Colin Foote PHD Thesis March 2007

New Methods for New Challenges in Electricity
Distribution Network Planning
Colin E. T. Foote
MEng
Submitted for the Degree

of
Doctor of Philosophy
Institute for Energy and Environment

Department of Electronic and Electrical Engineering
University of Strathclyde
Glasgow G1 1XW
United Kingdom
March 2007
The copyright of this thesis belongs to the author under the terms of the United
Kingdom Copyright Acts as qualified by University of Strathclyde Regulation 3.50.
Due acknowledgement must always be made of the use of any material contained in,
or derived from, this thesis.
i
Acknowledgements
I would like to thank Dr Graeme Burt and Professor Jim McDonald, my academic
supervisors, for providing guidance, support, encouragement and resources for this
research together with all manner of other opportunities.
I would also like to thank the rest of my colleagues within the Institute for Energy
and Environment at the University of Strathclyde who provided such a stimulating
and vibrant working environment. In particular, I extend thanks to Dr Graham Ault
for his guidance and encouragement.
I extend my gratitude to the organisations that have provided funding and technical
support for this research, namely East Midlands Electricity, ScottishPower, Rolls-
Royce, the European Commission and the Department for Trade and Industry.
Involvement in various projects has allowed me to meet people from a variety of
countries and organisations and many of them have contributed to this research
through discussion and insight. I thank them all for their contribution.
I would also like to thank my family, my friends and Lesley for all their support and
encouragement.
ii
Abstract
The planning and design of electricity distribution networks is facing new challenges
as new technologies are introduced and new demands are placed on the ageing assets
in developed countries around the world. In particular, the expansion of distributed
generation presents a range of challenges that distribution network planners must
resolve. This research offers a new perspective on the conventional approach to
distribution network planning and presents a comprehensive assessment of the new
challenges facing planners in the 21st century. A number of specific shortcomings in
the conventional approach are identified.
It is argued that the application of methods from other domains and the development
of new tools can help explore the shortcomings and update and improve the
conventional approach. Engineering design theory, decision support methods,
information management, scenario analysis, and dynamic modelling for power
system simulation have all been investigated and novel contributions have been made
in a number of areas to offer a valuable contribution to the development of electricity
distribution network planning.
Case studies and examples are used to examine distributed generation in detail,
compare its use as an alternative to conventional options, and assess its likely level of
penetration in future networks. Detailed dynamic studies of networks with wind
farms explore some of the most important issues of concern to grid operators.
Together, the application of methods from various domains show how the
conventional approach to electricity distribution network planning can be enhanced
so that distribution network operators are able to meet the challenges of the 21st
century.
iii
Table of Contents
Acknowledgements ...................................................................................................... ii
Abstract .......................................................................................................................iii
Table of Contents ........................................................................................................ iv
List of Figures ............................................................................................................. ix
List of Tables............................................................................................................. xvi
Abbreviations ..........................................................................................................xviii
1. Introduction .............................................................................................................. 1
1.1. Aim.................................................................................................................... 1
1.2. Principal Contributions ..................................................................................... 1
1.3. Publications ....................................................................................................... 4
2. Conventional Perspective on Distribution Network Planning ................................. 7
2.1. The Distribution Network Planning “Problem”................................................ 7
2.2. Model of the Conventional Approach to Distribution Network Planning ........ 8
2.2.1. Knowledge Modelling Methodology ......................................................... 8
2.2.2. Knowledge Models .................................................................................... 9
2.3. The Complete Model ...................................................................................... 26
2.4. Review of Chapter........................................................................................... 30
2.5. Chapter References ......................................................................................... 30
3. Drivers and New Directions in Distribution Networks.......................................... 33
3.1. Technology Drivers in Distribution Networks................................................ 34
3.1.1. Distributed Generation ............................................................................. 34
3.1.2. Energy Storage ......................................................................................... 38
3.1.3. Demand Side Technology Developments................................................ 39
3.1.4. Power Electronics..................................................................................... 39
3.1.5. Communications and Control Technologies............................................ 39
3.2. Commercial and Regulatory Drivers in Distribution Networks ..................... 40
3.3. Future Directions in Distribution Networks.................................................... 42
3.3.1. Network Architectures and Operation ..................................................... 42
3.3.2. Institutional and Organisational Changes ................................................ 46
iv
4. Shortcomings in the Conventional Approach and the Need for New Methods..... 53
4.1. The Use of Conventional Technologies and Methods .................................... 54
4.2. The Need for More Analysis........................................................................... 55
4.3. Organisational Separation and Loss of Control .............................................. 57
4.4. Difficulties in Formulating Strategies and Making Decisions........................ 58
4.5. The Need for Knowledge Management .......................................................... 59
5. Engineering Design Theory ................................................................................... 63
5.1. Solution-Neutral Problem Definition.............................................................. 64
5.2. Level of Risk and Innovation.......................................................................... 65
5.3. Decision Classifications .................................................................................. 65
5.4. Design Concurrency........................................................................................ 66
5.5. Alternative Designs......................................................................................... 66
5.6. Design Rationale ............................................................................................. 67
5.6.1. Structured Decision Making Methods ..................................................... 70
5.6.2. Generic Justifications ............................................................................... 71
5.6.3. Knowledge Modelling Methods............................................................... 72
6. Decision Support in Distribution Network Planning ............................................. 76
6.1. Multiple Criteria Decision Making ................................................................. 76
6.1.1. General Structure for MCDM .................................................................. 78
6.1.2. Calculation of Alternative Decision Ratios ............................................. 82
6.2. MCDM Case Study......................................................................................... 83
6.2.1. Development Issues ................................................................................. 84
6.2.2. Development Options .............................................................................. 85
6.2.3. Quantification........................................................................................... 86
6.2.4. Analysis.................................................................................................... 89
6.2.5. Decision Making ...................................................................................... 91
6.2.6. Sensitivity Analysis.................................................................................. 97
v
6.3. Review of Chapter......................................................................................... 107
6.4. Chapter References ....................................................................................... 108
7. Information Management in Distribution Network Planning .............................. 111
7.1. Novel Combination of Methods to Assist in Collecting Information ........... 112
7.1.1. Structured Approach .............................................................................. 113
7.1.2. Standard Formats ................................................................................... 113
7.1.3. Generic Data........................................................................................... 114
7.2. Representation and Exchange of Power System Models and Data............... 115
7.2.1. Models and Data in Power Systems....................................................... 116
7.2.2. Horizontal and Vertical Exchange of Power System Models and Data. 117
7.2.3. Representation and Exchange of Power System Data ........................... 118
7.2.4. Representation and Exchange of Power System Models....................... 119
7.3. Information Management for Distributed Generation .................................. 120
7.3.1. The Need for Information on DG........................................................... 121
7.3.2. A Structured Approach for Collecting Information on DG ................... 123
7.3.3. Examples of Distributed Generation Analysis....................................... 135
8. Scenario Analysis................................................................................................. 140
8.1. Methods for Managing Uncertainty and Risk............................................... 141
8.1.1. Scenario Analysis................................................................................... 141
8.1.2. Decision Trees........................................................................................ 142
8.1.3. Sensitivity Analysis................................................................................ 143
8.1.4. Probabilistic Choice Versus Risk Analysis............................................ 143
8.1.5. Flexibility as a Means of Dealing with Uncertainty .............................. 144
8.2. Distributed Generation Penetration Scenarios .............................................. 144
8.2.1. Methodology for Penetration Assessment ............................................. 145
8.2.2. Top-Down Approach ............................................................................. 146
8.2.3. Bottom-Up Approach............................................................................. 151
8.2.4. Combining Top-down and Bottom-up Results ...................................... 155
vi
9. Dynamic Modelling of Wind Farms .................................................................... 162
9.1. List of Symbols ............................................................................................. 164
9.2. PSS/E Requirements for Generator Modelling ............................................. 165
9.3. A Dynamic Model of a Wind Farm .............................................................. 165
9.3.1. Wind Speed Model................................................................................. 166
9.3.2. Aerodynamic Model............................................................................... 167
9.3.3. Mechanical Systems............................................................................... 168
9.3.4. Generator and Grid Interface ................................................................. 172
9.3.5. Control Systems ..................................................................................... 178
Power Quality ...................................................................................................... 182
9.3.6. Ancillary Systems .................................................................................. 191
9.3.7. Protection ............................................................................................... 192
9.3.8. Wind Farm Electrical Network .............................................................. 193
9.3.9. Electricity Network ................................................................................ 193
9.4. Model Validation or Verification.................................................................. 194
9.5. Case Studies .................................................................................................. 194
9.5.1. Aggregate Models of Wind Farms............................................................. 194
9.5.2. Study of the Effect of a High Penetration of Wind Farms......................... 197
10. Conclusions ........................................................................................................ 205
Appendix A. DG Penetration Survey Results .......................................................... 212
A.1. Scenario 1. Residential Ring, Germany ....................................................... 213
A.2. Scenario 2. Commercial Mesh, Germany .................................................... 214
A.3. Scenario 3. Mixed Radial, Germany ............................................................ 215
A.4. Scenario 4. Urban Meshed, UK ................................................................... 216
A.5. Scenario 5. Rural, Poland............................................................................. 217
A.6. Scenario 6. Urban Link, Poland ................................................................... 218
A.7. Scenario 7. Urban Radial, France................................................................. 219
A.8. Scenario 8. Rural, Italy................................................................................. 220
A.9. Scenario 9. Urban Radial, Italy .................................................................... 221
A.10. Scenario 10. Urban Link, Greece ............................................................... 222
vii
A.11. Scenario 11. Rural South Coast, Spain ...................................................... 223
A.12. Scenario 12. Rural Link, Netherlands ........................................................ 224
A.13. Scenario 13. Rural Ring, Belgium ............................................................. 225
A.14. Scenario 14. Urban Ring, Denmark ........................................................... 226
A.15. Scenario 15. Rural Ring, Austria ............................................................... 227
Appendix B. Aggregate Models of a Wind Farm .................................................... 228
B.1. Methodology for Study of Aggregate Models of a Wind Farm ................... 228
B.1.1. Test Networks........................................................................................ 228
B.1.2. Test Models ........................................................................................... 230
B.1.3. Test Procedure ....................................................................................... 231
B.2. Results for Study of Aggregate Models of a Wind Farm............................. 232
B.2.1. Study Set One ........................................................................................ 232
B.2.2. Study Set Two ....................................................................................... 235
B.2.3. Study Set Three ..................................................................................... 237
B.2.4. Study Set Four ....................................................................................... 240
Appendix C. The Effects of a High Penetration of Wind Farms ............................. 244
C.1. Methodology for Study of the Effects of a High Penetration of Wind Farms
.............................................................................................................................. 244
C.1.1. Test Network ......................................................................................... 244
C.1.2. Test Models ........................................................................................... 245
C.1.3. Test Scenarios........................................................................................ 249
C.1.4. Test Procedures ..................................................................................... 250
C.2. Results for Study of the Effects of a High Penetration of Wind Farms ....... 251
viii
List of Figures
Figure 2.1 – Symbols used in knowledge models........................................................ 9

