, with the caveat of losses in transmission.

98

Figure 3.3: Projected lines of consumption and generation until 2035 Conducting a sensitivity analysis for consumption and generation, we see that the y-axis crossing point and the slope have been changing every year. By projecting the lines after 1999, when we have 10 years data, we calculate the crossing point of the curves to see if the projected year of 2025 has been the expected breaking point when consumption will overcome generation. Table 3.2: Sensitivity analysis for the crossing-point year

Year '99 '00 '01 '02 '03 '04 '05 '06 '07 '08 '09 '10

Data points 10 11 12 13 14 15 16 17 18 19 20 21

Consumption B0 B1 5.513 0.783 5.658 0.746 5.835 0.706 6.006 0.669 5.967 0.677 6.084 0.655 6.157 0.642 6.205 0.634 6.219 0.632 6.259 0.626 6.391 0.607 6.438 0.601

R 0.96 0.963 0.963 0.963 0.97 0.972 0.975 0.978 0.981 0.983 0.982 0.984 99

Generation B0 B1 11.500 0.317 11.410 0.339 11.400 0.343 11.350 0.353 11.280 0.367 11.180 0.387 11.050 0.409 10.930 0.430 10.790 0.452 10.690 0.466 10.670 0.468 10.580 0.482

R 0.991 0.988 0.991 0.991 0.991 0.988 0.985 0.983 0.981 0.982 0.984 0.985

Crossing point 12.87 14.14 15.33 16.93 17.17 18.99 20.96 23.12 25.38 27.73 30.90 34.81

Table 3.2 shows the changes in the curves from 1999 to 2010, including the calculated crossing point. Table 3.2 shows how the expected crossing point year has been moving up in the future, as generation lines have been increasing strongly to prevent this deficit. This equilibrium point shift supports hypothesis H4c. For consumption, we can see that the y-axis crossing point has been increasing every year, while the slope is slowly decreasing. Based upon these observations we can infer that consumption is increasing but not as accelerated as expected, thanks to the energy saving efforts implemented in some countries. R-square has also been increasing, showing that we are explaining more efficiently the linear behavior of consumption in time.

Figure 3.4: Evolution of the crossing point from the sensitivity analysis For generation, the y-axis crossing point has been decreasing every year, while the slope is increasing; we can thus infer that generation is increasing at a more accelerated rate than expected. This seems to be the result of increasing generation sources to delay the feared moment of lack of electricity supply. The R-square has been moving along in, going up and down, but always predicting consistently an ever increasing generation effort. 100

After 2003 and 2008, there have been major changes in the generation of electricity that have pushed the crossing point back considerably. The increases moved from 1 year increase per year to almost 3 years per year now. Trying to better understand the changes in consumption, we conducted a sensitivity analysis for the main components of consumption: energy, households, industry and transport. Table 3.3: Consumption’s sensitivity analyses Year '99 '00 '01 '02 '03 '04 '05 '06 '07 '08 '09 '10

Data points 10 11 12 13 14 15 16 17 18 19 20 21

B0 0.278 0.292 0.302 0.311 0.318 0.323 0.328 0.335 0.340 0.347 0.355 0.358

Energy B1 0.037 0.033 0.031 0.029 0.028 0.027 0.026 0.025 0.024 0.023 0.022 0.021

R 0.898 0.894 0.898 0.900 0.907 0.915 0.922 0.924 0.929 0.928 0.926 0.931

Households B1 R 2.485 0.458 0.981 2.559 0.440 0.982 2.641 0.421 0.982 2.710 0.406 0.982 2.780 0.392 0.982 2.843 0.380 0.983 2.888 0.372 0.984 2.930 0.365 0.987 2.959 0.361 0.987 2.992 0.356 0.988 3.036 0.350 0.988 3.057 0.347 0.989 B0

Industry B1 2.663 0.270 2.714 0.258 2.792 0.240 2.881 0.221 2.763 0.244 2.809 0.236 2.829 0.232 2.825 0.233 2.803 0.236 2.801 0.237 2.878 0.226 2.898 0.223 B0

R 0.867 0.884 0.888 0.886 0.907 0.915 0.926 0.937 0.947 0.955 0.951 0.956

Transport B1 0.087 0.017 0.094 0.015 0.100 0.014 0.104 0.013 0.107 0.012 0.109 0.012 0.111 0.012 0.114 0.011 0.116 0.011 0.119 0.010 0.123 0.010 0.124 0.010 B0

R 0.798 0.797 0.799 0.806 0.823 0.842 0.853 0.863 0.872 0.876 0.876 0.885

Table 3.3 presents some data that is noteworthy: Energy, household and transportation consumptions are the main drivers of the path observed for consumption. Industry has been impacting demand with changes in 2003 and 2008—wherein consumption of electricity was reduced. In other words, industry consumption has been the element slowing down electricity consumption in the world. Although Industry B0’s are similar to the household consumption ones, the slope (B1) is much greater for household consumption; therefore, we can support hypothesis H4e that the homes ‘consumption of electricity is the factor with the largest growth.

101

Because generation of electricity can be further divided into 6 main categories (according to the UN database), we split generation and conducted a linear regression analysis for these 6 elements. The results are shown in Table 3.4. Table 3.4: Generation sources sensitivity analysis Year '99 '00 '01 '02 '03 '04 '05 '06 '07 '08 '09 '10

Data points 10.00 11.00 12.00 13.00 14.00 15.00 16.00 17.00 18.00 19.00 20.00 21.00

Thermal B0 B1 R 7.29 0.21 0.98 7.20 0.23 0.97 7.16 0.24 0.98 7.09 0.25 0.98 7.00 0.27 0.98 6.91 0.29 0.98 6.80 0.31 0.97 6.69 0.33 0.97 6.55 0.35 0.97 6.47 0.36 0.97 6.46 0.36 0.98 6.39 0.37 0.98

B0 2.17 2.19 2.21 2.23 2.24 2.23 2.22 2.20 2.18 2.17 2.15 2.13

Hydro B1 0.06 0.05 0.05 0.04 0.04 0.04 0.04 0.05 0.05 0.05 0.05 0.06

R 0.96 0.96 0.93 0.93 0.93 0.94 0.95 0.95 0.96 0.96 0.97 0.97

Nuclear B0 B1 R 1.93 0.06 0.99 1.93 0.06 0.99 1.93 0.06 0.99 1.94 0.06 0.99 1.96 0.05 0.99 1.96 0.05 0.99 1.97 0.05 0.99 1.98 0.05 0.99 2.00 0.05 0.98 2.02 0.05 0.97 2.04 0.04 0.95 2.05 0.04 0.95