Figure 2.2 – Task Model: Plan and design the distribution network ......................... 11
Figure 2.3 – Task Model: Identify requirements ....................................................... 12
Figure 2.4 – Task Model: Consider goals and responsibilities .................................. 13
Figure 2.5 – Task Model: Consider development drivers.......................................... 14
Figure 2.6 – Task Model: Consider new connections................................................ 15
Figure 2.7 – Task Model: Consider general load growth........................................... 16
Figure 2.8 – Task Model: Forecast load..................................................................... 16
Figure 2.9 – Task Model: Identify insecurities and exceeded ratings........................ 17
Figure 2.10 – Task Model: Consider asset replacement or refurbishment ................ 18
Figure 2.11 – Task Model: Assess condition of plant................................................ 19
Figure 2.12 – Task Model: Identify redundant network ............................................ 19
Figure 2.13 – Task Model: Consider network performance ...................................... 20
Figure 2.14 – Task Model: Combine requirements from different considerations.... 21
Figure 2.15 – Task Model: Design network additions and alterations ...................... 22
Figure 2.16 – Task Model: Determine substation, feeder and ancillary equipment
details ................................................................................................................. 23
Figure 2.17 – Task Model: Determine programmes of implementation.................... 24
Figure 2.18 – Task Model: Assess network access constraints ................................. 25
Figure 2.19 – Task Model: Revise policies, procedures and standards ..................... 25
Figure 2.20 – The complete model of the conventional approach, part 1 of 2 .......... 27
Figure 2.21 – The complete model of the conventional approach, part 2 of 2 .......... 28
Figure 2.22 – Text-only version of the complete model of the conventional approach
............................................................................................................................ 29
Figure 6.1 – General structure for multiple criteria decision making ........................ 79
Figure 6.2 – Impact of Varying Equipment Cost on Benefit / Cost Ratio ................. 98
Figure 6.3 – Impact of Assessment Lifetime on Benefit / Cost Ratio ....................... 99
Figure 6.4 – Impact of Varying Gas Cost on Benefit / Cost Ratio .......................... 100
Figure 6.5 – Impact of Varying Diesel Cost on Benefit / Cost Ratio ...................... 101
ix
Figure 6.6 – Impact of Varying Interruptions / Year Criterion Weight on Benefit /
Cost Ratio......................................................................................................... 103
Figure 6.7 – Impact of Varying Power Quality Criterion Weight on Benefit / Cost
Ratio ................................................................................................................. 104
Figure 6.8 – Impact of Varying Standalone Capability Criterion Weight on Benefit /
Cost Ratio......................................................................................................... 105
Figure 6.9 – Impact of Varying Visual Environmental Impact Criterion Weight on
Benefit / Cost Ratio.......................................................................................... 106
Figure 7.1 – Varying content and representation in power system models and data118
Figure 7.2 – Simple example of a possible operating chart for a distributed generator
.......................................................................................................................... 127
Figure 7.3 – Simple example of a possible response rate chart for a distributed
generator........................................................................................................... 128
Figure 7.4 – Simple example of a possible daily cost curve for a distributed generator
.......................................................................................................................... 129
Figure 9.1 – Components of a complete wind farm dynamic model ....................... 166
Figure 9.2 – Typical power versus wind speed characteristics for an 800kW wind
turbine .............................................................................................................. 168
Figure 9.3 - Two-mass representation of a wind turbine shaft ................................ 169
Figure 9.4 – Performance coefficient as a function of tip speed ratio with pitch angle
as a parameter................................................................................................... 180
Figure 9.5 – Response of electromagnetic torque to changes in rotor voltages....... 186
Figure 9.6 – Cascaded PI controller for rotor quadrature current ............................ 187
Figure 9.7 – Response of power and reactive power to changes in direct axis rotor
voltage .............................................................................................................. 189
Figure 9.8 – Simple PI control system for reactive power....................................... 190
Figure 9.9 – Real power in branch 102-103 for the three categories of study set one
.......................................................................................................................... 196
Figure 9.10 – Bus angle at bus 103 for the three categories of study set three........ 197
Figure 9.11 – Bus 1 voltage for medium load condition and different wind farm
conditions ......................................................................................................... 198
x
Figure 9.12 – Bus 30 voltage for low load condition and different wind farm
scenarios (different MW base levels)............................................................... 199
Figure 9.13 – Bus 30 wind farm output for low load condition and different wind
farm scenarios (different MW base levels) and the adjusted pattern of input
power................................................................................................................ 200
Figure B.1 – Test network for category A studies ................................................... 229
Figure B.2 – Test network for category B studies ................................................... 229
Figure B.3 – Test network for category C studies ................................................... 230
Figure B.4 – Real power in branch 102-103 for the three categories of study set one
.......................................................................................................................... 232
Figure B.5 – Reactive power in branch 102-103 for the three categories of study set
one .................................................................................................................... 233
Figure B.6 – Bus voltage at bus 103 for the three categories of study set one ........ 234
Figure B.7 – Bus angle at bus 103 for the three categories of study set one ........... 234
Figure B.8 – Real power in branch 102-103 for the three categories of study set two
.......................................................................................................................... 235
two.................................................................................................................... 236
Figure B.10 – Bus voltage at bus 103 for the three categories of study set two ...... 236
Figure B.11 – Bus angle at bus 103 for the three categories of study set two ......... 237
Figure B.12 – Real power in branch 102-103 for the three categories of study set
three.................................................................................................................. 238
three.................................................................................................................. 238
Figure B.14 – Bus voltage at bus 103 for the three categories of study set three .... 239
Figure B.15 – Bus angle at bus 103 for the three categories of study set three ....... 240
Figure B.16 – Real power in branch 102-103 for the three categories of study set four
.......................................................................................................................... 241
four ................................................................................................................... 241
Figure B.18 – Bus voltage at bus 103 for the three categories of study set four ..... 242
Figure B.19 – Bus angle at bus 103 for the three categories of study set four ........ 243
xi
Figure C.1 – Diagram of 30-bus network used in wind farm studies ...................... 244
Figure C.2 – Bus 1 voltage for no wind farms and different load conditions.......... 252
Figure C.3 – Bus 30 voltage for no wind farms and different load conditions........ 253
Figure C.4 – Bus 1 voltage for 0MW wind farm outputs and different load conditions
.......................................................................................................................... 254
Figure C.5 – Bus 30 voltage for 0MW wind farm outputs and different load
conditions ......................................................................................................... 255
.......................................................................................................................... 256
conditions ......................................................................................................... 257
conditions ......................................................................................................... 258
conditions ......................................................................................................... 259
conditions ......................................................................................................... 260
conditions ......................................................................................................... 261
conditions ......................................................................................................... 262
conditions ......................................................................................................... 263
conditions with high protection setting ............................................................ 264
conditions with high protection setting ............................................................ 265
Figure C.16 – Bus 1 voltage for low load condition and different wind farm
conditions ......................................................................................................... 266
conditions ......................................................................................................... 267
xii
Figure C.18 – Bus 1 voltage for medium load condition and different wind farm
conditions ......................................................................................................... 268
conditions ......................................................................................................... 269
Figure C.20 – Bus 1 voltage for high load condition and different wind farm
conditions ......................................................................................................... 270
conditions ......................................................................................................... 271
Figure C.22 – Bus 30 wind farm output for wind farm scenario 0 (0MW base level)
and different load conditions............................................................................ 272
Figure C.23 – Bus 2 generator output for wind farm scenario 0 (0MW base level) and
different load conditions .................................................................................. 273
Figure C.24 – Bus 30 voltage for wind farm scenario 0 (0MW base level) and
Figure C.25 – Bus 2 voltage for wind farm scenario 0 (0MW base level) and
Figure C.26 – Bus 30 wind farm output for wind farm scenario A (5MW base level)
Figure C.27 – Bus 2 generator output for wind farm scenario A (5MW base level)
Figure C.28 – Bus 30 voltage for wind farm scenario A (5MW base level) and
Figure C.29 – Bus 2 voltage for wind farm scenario A (5MW base level) and
Figure C.30 – Bus 30 wind farm output for wind farm scenario B (10MW base level)
Figure C.31 – Bus 2 generator output for wind farm scenario B (10MW base level)
Figure C.32 – Bus 30 voltage for wind farm scenario B (10MW base level) and
Figure C.33 – Bus 2 voltage for wind farm scenario B (10MW base level) and
xiii
Figure C.34 – Bus 30 wind farm output for wind farm scenario C (15MW base level)
Figure C.35 – Bus 2 generator output for wind farm scenario C (15MW base level)
Figure C.36 – Bus 30 voltage for wind farm scenario C (15MW base level) and
Figure C.37 – Bus 2 voltage for wind farm scenario C (15MW base level) and
Figure C.38 – Bus 30 wind farm output for wind farm scenario D (20MW base level)
Figure C.39 – Bus 2 generator output for wind farm scenario D (20MW base level)
Figure C.40 – Bus 30 voltage for wind farm scenario D (20MW base level) and
Figure C.41 – Bus 2 voltage for wind farm scenario D (20MW base level) and
Figure C.42 – Bus 30 wind farm output for low load condition and different wind
farm scenarios (different MW base levels) ...................................................... 292
Figure C.43 – Bus 2 generator output for low load condition and different wind farm
Figure C.45 – Bus 2 voltage for low load condition and different wind farm scenarios
(different MW base levels)............................................................................... 295
Figure C.46 – Bus 30 wind farm output for low load condition and different wind
farm scenarios (different MW base levels) and the adjusted pattern of input
power................................................................................................................ 296
Figure C.47 – Bus 2 generator output for low load condition and different wind farm
scenarios (different MW base levels) and the adjusted pattern of input power297
scenarios (different MW base levels) and the adjusted pattern of input power298
xiv
Figure C.49 – Bus 2 voltage for low load condition and different wind farm scenarios
(different MW base levels) and the adjusted pattern of input power............... 299
Figure C.50 – Bus 30 wind farm output for medium load condition and different
wind farm scenarios (different MW base levels) ............................................. 300
Figure C.51 – Bus 2 generator output for medium load condition and different wind
Figure C.54 – Bus 30 wind farm output for high load condition and different wind
Figure C.55 – Bus 2 generator output for high load condition and different wind farm
xv
List of Tables
Table 6.1 – Initial Construction Cost Data................................................................. 87

Table 6.2 – Ongoing Cost Data.................................................................................. 87
Table 6.3 – Reliability Data ....................................................................................... 88
Table 6.4 – Analysis for the Single Gas Turbine Option........................................... 90
Table 6.5 – Summary of Performance Results for all Criteria and all Options ......... 91
Table 6.6 – Normalised Results for all Criteria and all Options................................ 92
Table 6.7 – Evaluation Criteria Weight Values ......................................................... 93
Table 6.8 – Results of Decision Analysis .................................................................. 94
Table 6.9 – Results of Environmental Cost Analysis ................................................ 96
Table 7.1 – Information Requirements for Characterisation of Distributed Generation
and Examples of Specific Device Information and Generic Category Data.... 124
Table 8.1 Electricity from renewable sources in 1997 and targets for 2010 for EU
countries ........................................................................................................... 146
Table 8.2 – DG Penetration Forecast for Italy ......................................................... 147
Table 8.3 – DG Penetration Forecast for Poland ..................................................... 148
Table 8.4 – DG Penetration Forecast for the UK..................................................... 149
Table 8.5 – Summary of DG penetration forecasts.................................................. 150
Table 8.6 – Penetration forecasts for LV connected DG ......................................... 151
Table 8.7 – Summary of LV Grid Scenarios............................................................ 152
Table 8.8 – Forecast DG Penetration Rankings in the LV Grid Scenarios for 2010154
Table 8.10 – Possible strategy for defining DG penetrations in Scenario 1 ............ 157
Table 9.1 – The four main types of wind turbine generator .................................... 172
Table A.1 – Questionnaire results for scenario 1..................................................... 213
xvi
Table A.10 – Questionnaire results for scenario 10................................................. 222
Table B.1 – Doubly fed induction generator parameters ......................................... 230
Table B.2 – Rotor voltage controller parameters ..................................................... 231
Table B.3 – Division of power between machines in test categories B and C ........ 240
Table C.1 – Models used in the study of high penetration of wind farms ............... 245
Table C.2 – Doubly fed induction generator parameters ......................................... 245
Table C.3 – Rotor voltage controller parameters ..................................................... 246
Table C.4 – Pattern of power input variations ......................................................... 246
Table C.5 – Salient pole generator parameters ........................................................ 247
Table C.6 – Round rotor generator parameters ........................................................ 248
Table C.7 – Turbine governor parameters ............................................................... 248
Table C.8 – Simple excitation system parameters ................................................... 248
Table C.9 – Load values for the three scenarios ...................................................... 249
Table C.10 – Wind farm scenarios........................................................................... 250
Table C.11 – New pattern of input power variations ............................................... 296
xvii
Abbreviations
AC Alternating Current
BTU British Thermal Unit
CHP Combined Heat and Power
CIM Common Information Model
DC Direct Current
DFIG Doubly Fed Induction Generator
DG Distributed Generation
DNO Distribution Network Operator
E2I Electricity Innovation Institute
EPRI Electric Power Research Institute
EU European Union
GIS Geographic Information System
GSP Grid Supply Point
GWh Gigawatt Hours
kVA Kilo Volt Amperes
kW Kilowatt
kWh Kilowatt Hours
LV Low Voltage (below 1kV)
MCDM Multiple Criteria Decision Making
MW Megawatts
MWe Megawatts Electrical
NMS Network Management System
O&M Operation and Maintenance
PV Photovoltaic
SMART Simple Multiple Attribute Rating Technique
TWh Terawatt Hours
XML Extensible Markup Language
xviii
1. Introduction
Electricity distribution networks are among the most important elements of large-
scale infrastructure that underpin modern society. The planning and design of these
networks is becoming increasingly complex as new technologies are introduced and
new demands are placed on the ageing assets in developed countries around the
world. The time is ripe to offer new methods that can support distribution planners
as they attempt to meet the new challenges in their domain.
1.1. Aim
The thesis of this work is that there are new challenges facing electricity distribution
networks in the 21st century and that an assessment of the conventional approach to
planning and design will identify a number of shortcomings. Furthermore, in
exploring the application of a range of methods from other domains and the
development of new tools, it will be possible to specify ways in which the
conventional approach can be updated to meet the new challenges. In particular, the
use of engineering design theory, decision support methods, information
management, scenario analysis, and the development of new models for simulation
and analysis are proposed as valuable. This research and its conclusions are timely
given the radical changes in electricity distribution networks being discussed across
the developed world. In particular, the expansion of distributed generation (DG)
presents a range of challenges that distribution network planners must resolve.
1.2. Principal Contributions
In the research reported here, the conventional approach to the planning and design
of electricity distribution networks is analysed and discussed using a modified
knowledge modelling methodology. This novel assessment of the domain is
followed by an up-to-date review of the drivers and new directions in distribution
network planning, with a particular emphasis on DG and its impact. Given the
review of the conventional approach and the assessment of drivers and new
1
directions, specific shortcomings in distribution network planning and design are
identified.
Following this review of conventional practice, new challenges and shortcomings, a