B0 0.00 0.00 0.00 -0.01 -0.01 -0.01 -0.02 -0.02 -0.03 -0.04 -0.06 -0.07

Wind B1 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.01 0.01

R 0.91 0.89 0.90 0.89 0.89 0.88 0.89 0.89 0.88 0.87 0.86 0.86

Geothermal B0 B1 R 0.04 0.00 0.96 0.04 0.00 0.97 0.03 0.00 0.97 0.03 0.00 0.98 0.03 0.00 0.98 0.03 0.00 0.99 0.03 0.00 0.99 0.03 0.00 0.99 0.03 0.00 0.99 0.03 0.00 0.99 0.03 0.00 0.99 0.03 0.00 0.99

B0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.01

Solar B1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

R 0.84 0.87 0.90 0.92 0.93 0.92 0.89 0.87 0.85 0.80 0.75 0.58

Tide/Wave B0 B1 R 0.00 0.00 0.15 0.00 0.00 0.16 0.00 0.00 0.15 0.00 0.00 0.36 0.00 0.00 0.48 0.00 0.00 0.60 0.00 0.00 0.66 0.00 0.00 0.72 0.00 0.00 0.76 0.00 0.00 0.80 0.00 0.00 0.83 0.00 0.00 0.83

Table 3.4 shows that wind, solar, thermal and tide generation are the main drivers of the path, where the y-axis crossing point is being reduced every year while the slope is increasing. The only two sources that do not follow this pattern are nuclear energy and the other sources. Nuclear energy seems to be slowing down in time, although we know for a fact that following 2011 the amount of nuclear energy will be greatly reduced globally after the directives of some countries forbidding nuclear generation. Thermal generation is by far the largest contributor, but the other sources have been steadily generating electricity. Following 2004, however, they have reversed the trend on both slope and y-axis crossing point. 3.1.4

Global Trends on Energy Use and Availability

The situation might even be more critical because there are other ways that we are making use of electricity aside from the consumption by industry, transportation, households, and others. These other uses of electricity are forms of conversion to other energies, station use and station 102

losses, as well as exports of extra capacity. The trends of these uses are also growing, with conversion acting more aggressively than others. Export is the one activity that has not grown that much, but it reduces the availability of electricity for citizens. Table 3.5: Regression statistics for Generation and Imports Imports Generation

B0 (15.806) 10.577

B1 54.613 0.482

Std. Error 3.586 0.019

Beta t 0.961 15.229 0.985 24.877

Sig. R-square 0.000 0.924 0.000 0.970

Figure 3.5: Other uses of electricity to consider (source: UN data) In order to cover the shortage of electricity, some countries are turning to importations of energy, as a way of overcome the obstacles they face in not having enough electricity to meet current consumption rates. The growth of importations is so small, that in 2031 we expect importations to represent 1 out of 31.8 billion KWh of produced electricity, which is only 3.2%.

103

Figure 3.6: Importations of electricity and current trend

Figure 3.7: Trend of global use and availability of electricity (source: UN data) Adding importations to the production of electricity gave us the total amount of electricity available for all the nations in the world, and adding the consumption, conversion, losses, and exportations, we can clearly see the prognosis for the short and mid-term future. The

104

following statistics were used to calculate the projection line for available electricity, which includes gross production plus importations, and use of energy—which includes consumption, conversion, exports and losses. Table 3.6: Regression lines for Energy use and availability Available Use

B0 (20.963) (11.218)

B1 1.953 1.417

Std. Error 0.075 0.056

Beta t 0.986 25.913 0.986 25.35

Sig. R-square 0.000 0.972 0.000 0.971

Comparing the trends of the available and used electricity, we can see that the electricity crisis is not something that will come, but is already here. In 2004 both lines crossed, meaning that there has not been enough electricity for all people on the planet.

Figure 3.8: Projected lines of use and availability of energy (source: UN data) Conducting a sensitivity analysis for the use and availability lines, we came up with Table 3.7, which compares the crossing points for both lines beginning in 1999 until 2010. Just like in the consumption versus generation chart, the use and available energy data follows the same path. Use of energy is increasing, as the y-axis crossing point is increasing, but the slope is 105

tending downwards, while availability of energy y-axis crossing point is decreasing and the slope is growing. Based on the UN data, we already have a shortage of electricity supply since the period of 1997-2003 but it is hard to explain how the use of energy can be above the amount of available energy, so we are going to dig deeper to discover why. Table 3.7: Sensitivity analysis for crossing line year for use and available energy Year '99 '00 '01 '02 '03 '04 '05 '06 '07 '08 '09 '10

3.1.5

Data point 10 11 12 13 14 15 16 17 18 19 20 21

Use of Energy B0 B1 7.158 0.879 7.307 0.842 7.521 0.793 7.699 0.755 7.669 0.761 7.785 0.739 7.842 0.729 7.886 0.722 7.900 0.719 7.944 0.713 8.085 0.693 8.138 0.685

R 0.962 0.966 0.964 0.964 0.972 0.974 0.977 0.98 0.983 0.985 0.984 0.986

Available Energy B0 B1 12 0 12 0 12 0 12 0 12 0 11 0 11 0 11 0 11 0 11 0 11 0 11 0

R 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.98 0.98 0.98 0.99 0.99

Crossing point 8.516 9.080 9.671 10.289 10.403 11.029 11.612 12.184 12.764 13.328 13.995 14.690

Global Trends on Generation and Sources

Also using the secondary data available at the UN Database, we were able to analyze the sources of generation capacities and trends for the near future. Because the unit of measure for the installed capacity is in KW, a unit of power, and the units for produced electricity is in KWH, a unit of energy, we used the data to calculate the capacity factors per country per year in order to make our prognosis. Thermal energy is the generation process with the largest installed capacity and it is also the largest generator of electricity in the world, therefore, hypothesis H4d is supported because we do not expect all this capacity to be uninstalled but rather continue to function. 106

Figure 3.9: Growth of electricity capacity assuming 6870 hours per year (source: UN data)

Figure 3.10: Sources of production of electricity (source: UN data) Breaking the production of electricity into the major source of generation, we found that there are some other sources that are not considered in the major categories that the UN database includes, so we added a group named “others”.