number of methods from other domains are examined to determine how they might
support future planning and design of electricity distribution networks.
Engineering design theory offers a range of concepts and methods that can be applied
to electricity distribution network planning to address the shortcomings in the
conventional approach and help meet the new challenges of the 21st century. A
selection of these concepts and methods is considered, with an assessment of how
each of them relates to and influences distribution network planning. This
assessment has not been performed before. The result is particular emphasis being
put on the concept of design rationale, which is identified as being of primary
importance in improving distribution network planning.
Planning and design is ultimately all about making decisions. Decisions require
information to be gathered and may rely on various forms of analysis but tools are
also available to support the actual decision making process. The use of multiple
criteria decision making (MCDM) techniques is discussed, highlighting their value in
making explicit the identification, quantification and analysis of decision criteria. A
new general structure for MCDM-based planning is defined and its use is
demonstrated in a case study where MCDM techniques are used to assess the
financial viability and technical desirability of a number of DG options. This
includes the novel specification and use of alternative ratios within the MCDM
framework to provide a new perspective and enhance the information available to
decision makers. The case study demonstrates the assessment of novel solutions
based on DG alongside conventional grid reinforcement solutions.
Planning and design, whether for electricity distribution networks or for other
systems, involve the management and processing of information. However, the
management of information can impose a huge burden on analysts and decision
2
makers, and this burden is growing with the new challenges of the 21st century. A
novel combination of methods is suggested to reduce the burden of information
processing and thereby enhance the productivity of analysts and decision makers.
The modelling and analysis of new technologies and the greater exploitation of
resources through improved representation and exchange of power system models
and data is discussed. By way of demonstrating the ideas presented and further
examining the principal challenge in network planning, a comprehensive set of
information requirements for DG is defined.
Trends like industry restructuring and DG are increasing the level of uncertainty and
risk faced by distribution network planners. A number of methods for managing
uncertainty and risk are discussed, including scenario analysis. A novel scenario
development methodology is demonstrated with an assessment of the expected
growth in DG in low voltage grids. This spotlights the primary challenge currently
facing planners and shows how gathered information can be used to generate
scenarios to support further analysis.
Power system operators are concerned with the expansion of wind power and the
effect it will have on their networks. This requires appropriate modelling and
analysis of wind farms within power system simulation software. A comprehensive
analysis of the problem is presented, supporting the development of new wind farm
models for the PSS/E simulation environment. Studies are performed to assess the
equivalence of single and aggregate models of wind farms and also to assess the
effects of a high penetration of wind farms on distribution networks.
In summary, the principle contributions of this work are: the detailed examination of
electricity distribution network planning and the new challenges facing planners in
the 21st century; the identification of shortcomings in the conventional approach to
planning and design; and the novel application of a range of methods from other
domains and the development of new tools leading to conclusions on ways in which
the conventional approach can be updated to meet the new challenges.
3
1.3. Publications
A number of publications have arisen from the research and related work. These are
listed below.
(i) Currie,R.A.F., Ault,G.W., Foote,C. E. T., McDonald,J.R.; “Active Power Flow
Management Utilising Operating Margins for the Increased Connection of
Distributed Generation”; Accepted for publication by IEE Proceedings
Generation, Transmission and Distribution, Reference GTD-2006-0035.R1
(ii) Foote,C.E.T., Burt,G.M., Elders,I.M., Ault,G.W.; “Developing Distributed
Generation Penetration Scenarios”; International Conference on Future Power
Systems, FPS 2005, 16-18 November 2005, Amsterdam, The Netherlands
(iii) Foote,C.E.T., Roscoe,A.J., Currie,R.A.F., Ault,G.W., McDonald,J.R.;
“Ubiquitous Energy Storage”; International Conference on Future Power
(iv) Ault,G.W., Foote,C.E.T., McDonald, J.R.; “UK research activities on advanced
distribution automation”; IEEE Power Engineering Society General Meeting
2005, June 12-16, 2005, p.2365-2368
(v) Foote,C.E.T., Ault,G.W., Burt,G.M., McDonald,J.R., Silvestro,F.,
“Information Requirements and Methods for Characterising Distributed
Generation”, 18th International Conference and Exhibition on Electricity
Distribution, CIRED 2005, June 2005, Turin, Italy
(vi) Foote,C.E.T., Ault,G.W., McDonald,J.R., Beddoes,A.J., “The Impact of
Network Splitting on Fault Levels and Other Performance Measures”, 18th
International Conference and Exhibition on Electricity Distribution, CIRED
2005, June 2005, Turin, Italy
(vii) Currie,R.A.F., Ault,G.W., Foote,C.E.T., Burt,G.M., McDonald,J.R.,
“Fundamental Research Challenges for the Active Management of Distribution
Networks with High Levels of Renewable Generation”, Universities’ Power
Engineering Conference (UPEC) 2004, Bristol, U.K., Conference Proceedings
Volume 3, p.1024-1028
(viii) Mienski,R., Pawelek,R., Wasiak,I., Gburczyk,P., Foote,C., Burt,G., Espie,P.,
“Power Quality Improvement in LV Networks Using Distributed Generation”,
4
2004 International Conference on Harmonics and Quality of Power, Lake
Placid, New York, USA, September 2004
(ix) Mienski,R., Pawelek,R., Wasiak,I., Gburczyk,P., Foote,C., Burt,G., Espie,P.,
“Voltage Dip Compensation in LV Networks Using Distributed Energy
Resources”, 2004 International Conference on Harmonics and Quality of
Power, Lake Placid, New York, USA, September 2004
(x) McMorran,A.W., Ault,G.W., Elders,I.M., Foote,C.E.T., Burt,G.M.,
McDonald,J.R., “Translating CIM XML Power System Data to a Proprietary
Format for System Simulation”, IEEE Transactions on Power Systems, vol.19,
issue 1, February 2004, p.229-235, February 2004
(xi) Espie,P., Foote,C. E. T., Burt,G. M., McDonald,J. R., Wasiak,I., “Improving
Electrical Power Quality Using Distributed Generation: Part 1 - Assessing DG
Impact and Capability”, 7th International Conference on Electrical Power
Quality and Utilisation, September 2003
(xii) McMorran,A.W., Ault,G.W., Foote,C.E.T., Burt,G.M., McDonald,J.R., “Web
Services Platform For Power System Development Planning”, Aristotle
University of Thessaloniki, 38th International Universities Power Engineering
Conference, September 2003
(xiii) Ault,G.W., Foote,C.E.T., McDonald,J.R., “Distribution System Planning in
Focus”, IEEE Power Engineering Review, Volume 22, Number 1, January
2002, pp 60-62., January 2002
(xiv) Foote,C.E.T., Watson,A.S., Espie,P., Ault,G.W., Burt,G.M., McDonald,J.R.,
“An evaluation strategy for electricity distribution network planning and
design”, 3rd Mediterranean Conference and Exhibition on Power Generation,
Transmission, Distribution and Energy Conversion, MED POWER 2002,
Athens, Greece, November 2002
(xv) Espie,P., Foote,C.E.T., Ault,G.W., McDonald,J.R., “A Multiple Criteria Model
for Evaluating Distributed Generation Development Options”, Second
International Symposium on Distributed Generation: Power System and
Market Aspects, Stockholm, Sweden, October 2002
5
(xvi) Foote,C.E.T., Ault,G.W., Burt,G.M., McDonald,J.R, “Enhancing flexibility
and transparency in the connection of dispersed generation”, CIRED 2001,
Amsterdam, June 2001
(xvii) Foote,C.E.T., Ault,G.W., Burt,G.M., McDonald,J.R., Green,J.P, “Towards a
structured methodology for distribution network design applications”, 35th
Universities Power Conference, Belfast, September 2000
6
2. Conventional Perspective on Distribution Network Planning
In this Chapter, the conventional perspective of electricity distribution network

planning is presented. The first section briefly introduces the distribution network
planning “problem”, describing how planners must abide by regulations, standards
and guidelines. The rest of the Chapter presents a model of the conventional
approach to distribution network planning that has been developed through the novel
application of a modified version of a formal knowledge modelling methodology. In
the following Chapters, drivers and future directions in distribution networks are
discussed along with the resultant shortcomings that are exposed in this conventional
approach to distribution planning.
2.1. The Distribution Network Planning “Problem”
The basic purpose of electricity distribution networks is to deliver electrical power

from the energy source, which means the the transmission network, to consumers.
This must be achieved within acceptable standards of safety and quality and at an
acceptable cost. Normally, the requirements that must be met by distribution
network operators (DNOs) are defined in legislation, regulations, standards and other
formal documents. These objectives may be augmented further by the goals and
ambitions of the DNO itself. Electricity distribution networks are constrained by
what is technically possible, what society and communities permit, and the cost of
implementation. Thus, as with other complex engineering tasks, the basic electricity
distribution network planning “problem” is to meet, or get as close as possible to, all
the objectives while meeting all the constraints.
In electricity distribution network planning, objectives and constraints are concerned

mainly with safety, quality, reliability and cost. Most issues may be viewed as both
objectives and constraints. For example, most DNOs will claim to strive for ever-
improving safety while also having to meet statutory safety standards. As with most
complex problems, many of the objectives will be conflicting, e.g. minimise costs
versus technical improvements.
7
The detailed requirements vary from one area to another but in the simplest terms,
distribution networks are very similar all over the world. There are transformers,
switches and ancillary equipment in substations, which are connected to sources,
consumers and other substations by overhead lines and underground cables. In the
developed world, electricity supply systems have been in operation for more than a
century and now reach almost the entire population. Thus, there is a substantial
installed base of assets to build upon. Furthermore, many items of equipment have a
service life running into decades so may have to operate through widely varying
circumstances. Network planners must tackle the problem of extending and
renewing the system to meet all the demands put upon it, now and in the future.
2.2. Model of the Conventional Approach to Distribution Network Planning
In this section, the conventional approach to the planning and design of electricity
distribution networks is described using a modified knowledge modelling
methodology.
2.2.1. Knowledge Modelling Methodology
A formal modelling approach was adopted to provide some structure in the analysis
of the conventional approach, to facilitate its representation in diagrammatic form,
and to present a widely known subject in a novel way. The methodology is a
simplified version of the KADS methodology [2.1] and was used because of the
clarity that it can provide in modelling engineering activities. This type of modelling
approach has been found to be very effective in capturing and representing
knowledge in a diverse range of topics. It also facilitates the codification of
knowledge, which is necessary in the development of knowledge based or automated
systems. The full KADS methodology was not used because the purpose of the
analysis was only to understand the structure of the planning activity. Application of
the full KADS methodology was unnecessary and out of the scope of this work.
The KADS methodology was simplified by neglecting the inference and domain
layers to leave only the task layer. In the full methodology, the inference and domain
8
layers provide further detail on the knowledge resources and processes necessary to
fulfil tasks. The use of the task layer only allows a useful description of the activity
to be constructed, as demonstrated with the model below. Other researchers may
find a similar approach useful in assessing other engineering topics.
Distribution planning and design is represented below in terms of tasks that are
broken down into sub-tasks or methods. A task is a specific activity that must be
undertaken to satisfy its parent task. A method is a tool or technique that may be
used to satisfy its parent task. The symbols used are shown in Figure 2.1. The
rectangular boxes with double lines and rounded corners represent tasks and sub-
tasks. The ellipses represent methods. The arrow indicates that a task is broken
down further into sub-tasks or methods.
Figure 2.1 – Symbols used in knowledge models
Task Method
2.2.2. Knowledge Models
The models are presented in levels, where level one represents the highest level of
task. It was found that five levels were sufficient to represent the breakdown of
tasks. In each section an introductory explanation is provided together with
summary information identifying the parent task one level up and child tasks where
the model goes to another level of detail.
The models were compiled using a number of sources including literature [2.2-2.15]
and discussion with utility personnel [2.16, 2.17].
Level One: Plan and design the distribution network
The conventional approach to the task of planning and designing the distribution
network can be split into six broad sub-tasks (Figure 2.2). There is a need to identify
9
requirements and then design the necessary additions and alterations to the network.
These designs are then implemented in programmes of work. Through all planning
and design activities there is a need to maintain communication and exchange
information with others, be it within the organisation, with neighbouring networks,
with customers or with the regulator. As a check that the planning and design
process is satisfying requirements, there should be some review of plans actually
implemented, which will then inform future activities. Finally, the policies,
procedures and standards upon which planning and design is based may be revised
when appropriate. These tasks are continuous and ongoing as the electricity
distribution network is updated and extended to meet changing requirements.
Parent: None
Children: Level Two: Identify requirements
Level Two: Design network additions and alterations
Level Two: Determine programmes of implementation
Level Two: Revise policies, procedures and standards
10
Figure 2.2 – Task Model: Plan and design the distribution network
Identify requirements
Design network
additions and
alterations
Determine programmes
of implementation
Plan and design the

distribution network
Maintain
communication and
exchange information
with others
Review implementation
to see that
requirements are met
Revise policies,
procedures and
standards
Level Two: Identify requirements
In identifying the requirements of the distribution network (Figure 2.3), planners

must consider their basic goals and responsibilities. Further to that, there is a set of
development drivers that influence plans for the network and must be considered.
Finally, the range or requirements that emerge from considering goals and drivers
must be combined.
11
Parent: Level One: Plan and design the distribution network
Children: Level Three: Consider goals and responsibilities
Level Three: Consider development drivers
Level Three: Combine requirements from different considerations
Figure 2.3 – Task Model: Identify requirements
Consider goals and

responsibilities
Consider development
drivers
Combine requirements
from different
considerations
Level Three: Consider goals and responsibilities
Distribution network companies are normally subject to strong government

regulation and their basic goals and responsibilities will be defined in standards and
regulations (Figure 2.4). However, it is useful to also highlight specific goals and
responsibilities as being: maintain safety; maintain supply; optimise economic
efficiency; and minimise environmental impact. Regulatory regimes vary from place
to place, but a common factor is that regulators sometimes alter the incentives and
targets for distribution companies. Part of the planning task is to respond to these
changing requirements. Over time, standards and regulations are likely to become
more stringent, requiring better safety, improvements in the quality and reliability of
the electricity supply, and reduced environmental impact – or even active
environmental improvement. In many cases, regulators will demand all these
improvements while reducing the money available.
Parent: Level Two: Identify requirements

Children: None
12
Figure 2.4 – Task Model: Consider goals and responsibilities
Satisfy all relevant

standards and
regulations
Maintain safety
Consider goals and

Maintain supply
responsibilities
Optimise economic
efficiency
Minimise environmental
impact
Level Three: Consider development drivers
In the conventional approach to distribution network planning and design, six

principal development drivers can be identified (Figure 2.5): new connections; load
growth; asset replacement or refurbishment; network performance; interfacing with
the transmission system; and new products and concepts. The first four sub-tasks are
dealt with in more detail below. Transmission network developments are an
important driver on distribution networks because transmission-level effects cascade
down through the voltage levels. There must be open communication between
transmission and distribution companies. Finally, the development of distribution
networks have always been driven to some degree by the introduction of new
products or new concepts, e.g. covered conductors, or containerised substations.
New products and concepts must undergo testing and evaluation and they will be
introduced to the network gradually.
13
Children: Level Four: Consider new connections
Level Four: Consider general load growth
Level Four: Consider asset replacement or refurbishment
Level Four: Consider network performance
Figure 2.5 – Task Model: Consider development drivers
Consider new
connections
Consider general load

growth
Consider asset
replacement or
refurbishment
Consider development
drivers
Consider network
performance
Consider transmission
network developments
Consider new products

and concepts
Level Four: Consider new connections
In considering new connections to the network (Figure 2.6), planners must be

proactive in predicting where and when new connections will be made. But it is also
necessary to respond to applications for connection that may have been predicted or
14
not. In the conventional approach, as practiced by distribution companies over the
last few decades, new connections are mostly for new loads, like a new housing
estate or commercial development. Distribution network planners have always had
to consider the connection of generators but for most companies, over the last four or
five decades, this has been rare enough to be managed separately and on a purely
reactive basis.
Parent: Level Three: Consider development drivers

Children: None
Figure 2.6 – Task Model: Consider new connections
Predict new
connections
Consider new
connections
Assess applications for

new connections
Level Four: Consider general load growth
In conventional distribution network planning, an important element is forecasting

load growth (Figure 2.7). Greater demands on the network result in insecurities and
ratings being exceeded, which must be identified if action is to be taken.

Children: Level Five: Forecast load
Level Five: Identify insecurities and exceeded ratings
15
Figure 2.7 – Task Model: Consider general load growth
Forecast load
Consider general load

growth
Identify insecurities and

exceeded ratings
Level Five: Forecast load
In forecasting load a variety of approaches have been adopted but two of the most
widely used methods are forecasts based on economic growth estimates and forecasts
based on local knowledge and trending (Figure 2.8). Combining the results of
different methods can produce more confidence in the overall forecasts made.
Parent: Level Four: Consider general load growth

Children: None
Figure 2.8 – Task Model: Forecast load
Commercial
forecast based on
economic growth
estimates
Forecast load
Engineering
forecast based on
local knowledge
and trending
Level Five: Identify insecurities and exceeded ratings
The identification of insecurities and exceeded ratings relies on feedback from

experience of operating the network and on network modelling and analysis (Figure
2.9). These two methods broadly describe the ongoing analysis of the network’s
performance and are used in a number of tasks.
16
Feedback from operational experience might be concerned with plant where ratings
have been, or have been close to being, exceeded. Or operators might identify
portions of network that could be reconfigured to improve performance.
Network modelling and analysis covers the use of a wide range of simulation and
analysis software tools. Depending on the circumstances, this may cover the full
range of power system dynamics, from generation expansion planning to
electromagnetic transients, and also steady state analysis such as load flow.
However, in conventional distribution planning and design the range of tools used
would be limited; analysis would focus on static studies of voltage levels, power
flows and fault levels.
Parent: Level Four: Consider general load growth

Children: None
Figure 2.9 – Task Model: Identify insecurities and exceeded ratings
Feedback from
operational
experience
Identify insecurities and
exceeded ratings
Network modelling
and analysis
Level Four: Consider asset replacement or refurbishment
In most developed countries, where electricity distribution networks are mature and
assets are reaching the end of their predicted lives, asset management has become
more important as a development driver, often becoming the primary focus for
distribution companies. Determining whether assets need replacing or refurbishing
requires assessing the condition of the plant and identifying redundant network assets
(Figure 2.10). Many planners will also use statistical modelling and analysis of
distribution assets, which are numerous and suited to statistical studies.
17
Children: Level Five: Assess condition of plant
Level Five: Identify redundant network
Figure 2.10 – Task Model: Consider asset replacement or refurbishment
Assess condition of
plant
Consider asset
Identify redundant
replacement or
network
refurbishment
Statistical
modelling and
analysis of asset
replacement
Level Five: Assess condition of plant
Plant is most commonly assessed through visual inspection but there are also
diagnostic tools available to provide information to planners and designers (Figure
2.11). However, the tools available are not used comprehensively and in some cases
plant will only be identified as being in poor condition when it fails or its
performance deteriorates. Thus, planners must utilise feedback from operational
experience as well as inspections and diagnostic tools.
Parent: Level Four: Consider asset replacement or refurbishment

Children: None
18
Figure 2.11 – Task Model: Assess condition of plant
Visual inspections
of plant
Diagnostic tools to
Assess condition of
assess plant
plant
condition
Feedback from
operational
experience
Level Five: Identify redundant network
Identifying redundant network requires the same combination of feedback from

operational experience and network modelling and analysis as is required to identify
insecurities and exceeded ratings (Figure 2.12).
Parent: Level Four: Consider asset replacement or refurbishment

Children: None
Figure 2.12 – Task Model: Identify redundant network
Feedback from
operational
experience
Identify redundant
network
Network modelling
and analysis
Level Four: Consider network performance
The network performance measures that must be met will be defined in standards and
regulations. Determining whether these performance targets are being met rely once
again on feedback from operational experience and network modelling and analysis
19
(Figure 2.13). Any failure to meet performance targets will drive network
developments. The greatest pressure from customers and other stakeholders,
including the government, will come when there are significant network disruptions.
For example, severe storms can cause widespread disruption and prompt strong
criticism. Thus, an important objective is to try and minimise disruption when
storms occur and learn lessons from any past experience. The need to maintain
safety has already been identified as one of the specific goals and responsibilities of
distribution companies. However, it is useful to highlight how safety related issues
are an explicit development driver.