107

Figure 3.11: Capacity percentage -Production vs. installed capacity (source: UN data) The capacity has been growing way beyond the needed consumption of energy, but we are assuming a perfect world where we will have energy generation every hour of the whole year Table 3.8: Capacity factors Source Solar Nuclear Thermal Tide Wind Others

Capacity 26% 83% 54% 24% 23% 40%

To understand how low the capacity is, Figure 3.11 shows us the global capacity for every source of energy, focusing on percentages in the past ten years’ period. We created this focus because during the 90’s there were some testing and erratic data. The statistics show stability on most generation sources, except solar energy, that had a trend down that seems to be bouncing back up, but we will have to see to what degree solar energy recovers. Based on the capacity factors, we predict growth of installed capacity, based upon its impact on 108

consumption— as shown in the prior section. To assess the veracity of the data, we calculated the capacity per country and the production. We found some interesting facts. Namely, China does not report any solar capacity and is actually the largest producer. The plan in China for photovoltaic is so aggressive that it is expected to generate over 1,800 MW by the year 2020, with a stretch goal of 10,000 MW or more (Shen & Wong, 2009). 3.1.6

Global Sources of Generation Capacity versus Production

The following tables show the comparison of contributions by country for the generation compared to the production of electricity. It is noteworthy that some data seems incomplete or inaccurate as available within the United Nations Databases. Table 3.9: Hydro capacity (KWh) Hydro Country Capacity Contribution China 1,699,440 19.5% United States 884,961 10.2% Brazil 706,380 8.1% Canada 657,683 7.5% Japan 418,167 4.8% Russian Federation 410,003 4.7% India 329,621 3.8% Norway 254,066 2.9% France 220,875 2.5% Italy 188,515 2.2% Spain 164,119 1.9% Sweden 146,572 1.7% Turkey 138,680 1.6% Venezuela(Bolivar. Rep.) 128,089 1.5% Switzerland 120,187 1.4% Austria 111,261 1.3% Mexico 99,321 1.1% Germany 96,605 1.1% Argentina 87,994 1.0% Australia 83,553 1.0% Others 1772285 20.3% 8,718,378 100.0%

109

Table 3.10: Hydro production (KWh) Hydro Country Production Contribution China 722,172 20.5% Brazil 403,289 11.5% Canada 351,591 10.0% United States 286,333 8.1% Russian Federation 168,397 4.8% Norway 117,942 3.4% India 114,486 3.3% Japan 90,682 2.6% Venezuela(Bolivar. Rep.) 76,780 2.2% France 66,825 1.9% Sweden 66,501 1.9% Italy 54,406 1.5% Paraguay 54,065 1.5% Turkey 51,796 1.5% Spain 45,488 1.3% Austria 41,596 1.2% Colombia 40,624 1.2% Switzerland 37,825 1.1% Mexico 37,121 1.1% Argentina 33,897 1.0% Others 654,404 18.6% 3,516,219 100.0%

Table 3.11: Nuclear Capacity (KWh) Nuclear Country Capacity Contribution United States 886,223 26.5% France 553,019 16.5% Japan 428,890 12.8% Russian Federation 207,980 6.2% Germany 179,291 5.4% Korea, Republic of 155,192 4.6% Ukraine 121,195 3.6% Canada 110,945 3.3% China 96,360 2.9% United Kingdom 95,177 2.8% Sweden 78,639 2.3% Spain 65,034 1.9% Belgium 51,921 1.6% Other Asia 45,061 1.3% India 41,873 1.3% Czech Republic 34,164 1.0% Switzerland 28,496 0.9% Bulgaria 24,283 0.7% Finland 23,652 0.7% Brazil 17,581 0.5% Others 103,622 3.1% 3,348,598 100.0%

110

Table 3.12: Nuclear Production (KWh) Nuclear Country Production Contribution United States 838,931 30.4% France 428,521 15.5% Japan 288,230 10.5% Russian Federation 170,415 6.2% Korea, Republic of 148,596 5.4% Germany 140,556 5.1% Canada 90,659 3.3% Ukraine 89,152 3.2% China 73,880 2.7% United Kingdom 62,140 2.3% Spain 61,990 2.2% Sweden 57,828 2.1% Belgium 47,944 1.7% Other Asia 41,629 1.5% Czech Republic 27,998 1.0% Switzerland 26,339 1.0% India 26,266 1.0% Finland 22,800 0.8% Hungary 15,761 0.6% Bulgaria 15,249 0.6% Others 81,405 3.0% 2,756,289 100.0%

Table 3.13: Solar Capacity (KWh) Solar Country Capacity Contribution Germany 151,723 45.1% Spain 40,366 12.0% Japan 31,694 9.4% Italy 30,397 9.0% United States 29,547 8.8% Czech Republic 17,161 5.1% Belgium 7,919 2.4% France 7,823 2.3% Korea, Republic of 5,694 1.7% Greece 1,770 0.5% Australia 1,638 0.5% Lao People's Dem. Rep. 1,542 0.5% Egypt 1,226 0.4% Portugal 1,174 0.3% Switzerland 972 0.3% Canada 946 0.3% Austria 832 0.2% Reunion 782 0.2% Netherlands 771 0.2% United Kingdom 675 0.2% Others 2,036 0.6% 336,688 100.0%

111

Table 3.14: Solar Production (KWh) Solar Country Production Contribution China 34,558 51.3% Germany 11,682 17.3% Spain 7,105 10.5% United States 3,934 5.8% Japan 3,799 5.6% Italy 1,906 2.8% Korea, Republic of 772 1.1% Czech Republic 616 0.9% France 564 0.8% Belgium 560 0.8% Australia 277 0.4% Portugal 211 0.3% Egypt 206 0.3% Lao People's Dem. Rep. 193 0.3% Canada 158 0.2% Greece 158 0.2% Austria 89 0.1% Switzerland 83 0.1% Reunion 76 0.1% Israel 70 0.1% Others 374 0.6% 67,391 100.0%

Table 3.15: Thermal Capacity (KWh) Thermal Country Capacity Contribution United States 6,975,614 22.6% France 6,797,760 22.0% Japan 1,615,510 5.2% Russian Federation 1,437,674 4.7% Germany 1,344,003 4.3% Korea, Republic of 709,201 2.3% Ukraine 663,167 2.1% Canada 637,167 2.1% China 529,332 1.7% United Kingdom 503,323 1.6% Sweden 463,172 1.5% Spain 426,621 1.4% Belgium 413,744 1.3% Other Asia 406,236 1.3% India 395,102 1.3% Czech Republic 384,908 1.2% Switzerland 354,780 1.1% Bulgaria 332,214 1.1% Finland 308,229 1.0% Brazil 283,587 0.9% Others 5,924,787 19.2% 30,906,132 100.0%

112

Table 3.16: Thermal Production (KWh) Thermal Country Production Contribution China 3,331,928 22.5% United States 3,136,499 21.2% India 813,787 5.5% Japan 729,916 4.9% Russian Federation 698,709 4.7% Germany 411,569 2.8% Korea, Republic of 342,851 2.3% United Kingdom 302,019 2.0% South Africa 242,358 1.6% Saudi Arabia 240,067 1.6% Italy 231,249 1.6% Australia 223,987 1.5% Iran(Islamic Rep. of) 223,266 1.5% Mexico 220,080 1.5% Other Asia 197,113 1.3% Indonesia 159,560 1.1% Canada 155,959 1.1% Turkey 155,828 1.1% Thailand 153,953 1.0% Poland 152,505 1.0% Others 2,667,005 18.0% 14,790,208 100.0%