Children: None
Figure 2.13 – Task Model: Consider network performance
Feedback from
operational
experience
Network modelling
and analysis
Consider network
performance
Experience of
significant network
disruptions
Safety related
issues
Level Three: Combine requirements from different considerations
The multiple goals and responsibilities of distribution companies and the multiple
development drivers for the network result in a complex set of interrelated
requirements for planning and design. An important task is the combining of these
20
requirements into manageable blocks (Figure 2.14). This is most commonly done on
the basis of geography, network impact and time horizon.

Children: None
Figure 2.14 – Task Model: Combine requirements from different considerations
by geography
(Geographic
Information System)
by network impact
(Network Management
System)
from different
considerations
Combine
requirements by
time horizon
Split requirements
into manageable
blocks
Level Two: Design network additions and alterations
The conventional approach to distribution network design focuses on the three

elements that make up the network: substations, feeders and ancillary equipment
(Figure 2.15). These high-level descriptions disguise the considerable detail
involved. In addition, network design also takes into account how the network
components will be configured or how they might be reconfigured. Network
operation also influences designs; e.g. designers will sometimes specify switching
methodologies to accompany network designs.
21
Children: Level Three: Determine substation, feeder and ancillary equipment
details
Figure 2.15 – Task Model: Design network additions and alterations
Determine substation
site, capacity and
configuration
Determine feeder route,

capacity and
connection
Design network
additions and
alterations
Determine ancillary
equipment type and
location
Determine configuration
and operational
methodologies
Level Three: Determine substation, feeder and ancillary equipment details
For all three elements of the conventional approach to physical network design, the
methods used are similar (Figure 2.16). Academic researchers have proposed vary
many complex optimisation methods based on esoteric mathematics or the latest,
fashionable technique [2.15]. Despite this, distribution company engineers are more
likely to turn to simpler and more robust approaches. These include referring to
written procedures or accepted “rules of thumb”. Conventionally, lists of standard
designs and components have been used, although there has always been scope to
gradually incorporate new products and concepts. To reduce the burden of analysis,
prepared tables and charts have emerged that specify design details such as
conductor diameters to meet certain conditions. However, designers will perform
some network modelling and analysis where it is required. The application of
personal knowledge, experience and engineering judgement has always been
22
important. In particular, engineers will apply experience of similar projects that they
have encountered in the past.
Parent: Level Two: Design network additions and alterations

Children: None
Figure 2.16 – Task Model: Determine substation, feeder and ancillary equipment details
Follow company policy

and procedure or
accepted “rules of
thumb”
Utilise standard
Determine substation designs and
site, capacity and components (gradually
configuration incorporate new
products)
Determine feeder route, Refer to prepared

tables and charts (with
capacity and liberal tolerances and
connection safety factors)
Determine ancillary
Network modelling
equipment type and
and analysis
location
Apply knowledge,
experience and
engineering
judgement
Level Two: Determine programmes of implementation
Having identified requirements on the network and designed the additions and
alterations necessary, these changes must be implemented, and revised, as
programmes of work (Figure 2.17). Performing construction or significant
maintenance work in the network will require outages so the constraints on network
access must be assessed. In addition, the availability of the resources required to do
the work may influence the work programmes. This will include people like
contractors and authorised personnel, machinery and materials, and, perhaps most
23
important of all, finance. Finally, distribution networks exist in a dynamic
environment and all plans must be revised as conditions change.

Children: Level Three: Assess network access constraints
Figure 2.17 – Task Model: Determine programmes of implementation
Assess network access

constraints
Determine programmes Assess implementation

of implementation resource constraints
Revise plans as
conditions change
Level Three: Assess network access constraints
Assessing network access constraints requires feedback from operational experience

and appropriate network modelling and analysis (Figure 2.18). In particular,
planners must ensure contingency plans are in place to meet security and reliability
requirements while work is being undertaken on the network.
Parent: Level Two: Determine programmes of implementation

Children: None
24
Figure 2.18 – Task Model: Assess network access constraints
Feedback from
operational
experience
Assess network access Network modelling

constraints and analysis
Contingency
analysis
Level Two: Revise policies, procedures and standards
Planning and design of the distribution network relies very heavily on policies,
procedures and standards. These can be internal to the distribution company and
external, perhaps as industry-wide agreements or as part of statutory regulations.
These policies, procedures and standards are revised as issues emerge in the planning
and design of networks (Figure 2.19). These tasks must look to the future to identify
emerging issues and possible externally imposed changes to ensure that policies,
procedures and standards are in place for when they are needed.

Children: None
Figure 2.19 – Task Model: Revise policies, procedures and standards
Revise internal policies,

procedures and
standards
Revise policies,
procedures and
standards
Propose or support changes
to industry and statutory
policies, procedures and
standards
25
2.3. The Complete Model
The complete model is shown in the fewest separate parts as is practical for this
document in Figure 2.20 and Figure 2.21. The model can also be described in text-
only form as shown in Figure 2.22.
The knowledge modelling methodology as applied here results in a complete model

that is concise and clear, providing a new description of the conventional approach to
distribution network planning and design. Although based on a review of literature,
supplemented through discussions with distribution network operators and
experienced practitioners, this is a novel representation that provides an opportunity
to assess the conventional approach in new ways.
While the model is simply a description of the conventional approach, not being
codified in any way other than as in the diagrams, this new description allows
specific parts of the process to be highlighted and seen in context. When considered
alongside the drivers and new directions in the next Chapter the model exposes some
shortcomings in the conventional approach. These shortcomings and the methods
proposed for addressing them are discussed in more detail in later Chapters.
However, even before the new challenges facing DNOs are considered, the model
points to some important aspects of and weaknesses in the conventional approach.
“Network modelling and analysis” and “Feedback from operational experience” both
appear in the model a number of times, highlighting the importance of these
activities. While there are tasks to “Predict new connections” and “Forecast load”,
the process is largely a reactive one based on external drivers. This is appropriate
since the network should reflect the needs of its users but it can place a burden on the
DNO to perform very complex planning and design tasks in a short time. While
“Consider new products and concepts” is one of the development drivers, the model
shows how network additions and alterations are designed using conventional
methods and solutions. Overall, the model shows that distribution network planning
involves a complex set of tasks and methods that mean highly prescriptive and
restrictive methods are inappropriate.
26
Figure 2.20 – The complete model of the conventional approach, part 1 of 2
Satisfy all relevant

standards and
regulations
Maintain safety
Consider goals and

Maintain supply
responsibilities
Optimise economic
efficiency
Minimise
environmental impact
Predict new
connections
Consider new
connections
Assess applications
for new connections
Commercial
forecast based
on economic
growth
estimates
Forecast load
Engineering
forecast based
on local
knowledge and
Consider general trending
load growth
Feedback from
operational
Plan and design the Consider experience
distribution network development drivers
Identify insecurities
and exceeded ratings
Network
modelling and
analysis
Visual
inspections of
plant
Diagnostic tools
Assess condition of
to assess plant
plant
condition
Feedback from
operational
Consider asset experience
replacement or
refurbishment
Feedback from
operational
experience
Identify redundant
network
Network
modelling and
analysis
Statistical
modelling and
analysis of asset
replacement
Feedback from
operational
experience
Network
modelling and
analysis
Consider network
performance
Experience of
significant
network
disruptions
Safety related
issues
Consider
transmission network
developments
Consider new
products and
concepts
27
Figure 2.21 – The complete model of the conventional approach, part 2 of 2
Combine
requirements by
geography
(Geographic
Information
System)
Combine
requirements by
network impact
Combine (Network
requirements from Management
Identify requirements System)
different
considerations
Combine
requirements by
time horizon
Split
requirements
into manageable
blocks
Follow company
policy and
procedure or
accepted rules
of thumb Ó
Utilise standard
designs and
Determine substation components
site, capacity and (gradually
configuration incorporate new
products)
Determine feeder Refer to

route, capacity and prepared tables
connection and charts (with
liberal tolerances
and safety
Determine ancillary factors)
Design network equipment type and
additions and location
Network
alterations modelling and
analysis
Apply
knowledge,
experience and
engineering
judgement
Plan and design the Determine

distribution network configuration and
operational
methodologies
Feedback from
operational
experience
Assess network Network

access constraints modelling and
analysis
Contingency
analysis
Determine
programmes of
implementation
Assess
implementation
resource constraints
Revise plans as
conditions change
Maintain
communication and
exchange
information with
others
Review
implementation to
see that
requirements are met
Revise internal
policies, procedures
and standards
Revise policies,
procedures and
standards Propose or support
changes to industry
and statutory
policies, procedures
and standards
28
Figure 2.22 – Text-only version of the complete model of the conventional approach
Plan and design the distribution network

o Identify requirements
Consider goals and responsibilities
• Satisfy all relevant standards and regulations
• Maintain safety
• Maintain supply
• Optimise economic efficiency
• Minimise environmental impact
Consider development drivers
• Consider new connections
o Predict new connections
o Assess applications for new connections
• Consider general load growth
o Forecast load
Commercial forecast based on economic growth estimates
Engineering forecast based on local knowledge and trending
o Identify insecurities and exceeded ratings
Feedback from operational experience
Network modelling and analysis
• Consider asset replacement or refurbishment
o Assess condition of plant
Visual inspections of plant
Diagnostic tools to assess plant condition
o Identify redundant network
Network modelling and analysis
o Statistical modelling and analysis of asset replacement
• Consider network performance
o Feedback from operational experience
o Network modelling and analysis
o Experience of significant network disruptions
o Safety related issues
• Consider transmission network developments
• Consider new products and concepts
Combine requirements from different considerations
• Combine requirements by geography (Geographic Information System)
• Combine requirements by network impact (Network Management System)
• Combine requirements by time horizon
• Split requirements into manageable blocks
o Design network additions and alterations
Determine substation site, capacity and configuration
Determine feeder route, capacity and connection
Determine ancillary equipment type and location
• Follow company policy and procedure or accepted “rules of thumb”
• Utilise standard designs and components (gradually incorporate new products)
• Refer to prepared tables and charts (with liberal tolerances and safety factors)
• Network modelling and analysis
• Apply knowledge, experience and engineering judgement
Determine configuration and operational methodologies
o Determine programmes of implementation
Assess network access constraints
• Feedback from operational experience
• Network modelling and analysis
• Contingency analysis
Assess implementation resource constraints
Revise plans as conditions change
o Maintain communication and exchange information with others
o Review implementation to see that requirements are met
o Revise policies, procedures and standards
Revise internal policies, procedures and standards
Propose or support changes to industry and statutory policies, procedures and standards
29
2.4. Review of Chapter
In this Chapter, electricity distribution network planning was introduced with a novel
analysis of the conventional approach using a modified knowledge modelling
methodology. The methodology is a simplified version of the KADS methodology
and was used because of the clarity that it can provide in modelling engineering
activities. The full KADS methodology was not used because the purpose of the
analysis was only to examine the structure of the planning activity. The models
identified and examined the main tasks in network planning, with reference to a wide
range of sources. This analysis reveals the structure of the distribution network
planning activity and supports its further assessment and proposed modifications in
later chapters, particularly the identification of shortcomings in Chapter 4.
The complete set of knowledge models provide a novel perspective on the

conventional approach to electricity distribution network planning, highlighting some
of the important features and weaknesses. It serves as a useful tool for learning and
facilitates reassessment in the light of new challenges, which are discussed in the
next Chapter. In particular, the models highlight the role of diverse factors in
network planning and illustrate how the optimisation of substation site and size,
feeder route and capacity and ancillary devices – so often the focus of academic
work on distribution network planning – play a relatively small role in the overall
task facing planners. The models also indicate the role of network modelling and
analysis, which is identified in Chapter 4 as one of the shortcomings that requires
particular attention.
2.5. Chapter References
2.1. Schreiber,G., Akkermans,H., Anjewierden,A., de Hoog,R., Shadbolt,N., Van

de Velde,W., Weilinga,B.; “Knowledge Engineering and Management: The
CommonKADS Methodology”; 2000, Massachusetts Institute of Technology;
ISBN 0-262-19300-0
2.2. C.R.Bayliss; “Transmission and Distribution Electrical Engineering”;
Butterworth-Heinemann; 1996; ISBN 0750622873; D621.319BAY
30
2.3. Lakervi,E., Holmes,E.J.; “Electricity distribution network design”; 2nd edition;
1996; Peter Peregrinus Ltd.; ISBN 0863413099
2.4. CIRED Experts Group, Session No.1 – Group Report; Survey of CIRED
members; May 1997
2.5. Stowell,P.; “Manweb Distribution”; Journal IEE, Jan 1958, p.15-21
2.6. UNIPEDE Distribution Study Committee 50.04.DISNET; “Distribution
network configuration and design – Applied practices, likely trends and
evaluation of technical solutions”; September 1995; Ref.: 05004Ren9540
network configuration and design – Network design - Applied practices in
European countries”; September 1995; Ref.: 05004Ren9539
network configuration and design – Evaluation of technical solutions”;
September 1995; Ref.: 05004Ren9538
network configuration and design – Likely trends in distribution systems”;
September 1995; Ref.: 05004Ren9537
2.10. Burke,J.J.; “Power Distribution Engineering: Fundamentals and Applications”;
1994; Marcel Dekker, Inc.; ISBN 0-8247-9237-8; D621.319BUR
2.11. Pansini,A.J.; “Electrical distribution engineering”; 2nd ed; 1992; The Fairmont
Press, Inc.; ISBN 0-88173-121-8; D621.319PAN
2.12. Raytheon Engineers & Constructors, EBASCO Division, Electric Power
Systems; “Electric Distribution Systems Engineering Handbook”; 3rd ed.;
1992; McGraw Hill, Inc.; ISBN 0-07-607079-4
2.13. Eggleton,M.N., Mazzoni,M., Van Geert,E., Van Der Meijden,M.A.M.M.,
Kling,W.L.; “Network structure in sub-transmission systems. Features and
practices in different countries.”; CIRED 12th International Conference on
Electricity Distribution, 1993, Subject Area 6: Design and Planning of Public
Supply Systems; IEE Conference Publication No.373, p.6.9.1-6.9.8
2.14. Carr, J., McCall, L.V.; “Divergent Evolution and Resulting Characteristics
Among the World’s Distribution Systems”; IEEE Trans. on Power Delivery,
vol.7, no.3, July 1992, p.1601-1609
31
2.15. Willis, H.L., Northcote-Green, J.E.D.; “Comparison of Several Computerized
Distribution Planning Methods”; IEEE Trans. on Power Apparatus and
Systems, vol.PAS-104, no.1, January 1985, p.233-240
2.16. Foote, C.; “Planning and Design of the Distribution Network”; February 2000;
CEPE Report EME/EP/KT/1999-002A; University of Strathclyde
2.17. Foote, C.; “Review of Distribution Planning and Design Techniques”; March
2000; CEPE Report EME/EP/REV/2000-002; University of Strathclyde
32
3. Drivers and New Directions in Distribution Networks
In this Chapter, the drivers and new directions in electricity distribution network
planning, operation and management are examined. This review of current trends
helps expose the shortcomings in the conventional approach to distribution networks
described in the previous Chapter. These shortcomings are discussed explicitly in
the next Chapter and provide the justification for the application of new tools and
techniques, which is presented in later Chapters. The choice of which tools and
techniques to test and develop was driven by this assessment of new challenges in
distribution network planning and by the new directions in the electricity industry
and technology in general.
Electricity distribution networks connect sources of electrical energy, typically

transmission grid substations, to consumers of electrical energy, e.g. industrial,
commercial and domestic consumers. Distribution networks have traditionally been
designed for uni-directional power flow, from higher to lower voltages. Distribution
networks represent a huge installed asset base, with thousands of kilometres of
individual circuits at low voltage – distribution networks in the UK have a total
length of 767,376 kilometres [3.1]. These networks are actively monitored only at
the higher voltage levels. Due to the large scale, a “fit-and-forget” approach has
often been adopted, with assets installed to accommodate an expected future growth
in demand. In general, distribution networks have proved very effective, delivering
electricity at reasonable cost, quality and safety.
Established distribution networks in developed countries are now undergoing

considerable change. A range of drivers is influencing their management and
development. Broadly, drivers for change in electricity distribution networks can be
split in two: technology drivers and commercial/regulatory drivers. These drivers
present a range of new challenges to distribution companies and will result in
distribution networks being taken in new directions.
33
3.1. Technology Drivers in Distribution Networks
Technology drivers can be separated into power technologies and communications

and control technologies [3.21, 3.22]. Power technology drivers include distributed
generation, energy storage, the demand side and power electronics. All of these
technology areas have seen advances in recent years and are affecting electricity
distribution networks. Communications and control technologies that are driving
change in distribution networks include the Internet, mobile communications,
automation and the ever-increasing processing power of computer systems. The
sections below discuss these technology drivers in more detail.
3.1.1. Distributed Generation
Distributed generation (DG) refers to generators connected directly to distribution

networks. Typically, DG installations are smaller than conventional power stations,
which are connected at transmission level, and may utilise unconventional energy
sources or conversion methods, such as renewable sources or power electronics. DG
is also sometimes called embedded generation because it is embedded in distribution
networks, or dispersed generation because small generators can be seen as dispersed
around the system rather than being centralised like large, conventional power plants.
In this document the terms distributed generation or DG will be used.
Distributed generation is probably the single, most significant issue driving change in
modern electricity distribution networks at the start of the 21st century.
Consequently, there is a large volume of published work covering all aspects of DG
and its impact on distribution networks and concise summaries of most of the issues
are available [3.2, 3.3, 3.4, 3.5, 3.6]. Industry rules and standards have been
established for many aspects of DG although many are being reviewed in light of the
expansion of DG [3.7, 3.8, 3.9]. This highlights the importance of the level one task
in the knowledge model of the conventional approach: “Revise policies, procedures
and standards”.
The amount of small-scale generation connected to distribution networks is