Table 3.17: Wind Capacity (KWh) Wind Country Capacity Contribution United States 342,814 26.3% Germany 238,351 18.3% Spain 181,884 14.0% China 92,856 7.1% France 52,201 4.0% Italy 50,755 3.9% United Kingdom 47,111 3.6% Canada 34,803 2.7% Denmark 33,306 2.6% Portugal 33,253 2.6% Japan 20,095 1.5% Netherlands 19,596 1.5% Sweden 17,686 1.4% Australia 16,329 1.3% Ireland 12,317 0.9% Turkey 11,563 0.9% Greece 11,370 0.9% Poland 9,706 0.7% Austria 8,559 0.7% Belgium 7,989 0.6% Others 59,559 4.6% 1,302,104 100.0%

113

Table 3.18: Wind Production (KWh) Wind Country Production Contribution United States 95,148 29.8% China 44,622 14.0% Spain 44,165 13.8% Germany 37,793 11.8% United Kingdom 10,183 3.2% France 9,969 3.1% Canada 9,557 3.0% Portugal 9,182 2.9% Italy 9,126 2.9% Denmark 7,809 2.4% Australia 4,798 1.5% Netherlands 3,993 1.2% Japan 3,962 1.2% Sweden 3,502 1.1% Turkey 2,916 0.9% Ireland 2,815 0.9% Greece 2,714 0.8% Austria 2,064 0.6% Poland 1,664 0.5% New Zealand 1,634 0.5% Others 12,144 3.8% 319,760 100.0%

3.1.7

Diffusion of Renewable Generation

Because the literature review presented the importance of distributed generation, this dissertation is going to analyze the diffusion of the different sources of generation that use renewable resources in the US. Table 3.19 shows the breakdown of the generation of electricity. Table 3.19: Evolution of Sources of Electricity Generation in the US Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Natural Gas 29.8% 34.5% 37.5% 38.5% 39.2% 39.4% 39.5% 39.3% 39.1% 39.2% 39.5%

Coal 37.0% 34.8% 33.0% 32.5% 32.0% 31.7% 31.4% 31.0% 30.7% 30.5% 30.2%

Nuclear 11.6% 10.9% 10.5% 10.3% 10.2% 10.2% 10.1% 10.0% 9.9% 9.7% 9.6%

Hydro Hydro Wood & Other Geo- Other Solar & Other Petroleum Wind Conv. Pump. Stor. Der. Fuels Biomass thermal Gas PV Sources 9.3% 7.8% 0.5% 2.3% 0.7% 0.4% 0.3% 0.2% 0.0% 0.1% 8.8% 6.6% 0.5% 2.3% 0.6% 0.4% 0.2% 0.2% 0.0% 0.1% 8.3% 6.4% 0.6% 2.2% 0.6% 0.4% 0.2% 0.2% 0.0% 0.1% 8.1% 6.1% 0.7% 2.2% 0.6% 0.4% 0.2% 0.2% 0.0% 0.1% 7.9% 6.0% 0.9% 2.2% 0.6% 0.4% 0.2% 0.2% 0.0% 0.1% 7.9% 5.9% 1.1% 2.2% 0.6% 0.4% 0.2% 0.2% 0.0% 0.1% 7.8% 5.6% 1.7% 2.2% 0.7% 0.4% 0.2% 0.2% 0.1% 0.1% 7.7% 5.7% 2.4% 2.2% 0.7% 0.4% 0.2% 0.2% 0.1% 0.1% 7.7% 5.5% 3.3% 2.2% 0.7% 0.4% 0.2% 0.2% 0.1% 0.1% 7.6% 5.4% 3.8% 2.1% 0.7% 0.4% 0.2% 0.3% 0.1% 0.1% 7.5% 4.9% 4.3% 2.1% 0.7% 0.4% 0.2% 0.2% 0.1% 0.1%

114

Table 3.19 clearly shows that most trends are going downwards with the exception of Natural Gas, Wind, and Solar generations— which are consistently trending upwards. Because gas is not a renewable resource, we are going to conduct Bass’ diffusion analysis (as detailed in section 2.6.2) only for the wind and solar electricity generation processes by state in the US using the data available at the US Energy Information Administration (EIA). 3.1.7.1

Wind Generated Electricity in the US

We are seeing Texas as the largest contributor for the electricity generated with wind, with Iowa following at a distance with more than one third of the Texas generation; but Iowa’s growth is important with a much faster diffusion speed than anyone else in the list. California shows an imitation index of zero, which means that only innovators are implementing wind energy because the curve shows a flattened curve, as this state has been generating electricity using wind mills long before anyone else. Therefore, we do not see as a rapid growth in California as Washington and Minnesota, which will reach maturity much faster. Table 3.20: Diffusion Indexes for Wind Electricity Generation in the US by States

Wind Energy Cum. Level Maturity Level Innovation US (KWh) (KWh) Index USA 120,177 427,082 0.00007 Texas 30,548 45,580 0.00026 Iowa 10,709 19,924 0.00017 California 7,752 98,123 0.00327 Minnesota 6,726 12,463 0.00032 Washington 6,262 10,272 0.00627 Illinois 6,213 9,101 0.00110 Oklahoma 5,605 23,135 0.00491 North Dakota 5,236 6,193 0.00065 Colorado 5,200 6,475 0.00137 Oregon 4,775 6,575 0.00059

115

Imitation Index 0.3424 0.4873 0.4512 0.3179 0.4102 0.7206 0.2851 0.8862 0.5436 0.5166

Diffusion Speed 4,711 1,884 2,719 996 65 654 58 1,373 398 869

Table 3.21: t-test for top 10 States using Wind generated Electricity Test Value = 0 Wind Cum M P Q

t

df

Sig. tailed)

3.611 2.614 2.693 6.057

9 9 9 9

.006 .028 .025 .000

(2- Mean Difference 8902.68721 23784.20434 .00187 .46315

95% C.I. of the Difference Lower

Upper

3325.1518 3205.0345 .0003 .2902

14480.2226 44363.3742 .0034 .6361

Doing t-tests for the states’ cumulative, maturity level, and innovation and diffusion indexes, we find them to be significantly below 2.8%. The top ten states show that the diffusion of wind generated energy will reach maturity around the year 2050, when some scholars are predicting that renewable energy will represent 50% of all electricity in the US (Lehr, 2013).

Figure 3.12: Diffusion of Wind Generation by State

116

3.1.7.2

Solar Generated Electricity in the US

For solar energy, California is undoubtedly the largest contributor with 48.9% of all solar electricity in the country. California has been a pioneer in renewable resources electricity, so we are also seeing an imitation index of zero, as well as for wind energy. This means that the curve is flattened and all the growth is identified as innovator growth that will take much longer to mature. Nevada is second on order, but its growth is strongly driven by imitation, which means that it will reach maturity in 2018. New Mexico is a state that, although it is just beginning with the solar energy experience, presents an imitation factor of 3.02, which means that this factor is steepening the curve. Thus New Mexico will reach maturity comparatively much faster, in 2015.

Figure 3.13: Diffusion of Solar Generation by State

117

Table 3.22: Diffusion Indexes for Solar Electricity Generation in the US by States

Cum. Level Maturity Level Innovation Imitation (KWh) (KWh) Index Index 1,817.696 35,876.126 0.0012 0.0177 888.829 986.974 0.0641 291.225 329.745 0.0521 0.6701 127.802 401.676 0.0002 3.0238 125.726 359.213 0.0130 0.9238 104.636 317.011 0.0019 0.8971 83.349 409.752 0.0000 0.8997 68.878 447.292 0.0006 1.1779 28.639 86.987 0.0000 0.4226 23.144 957.176 0.0001 1.0459 17.380 21.606 0.0071 1.3253

Solar US California Nevada New Mexico Florida Colorado Arizona New Jersey Texas Pennsylvania North Carolina

Diffusion Speed 32,097 31,751 5,729 8,644 30,807 28,780 54,652 24,707 3,932 37,694 37,694

For solar generation, the t-tests results are significantly below 10%. The maturity level and q index are the most significant, at below 5%, which demonstrates that the analysis is correct and there will be a quick diffusion process for solar energy in the US Table 3.23: t-test for top 10 States using Solar generated Electricity Test Value = 0 Solar Cum M P Q 3.1.8

T 2.117 4.313 1.846 4.131

df 9 9 9 9

Sig. (2-tailed) .063 .002 .098 .003

Mean Difference 175.95814 431.74955 .01390 1.03862

95% C.Il of the Difference Lower Upper -12.0488 363.9651 205.2937 658.2054 -.0031 .0309 .4699 1.6074

Global Prognosis

Coal is one of the major concerns in the world for the emission of CO2 to the environment, as it contributes to the greenhouse effect— the process responsible for global warming. It is concerning to see the tremendous growth of energy generation with the use of coal in China, while in the United States coal use is decreasing, as shown in Figure 3.14.

118

Figure 3.14: Energy generation using coal (source: World Bank database) Looking at the generation of electricity using hydro power, we see China growing tremendously. China is the number one producer of hydro power and it is still growing strong, while the other countries have remained in a steady state using these natural resources. We also see Brazil growing slowly in this area, after the investments referenced in the section 3.1.2.

Figure 3.15: Electricity generated using hydro power (source: World Bank database)

119

Natural gas has been identified as one of the cleanest sources of energy, causing less harm to the environment; the US is the leader in its implementation. We can see that Russia is slowly growing in second place but it would take a long time to take the leadership position.

Figure 3.16: Electricity generated with natural gas (source: World Bank database)

Figure 3.17: Electricity generated using nuclear plants (source: World Bank database)

120

Nuclear energy had been steadily growing throughout the world, but following the Fukushima incident there are some countries that are reversing their efforts toward nuclear energy until preventive measures can be implemented to prevent disasters.

Figure 3.18: Electricity generated burning oil (source: World Bank database)

Figure 3.19: Global energy generation sources (source: World Bank database)

121

Looking at Figure 3.18, it is evident that the amount of energy produced using oil has been drastically reduced. Despite this global trend, countries such as Saudi Arabia and Iran are still rapidly increasing this generation processes. Figure 3.19 clearly shows that coal is the largest source used to generate electricity and will continue to be for a long time, until regulators begin inhibiting its growth. Natural gas is a strong second place that is growing, but not nearly as fast as coal. Hydro is clearly third and is also growing, with a big push from China. Nuclear energy will completely change its trend as more countries are prohibiting its use, and oil is also going down. Figure 3.20 shows that the leadership on electricity generation will shortly be changed from the United States to China, as Chinese growth is exponential compared to the US growth. The potential of the Chinese market has not been fully understood, as the communist government is investing all necessary capital to electrify the country and allow foreign industries to be located there, while the nationals have access to utilities that can improve their quality of life.

Figure 3.20: Worldwide top electricity generation producers 122

Table 3.24: Hypotheses H4 Results

The results of the proposed hypotheses in section 3.1.1 are presented in Table 3.24, which presents a challenging future for human society as consumption keeps growing and generation is mostly based on thermal generation that emits large amounts of CO2 to the atmosphere. Some actions are being taken by consumers, utility companies and governments that are delaying, but not stopping, the upcoming point where demand will exceed generation. Unfortunately much of the investment continues to support macro-generation, when one of the possible solutions (distributed generation) is not being supported as much.

3.2 ADVANCED METERING INFRASTRUCTURE BACKGROUND There have been important changes in the electricity sector of many countries. Another area that has been liberated is the telecommunication sector, beginning in the early 90’s. The market liberalization movements have forced industry and governments to shift from the basic assessment of production costs to the more complicated measurement of productivity. Energy and telecommunications are now coming hand in hand with the growth of Smart 123

Grids, placing consumers in a much better position as they can participate in this sector in ways that in the past used to be monopolistically controlled by utility companies. Most diffusion studies have been conducted at national levels for ICT technologies, but there is ample room to conduct studies at regional clusters of nations (Peres et al., 2010). Multi-national diffusion model-based studies of innovation exist; however, diffusion of ICT, along with electricity innovations in developing nations, is an area of opportunity according to Peres et al. (2010). This dissertation has chosen advanced metering infrastructure (AMI) as the Smart grid Distribution technology to measure SG diffusion in the US, individual states, and available information on worldwide data. In this dissertation we use Van den Bulte’s (2004) definition of diffusion speed as the speed at which new products gets adopted and diffused through the market. We investigate the relation between speed of diffusion and strong regional clustering, which can be introduced by physical proximity and other similar aspects of these states due to a “spill-over effect” introduced by diffusion in neighboring entities (Mohan Babu & Ganesh, 1997). We are clustering states by geographical location, as has been typically done in the United States by the Census Bureau: Table 3.25: Regional Clustering for the US Midwest Illinois Ohio Indiana South Dakota Iowa Wisconsin Kansas Michigan Minnesota Missouri Nebraska North Dakota

Northeast Connecticut Maine Massachusetts New Hampshire New Jersey New York Pennsylvania Rhode Island Vermont

South Alabama Mississippi Arkansas North Carolina Delaware Oklahoma District Columbia South Carolina Florida Tennese Georgia Texas Kentucky Virginia Louisiana West Virginia Maryland

124

West Alaska Oregon Arizona Utah California Washington Colorado Wyoming Hawaii Idaho Montana Nevada New Mexico

This section of the dissertation seeks to answer the following questions: How does the speed at which AMI are being adopted and diffused through the United States vary across regions and states? Using Bass’ diffusion model, as presented in section 2.6.2, we discuss the multi-level model for a panel of diffusion time series of smart meters. The Bass model assumes a population of m potential adopters, where, in the context of citations, we will associate m with the maturity level or in this case the maximum amount of smart meters to be installed at that entity. In our context, adopters should be viewed as consumers that have a smart meter installed on site. It is important to note that if p is zero, the ratio will tend to infinite, so we use log of q/p. Table 3.26: Ownership of the utility Company Private Owned

Transmission

Cooperatively Owned Utility Investor Owned Utility

Regional Transmission Organization/ Independent System Operator

Transmission

3.2.1

Government Owned Federal Utility Municipal Power Agency Municipally Owned Utility State Utility Political Subdivision

Provider Curtailment Service Provider Retail Power Marketer Wholesale Power Marketer

Hypotheses H5

Van den Bulte (2004) uncovered interesting evidence about diffusion speed in a metaanalysis of “diffusion speed” research. This study provides product managers and developers with a useful quantitative tool to measure diffusion for products and regions. The present study shows that some states and utility companies, who have adopted smart meters later in the game, have certain advantages because earlier adopters have discovered and already solved set-up issues. This study includes innovator states and companies that are expected to be early adopters

125

of AMI. For instance, California is expected to be one of the leaders in the implementation of smart meters. However, it is important to emphasize the number of meters installed in that state, such that we can see if the diffusion is significant by state or by utility companies’ ownership. Thus, we propose the following hypothesis: H5a: The diffusion speed q/p of a cluster of geographically closer states will be significantly greater than those from other regions of the United States. Libai, Muller & Peres (2005) extended their research questions to explore responsive budgeting strategies in which firms dynamically allocate their marketing efforts according to developments in the market. Many other questions are still waiting to be answered and issues such as regulation (Stremersch & Lemmens, 2009), international competition, and the optimal marketing mix of growing international markets can be further explored. H5b: Utility companies’ ownership is a key factor in the deployment of smart meters in the US. We expect that states with higher urban concentrations will be the first to have AMI implemented, as it is much more convenient for the utility companies to have these meters installed close by— so that they can monitor them constantly and provide better feedback to consumers. If this expectation is correct, the following hypothesis can be supported: H5c: States with the larger urban concentration will be the first ones to implement AMI. 3.2.2

Implementation Progress

For this study we use secondary data sources from the US 2010 population census to document the urban concentration indexes for a classification that seems to make sense.

126

3.2.3

Bass’ Diffusion Model for AMI/AMR

The first step of the process is to calculate the p, q and m diffusion speed using the Bass model and the differential solution proposed by Boswijk and Franses (equation 4), as shown in section 2.6.2. Using the Excel Solver software, the curves were approximated to non-linear distributions, following Bass’ diffusion model formula. Once the values of these parameters are obtained, we calculate the ratio imitators/innovators of AMI diffusion for the hypotheses testing. 3.2.3.1

Geographical Clusters of States Table 3.27: Urban concentration indexes for the US

inhabitants inhabitants /sq. mile / km2 District of Columbia 10,357.0 3,999.0 New Jersey 1,205.0 465.0 Rhode Island 1,016.0 392.0 Massachusetts 852.1 329.0 Connecticut 741.4 286.3 Maryland 606.2 234.1 Delaware 470.7 181.7 New York 415.3 160.3 Florida 360.2 139.1 Pennsylvania 285.3 110.2 California 282.5 109.1 Ohio 244.2 94.3 Illinois 231.9 89.5 Hawaii 216.8 83.7 Virginia 207.3 80.0 North Carolina 200.6 77.5 Indiana 182.5 70.5 Michigan 174.8 67.5 Georgia 172.5 66.6 South Carolina 157.1 60.7 Tennessee 156.6 60.5 New Hampshire 147.0 56.8 Kentucky 110.0 42.5 Wisconsin 105.2 40.6 Louisiana 105.0 40.5 Washington 102.6 39.6

inhabitants inhabitants / /sq. mile km2 Texas 98.1 37.9 Alabama 94.7 36.5 Missouri 87.3 33.7 West Virginia 77.1 29.8 Vermont 67.7 26.2 Minnesota 67.1 25.9 Mississippi 63.5 24.5 Arizona 57.1 22.0 Arkansas 56.4 21.8 Oklahoma 55.2 21.3 Iowa 54.8 21.2 Colorado 49.3 19.1 Maine 43.0 16.6 Oregon 40.3 15.6 Kansas 35.1 13.6 Utah 34.3 13.2 Nevada 24.8 9.6 Nebraska 24.0 9.3 Idaho 19.2 7.4 New Mexico 17.2 6.6 South Dakota 10.9 4.2 North Dakota 9.9 3.8 Montana 6.9 2.6 Wyoming 5.9 2.3 Alaska 1.3 0.5

State

State

127

Following the representation, as advocated in Boswijk and Franses (2005), the estimation of diffusion speeds for the states is first calculated. Using the resulting diffusion speed for each state, we make the categorization on the clusters included in our sample as geographically located (West, South, Midwest & Northeast). Accordingly, this distinction is shown in Table 3.28. The clustering was used to observe actual diffusion curves in each case (Figure 3.21). Looking at the cluster results in Table 3.28; it is evident that the South cluster is the fastest growing sector, with a much higher diffusion rate (q/p), followed by the west, while the Midwest and Northeast are well behind. The imitation index (q) flattens the curve so for the South and West the q-indexes are much higher, meaning that the curves grow up quickly and flatten, while for the other 2 clusters, the curves slowly grow continuously. This analysis shall prove the hypothesis H5a about a cluster having higher diffusion speed, being faster than others.

Figure 3.21: Diffusion Speeds of AMI for different states clusters Probing deeper into the cluster analysis, we can see in Table 3.30 that the GDP averages of all states included in every cluster are very close. Conducting a t-test for the clusters, the result 128

shows that only the urban concentration of inhabitants per square-mile is not significant comparing the states in the cluster. We will be cautious about this factor in the analysis. Table 3.28: Cluster Totals for Diffusion Cluster Northeast Midwest South West

Cum. Level (Millions) 5.038 1.328 18.258 20.107

Maturity (Millions) 23.854 27.550 52.441 36.405

Innovation Imitation 0.0127 0.0467 0.0093 0.0598 0.0044 0.5236 0.0094 0.4629

Diffusion Rate 3.67 6.45 118.52 49.06

Table 3.30 shows the averages of all the states included in every cluster. Analyzing the data, we can conclude that the imitation index of the Midwest is the highest one, but in general the cluster in total is very low for q. Therefore we need to analyze the standard deviations to study dispersion within the cluster. Table 3.29: t-test Results for Clusters Test Value = 0

Maturity Level (m) Innovation Index (p) Imitation Index (q) Population Millions GDP per capita Inhabitants per sq. mile

t 16.517 6.307 3.802 25.376 32.625 1.045

df 3 3 3 3 3 3

Sig. (2tailed) .000 .008 .032 .000 .000 .373

Mean Difference 2.69793E6 .0067000 .4255000 6.00925 $48,872.750 136.91850

95% Confidence Interval of the Difference Lower Upper 2.1781E6 3.2178E6 .003320 .010080 .069319 .781681 5.2556 6.7629 $44,105.38 $53,640.12 -280.0801 553.9171

Table 3.31 shows that the Midwest imitation index q is the factor with the lowest standard deviation. Therefore, we can expect to find very different states within each cluster that bias the average of the states within the cluster. While doing the t-test (Table 3.32) we discovered that although the GDP averages are equivalent, the standard deviation is not, so we 129

have to be careful while analyzing this factor too. The diffusion indexes and level are equivalent; therefore we validate the analysis of a higher diffusion rate in the South cluster compared to all others. The next step is to analyze the geographical clusters moving from four-regions into ninedivisions, to determine if by further selecting closer states we can find differences. Table 3.30: Average statistics by States’ Clusters Average Maturity Level (m) Innovation Index (p) Imitation Index (q) Population (Millions) GDP per capita Inhabitant/sq. mile

Northeast Midwest South West 2,833,187 2,623,135 3,051,408 2,283,995 0.0057 0.0093 0.0044 0.0074 0.5853 0.1058 0.5754 0.4355 6.144 5.575 6.618 5.700 $ 51,052 $ 45,626 $ 51,759 $ 47,054 530 3.583 5.706 8.385

Table 3.31: Standard Deviation statistics by States’ Clusters Standard Deviation Maturity Level (m) Innovation Index (p) Imitation Index (q) Population (Millions) GDP per capita Inhabitant/sq. mile

Northeast Midwest South West 2,717,328 1,996,887 2,908,508 3,769,678 0.0146 0.0106 0.0042 0.0078 0.7223 0.1742 0.5573 0.5063 6.476 4.065 6.482 9.696 $ 8,253 $ 4,448 $ 32,997 $ 10,179 435 0.515 0.849 0.506

Table 3.32: Standard Deviation t-test for States’ Clusters Test Value = 0

T Maturity Level 7.813 Innovation Index 4.229 Imitation Index 4.264 Population Millions 5.782 GDP per capita 2.165 Inhabitants per sq. mile 1.006

df 3 3 3 3 3 3

Sig. (2- Mean tailed) Difference .004 2.84810E6 .024 .0093000 .024 .4900250 .010 6.67975 .119 $13,969.250 .389 109.21750 130

95% Confidence Interval of the Difference Lower Upper 1.6880E6 4.0082E6 .002301 .016299 .124288 .855762 3.0032 10.3563 $-6,568.24 $34,506.74 -236.3777 454.8127

Table 3.33: Diffusion Results by States’ Division

Division New England Mid-Atlantic East North Central West North Central South Atlantic East South Central West South Central Mountain Pacific

Cum. Level Maturity Level Innovation (Millions) (Millions) Index 1.174 5.348 0.0010 3.864 18.724 0.0133 1.236 22.024 0.0084 0.092 6.059 0.0069 9.865 27.320 0.0024 1.569 9.061 0.0091 6.824 16.060 0.0056 2.889 16.626 0.0089 17.218 19.779 0.0058

Imitation Index 0.6341 0.1086 0.6007 0.3014 0.5696 0.2886 0.6802

Diffusion Speed 650.82 12.99 254.69 33.13 102.46 32.27 116.49

Looking at the cluster results in Table 3.33, it is evident that the New England Division is the fastest growing area, with a much higher diffusion rate (q/p), followed by the South Atlantic, the Pacific and West South Central. The imitation index (q) flattens the curve, so the Pacific and West South Central Divisions are the ones with the higher q-indexes, along with New England.

Figure 3.22: Diffusion Curves by Cluster of States by Division

131

The Pacific and West South Central Divisions include the top 2 states on amount of installed smart meters: California and Texas. In New England, the states of Maine and Vermont are really diffusing AMI in a short time, so their statistics look very good. Table 3.34: t-test Results for Divisions Test Value = 0

Maturity Level (m) Innovation Index (p) Imitation Index (q) Population (Millions) GDP per capita Inhabitants per sq. mile

T 5.294 5.186 3.854 5.169 6.721 2.291

df 8 8 8 8 8 8

Sig. (2tailed) .001 .001 .005 .001 .000 .051

Mean Difference 3.03358E6 .0070603 .3921979 6.78515 $277,620.889 342.60350

95% C.I. of the Difference Lower Upper 1.7122E6 4.3549E6 .003921 .010200 .157540 .626856 3.7584 9.8119 $182,369.88 $372,871.90 -2.2107 687.4177

It is important to notice that the Pacific Division is very close to complete diffusion, with 87%, West South Central follows with 42%, while all others fall below 40% implementation. The Mid-Atlantic and West North Central have a q-index of zero, representing a lack of imitation. That is, only innovators are implementing AMI in these Divisions. Table 3.35: Average statistics by States’ Clusters Average

New England

MidAtlantic

East North West North Central Central

South Atlantic

East South West South Mountain Central Central

Pacific

Maturity Level (m) 1,250,566 5,998,428 4,467,105 1,306,014 3,123,872 1,896,334 4,043,438 1,203,160 4,013,333 Innovation Index (p) 0.0011 0.0149 0.0078 0.0103 0.0038 0.0058 0.0043 0.0063 0.0093 Imitation Index (q) 0.8779 0.0000 0.1748 0.0565 0.6745 0.2652 0.6629 0.5235 0.2946 Population (Millions) 2.400 13.633 9.280 2.929 6.644 4.100 9.075 3.025 9.980 GDP per capita $ 300,249 $ 159,223 $ 215,253 $ 332,257 $ 544,825 $ 150,798 $ 184,286 $ 348,576 $ 263,121 Inhabitant/sq. mile 478 635.200 187.720 41.292 1,400.962 106.188 78.680 26.812 128.699

The worst cases are New Jersey, New York (Mid-Atlantic), Iowa, Minnesota and the Dakotas (West North Central Div.) Probing deeper, we can see in Table 3.35 that all variable

132

averages of all divisions are very close. By conducting a t-test; the results show that all the variables are significant, with urban concentration significant with 5.1%. Table 3.36: Standard Deviation statistics by States’ Divisions New MidEast North West North South East South West South Mountain Pacific England Atlantic Central Central Atlantic Central Central Maturity Level (m) 1,051,759 2,056,161 1,490,771 985,627 2,846,423 986,584 4,408,064 748,542 5,964,579 Innovation Index (p) 0.0026 0.0254 0.0110 0.0111 0.0047 0.0050 0.0030 0.0072 0.0091 Immitation Index (q) 0.7256 0.0000 0.2423 0.0984 0.6866 0.3096 0.3679 0.5464 0.4541 Population (Millions) 2.263 5.361 3.092 2.075 5.745 2.235 10.703 1.860 15.452 GDP per capita $ 9,338 $ 6,729 $ 4,695 $ 3,479 $ 44,085 $ 1,445 $ 9,188 $ 10,388 $ 7,677 Inhabitant/sq. mile 440 497.724 55.104 29.476 3,362.766 38.778 26.546 18.788 118.476 Standard Deviation

Table 3.35 shows the averages of all the states included in every division. Analyzing the data, we can see that the imitation index of the New England is the highest one, followed by the West South Central and Mountain Divisions. Mountain is heavily pushed by Nevada, which is doing a quick rollout of AMI. Table 3.36 presents the standard deviation of the statistics from every Division. It is evident that there is a big spread of the AMI diffusion and general statistics from the Divisions. Most q-indexes standard deviations are high, with the exceptions of MidAtlantic and West North Central that tend to zero, as there are no imitators but only innovators implementing AMI in these Divisions. Table 3.37: Statistics from the Individual States States Stats CA TX FL GA PA AZ AL OR MI OK Maturity Level (m) 14,472,050 10,643,823 9,596,693 3,887,502 6,223,840 2,701,242 2,190,337 1,721,751 4,529,777 1,813,281 Innovation Index (p) 0.0013 0.0045 0.0002 0.0006 0.0442 0.0118 0.0118 0.0238 0.0023 0.0068 Immitation Index (q) 1.0643 0.6915 1.0425 1.1883 0.0000 0.5869 0.5869 0.3503 0.4905 0.5029 Population (Millions) 37.300 25.100 18.800 9.700 12.700 6.400 4.800 3.800 9.900 3.800 GDP per capita $ 51,914 $ 58,099 $ 40,106 $ 41,711 $ 45,323 $ 40,828 $ 36,333 $ 44,447 $ 37,616 $ 42,237 Inhabitant/sq. mile 283 98.070 360.200 172.500 285.300 57.050 94.650 40.330 174.800 55.220

It is important to note the outlier of urban concentration in DC, where the number of inhabitants per square mile is 10,357, which biases the statistics—thus making it insignificant for

133

this data. Performing a t-test (Table 3.38), we confirm that the standard deviation for urban concentration is not significant, so we remain careful in regard to this factor. Table 3.38: Standard Deviation t-test for States’ Divisions Test Value = 0

Maturity Level (m) Innovation Index (p) Imitation Index (q) Population (Millions) GDP per capita Inhabitant per sq. mile

T 3.776 3.757 4.601 3.447 2.520 1.408

df 8 8 8 8 8 8

Sig. (2tailed) .005 .006 .002 .009 .036 .197

Mean Difference 2.28206E6 .0087781 .3812257 5.42067 $10,780.584 509.68542

95% C.I. of the Difference Lower Upper 888,327.1944 3.6758E6 .003390 .014166 .190145 .572307 1.7942 9.0471 $916.38 $20,644.79 -324.9295 1344.3004

Table 3.39: Top 10 States Diffusion Statistics State CA TX FL GA PA AZ AL OR MI OK

Cum. Level Maturity Level Innovation (Millions) (Millions) Index 12.101 14.472 0.001 5.692 10.644 0.004 5.882 9.597 0.000 2.460 3.888 0.001 3.863 6.224 0.044 0.862 2.701 0.012 1.491 2.190 0.012 0.833 1.722 0.024 1.090 4.530 0.002 0.804 1.813 0.007

Imitation Index 1.0643 0.6915 1.0425 1.1883 0.5869 0.5869 0.3503 0.4905 0.5029

Diffusion Speed 834.77 154.29 5,052.94 1,987.45 49.74 49.74 14.70 209.00 73.69

Some noteworthy comments from Table 3.39 are that Pennsylvania shows growth following the exponential distribution, because q equals zero, while Florida and Georgia show minimum innovation indexes with high imitation indexes, such that p equals zero; the diffusion reduces to the logistic distribution. In other words, Pennsylvania is showing a flat behavior. Figure 3.23 shows the important growth of AMI in the US thanks to the states of California, 134

Texas, Florida and Georgia. They are expected to reach maturity from 2019 through 2022. Pennsylvania is continuously growing but it will not reach maturity until around the year 2144.

Figure 3.23: Diffusion curves for the Top Ten States

3.2.3.2

Utility Company Ownership

Looking at the results in Table 3.40, it is evident that the Government Owned utilities are the ones diffusing AMI faster than the Privately Owned utilities so can support hypothesis H5b stating that utility company ownership is important for the deployment of smart meters. Table 3.40: Diffusion Statistics by Company Ownership

Privately Owned Government Owned Providers Transmission

Cumm. Level Maturity Level Innovation Imitation Diffusion (Millions) (Millions) Index Index Speed 29.494 117.342 0.0034 0.5395 160.93 3.276 17.253 0.0027 0.5177 192.63 0.610 1.790 0.0000 1.5288 ∞ 0.000 0.003 0.0049 0.0000 -

135

Despite this fact, investors have implemented more than a quarter of the total number of meters, and the innovation and imitation indexes are higher for the private investors. Because the injection of ARRA funds in AMI has been substantial; therefore, we see important growth in government sector utilities. To further prove this point, we are going to separate the 4 categories into 7 more specific ownerships.

Figure 3.24: Diffusion Speeds of AMI for different states clusters Table 3.41: Diffusion Statistics for Ownership Categories Ownership Private Investors Cooperatives Municipalities Federal Utilities Political Subdivisions Power Marketer State Utility Curtailment Serv. Prov. Transmission

Cum. Level Maturity Level Innovation Imitation (Millions) (Millions) Index Index 25.533 105.572 0.0010 0.7446 3.961 11.769 0.0617 1.448 11.015 0.0020 0.4862 0.971 4.164 0.0010 0.7324 0.815 2.277 0.0104 0.4348 0.604 1.784 0.0000 1.5320 0.043 0.491 0.0000 1.2769 0.006 0.006 0.0000 1.9050 0.000 0.003 0.0049 -

136

Diffusion Speed 757 242 713 42 153,200 125,276 190,505 -

Figure 3.25: Diffusion curves by Utility Company Ownership We can easily see in Table 3.41 that Cooperatives have an imitation index q equal to zero, that is, they have no imitators, but only innovators. On the other hand, Power Marketers, State Utilities, and CSP have p approaching zero; therefore, the distribution is shifting into the future. We can prove that utilities diffusing more AMI are Federal Utilities, due to ARRA investments.

3.2.3.3

Urban Concentration Analysis

In Table 3.42, we can see that states with an urban concentration of 501 to 1500 inhabitants per square mile do not have imitators, but only innovators. Among these states are: Connecticut, Maryland, Massachusetts, New Jersey and Rhode Island. States with concentrations between 251 and 500 are the ones with the highest imitation index— the states in this category

137

are California and Florida. Due to the large geographical extension, Texas is in the less than 100 category.

Figure 3.26: Diffusion Speeds for AMI depending on Urban Concentration Table 3.42: Diffusion Statistics for Urban Concentration

Concentr.