increasing in many countries [3.13]. This is a result of a combination of factors
34
including changes to market and regulatory structures, advances in generation
technology and changing attitudes with respect to the environment. In an
increasingly competitive energy market, the modularity and short lead times of DG
increase its attractiveness in comparison to larger investments by reducing the risk
associated with individual projects. The vertically integrated utilities of some
countries also see DG as an opportunity to offset investments in transmission and
distribution infrastructure. This will require novel approaches to network design and
management and enhancements to the “Network Modelling and Analysis” method
used at various points in the knowledge model of the conventional approach. With
appropriate operation and control, DG has the potential to improve voltage profiles,
reduce losses, and improve operational flexibility.
3.1.1.1. Technical Issues
Distributed generation can affect the distribution network in many ways. Generator
performance and the impact on network performance depend very much on network
and generator characteristics. All aspects of a network’s performance, from network
power flows to reliability, may be influenced although the most important issues vary
from case to case. The technical issues most frequently highlighted for close
examination are network capacity, electrical losses, fault levels and stability.
Generally, technical solutions are available to problems faced in connecting DG but
they add costs, which raise important questions about who pays, as discussed in the
next section. Various tools and techniques are required to support the analysis of
different scenarios [3.2], and some solutions are new or have not been previously
used by DNOs.
3.1.1.2. Economic Issues
Economic conditions influence the installation and operation of DG. Most simply,
growth in the economy increases the overall demand for electricity. For the
economic reasons mentioned above of risk reduction and better matching supply to
demand, new capacity to replace retired generators or meet future increases in
demand is likely, in the near-term, to come from smaller-scale installations. The
comparative economic benefits of economies of scale in power plants have been
35
reduced by improvements in small-scale generator technologies. The argument for
smaller-scale, and therefore distributed, generation is enhanced by government
incentives for environmentally friendly sources like renewable energy and combined
heat and power [3.10].
The viability of DG schemes is dependent on a number of other economic factors,

including the energy trading mechanism and the costs imposed by the DNO. The
cost of connecting to a network can be a controversial subject because all parties
must ensure that they can recover any costs incurred through their revenue streams.
For DG operators this means the sale of energy, the value of which will depend on
the market rate, unless some form of subsidy or guaranteed price is on offer.
DG can influence both the capital and operational expenditure of distribution

companies. They must be confident that any investment in their network to
accommodate DG will be fully recovered, either from the DG operator directly or
through future payments from users of the system. The opportunity exists for
distribution companies to make investment in the connection of DG to reduce
operational expenditure and reduce the need for investment in areas such as load
related reinforcement, asset replacement and performance improvement. As
monopolies, distribution companies are usually subject to price control regulation of
some sort. If the price control mechanism does not make allowance for funding
contracts with DG as an alternative to investment in network assets then DG may not
be financially attractive to distribution companies.
Where the use of DG proves more expensive than conventional solutions, this
additional cost must be covered. Whether through higher energy prices charged by
the DG operator, higher network use of system charges by the DNO, or taxes to pay
for government subsidies, the higher costs will ultimately be passed on to the
consumers of electrical energy.
Economic issues, and others affecting DG, are subject to regulation and may change.
Thus, planners must be aware of existing regulations and possible future changes.
36
3.1.1.3. Environmental Issues
Environmental issues are amongst the principal drivers for DG [3.6]. The incentive
to reduce emissions is provided by pollution taxes and other government-imposed
incentives. The use of combined heat and power (CHP), and technologies such as
wind turbines and photovoltaics help reduce emissions; and these technologies are
more suited to small-scale generation.
The potential for renewable and CHP generation in a particular area is a function of
the local environment. Some areas may provide large amounts of wind energy while
other areas may provide more solar energy. The potential for combined heat and
power plants depends on the requirements of suitable nearby heat loads. Another
environmental issue is planning permission. One of the perceived advantages of
conventional large-scale power stations was their distance from population centres.
DG is different and may face opposition due to the impact of generators on their
immediate environment.
3.1.1.4. Government Policies
Many governments of developed countries have energy policies similar to that of the
UK, namely to ensure secure, diverse and sustainable supplies of energy at
competitive prices [3.10,3.14]. The response to climate change is a central element
to policy in many countries and it is widely recognised that renewables and CHP
have a significant role to play in reducing emissions. In the case of the UK, the
government also sees renewables and CHP as contributing to employment and rural
development.
3.1.1.5. Distributed Generation Technologies
Distributed generation is a very broad term and can cover a wide range of generation
technologies [3.11, 3.12, 3.13, 3.14]. Renewables are often the focus of discussion
but there is also much research and development of small-scale, non-renewable
technologies, particularly micro turbines and fuel cells. These technologies are made
37
particularly attractive when used to provide combined heat and power. All of these
DG technologies are under continuous development – a problem in itself for
distribution companies trying to keep up with the latest technologies.
3.1.2. Energy Storage
One of the primary assumptions in the design and operation of electricity supply
systems is that electrical energy cannot be stored. In reality, energy storage has been
utilised for a long time, mainly in the form of pumped-storage hydro schemes,
although such schemes represent a tiny fraction of total system load. Various forms
of battery technology have long been used as backup supplies for the short-term
support of critical loads. This includes emergency lighting and uninterruptible power
supplies.
The range of effective energy storage technologies is expanding [3.15, 3.16, 3.17].
There have been improvements in battery and flywheel technology. Perhaps the
most exciting advances in storage technologies are associated with regenerative fuel
cells and flow batteries. These technologies generate electricity directly from
chemical reactions, and through management of the chemical products used in the
reaction can provide a flexible means of energy storage. Flow batteries are finding
applications from critical loads to system support at distribution and transmission
level.
The advances in energy storage technologies offer distribution companies new

options and new challenges in network design and operation. By undermining the
primary assumption upon which modern electricity systems have been designed,
energy storage is potentially the most disruptive of technology drivers. For example,
in the future it may be possible to perform some degree of load levelling at all levels
on the power system, thereby utilising assets more efficiently [3.18]. But this will
require big changes in system design and operation.
38
3.1.3. Demand Side Technology Developments
Demand side technology developments present new challenges to distribution

companies in a number of ways [3.17,3.21]. Increasingly, some loads require higher
standards of power quality and reliability. Distribution companies sometimes have
to design their system differently or install additional equipment to meet the
requirements of special customers. While electricity consumption is still rising in the
developed world, environmental and financial concerns mean there is an increasing
focus on energy efficiency. The implementation of energy-saving schemes can
affect patterns of electricity usage, perhaps shifting energy use between times of the
day or reducing the overall demand. But perhaps the greatest potential impact on the
way distribution systems are designed and operated is the prospect of more active
management of loads. As with distributed generation, integration of distributed load
management resources with network management offers potential benefits in asset
utilisation and avoided network expansion.
3.1.4. Power Electronics
Advances in materials, designs and control methods have resulted in great

improvements in power electronics [3.21]. This has made more options available in
various applications, including AC/DC conversion, variable-speed drives and
generators, switching, and ancillary devices like dynamic voltage restorers. These
new and improved technologies are being used more widely, by electricity
consumers and generators, and within transmission and distribution.
3.1.5. Communications and Control Technologies
In the last two decades there has been a massive expansion of new technologies and
market changes in the telecommunications industry. The Internet and related
systems make possible the concept of ubiquitous communications, all the time
reducing costs while expanding the services available.
Electricity distribution companies can now utilise a wide range of communications

media. These include the traditional land-based telephone networks, the new mobile
39
telephone networks, satellite communications, microwave and radio links, and power
line carrier. In various combinations, these communications technologies offer
unprecedented scope for monitoring and control of electricity distribution networks,
loads and generators.
The growing power and falling costs of computer technology have matched the
changes in communications. This has brought ever more complex computations
within reach of distribution network engineers, both in real time for system operation
purposes and off-line for simulation, analysis and network planning and design.
Allied to the new communications technologies, this computational power makes
new approaches to network planning and system operation possible, even where
there are hundreds or even thousands of small, distributed generators.
Control of distribution networks has also been enhanced by new technologies for
network automation. Improvements in automated sectionalising and on-line tap-
changers provide new options. The developments can be characterised as bringing
transmission technologies to the distribution level.
Finally, advances in control system design and other algorithms and methods make
possible more active and more precise control of numerous and distributed resources.
3.2. Commercial and Regulatory Drivers in Distribution Networks
In the final decades of the 20th century, commercial and regulatory drivers have
probably had a greater effect on the way distribution companies are run than any
advances in technology [3.1,3.6]. Distribution companies have been influenced by
changes in the electricity industry specifically, the energy industry more broadly, and
ultimately in the way large industries and essential services are perceived by
governments and markets across the world.
The privatisation of state-owned industries and associated restructuring, re-regulation

and introduction of competition has had a profound effect on the energy sector and
the electricity industry in particular. As natural monopolies, as providers of an
40
essential service, and as the part of the industry closest to customers, distribution
companies have probably been shielded more than any other part of the industry
from these changes; but their effect has still been felt. Where industry restructuring
has been taken furthest, the monopoly role of distribution network operator has been
clearly separated from other roles that can be made open to competition. However,
the approach to network planning has remained essentially the same. The knowledge
model of the conventional approach is still valid although some of the details, such as
the requirements of standards and regulations, have changed.
For example, the regulatory regime imposed upon a distribution company determines
the incentives and goals for company managers. This will influence the approaches
taken to asset management and network performance. The push for greater returns
on investments and a new focus on shareholder value has undoubtedly resulted in
some improvements in efficiency and value for money in distribution companies, but
it has also resulted in investment in network assets being reduced to an unsustainable
level [3.1]. With distribution network operators cut back to the bare minimum just to
maintain their existing system, resources are not currently available for the changes
that are required to accommodate distributed generation.
New regulations and government policies on the environment and energy, allied to
technical advances and market opportunities, have resulted in the growth of
renewable energy and combined heat and power. These distributed generation
technologies present a range of new challenges to distribution companies, as
discussed above. If distribution companies are going to play their role in meeting
government objectives, then appropriate commercial and regulatory incentives must
be put in place [3.30].
Governments still exert considerable control and influence over their domestic
energy sectors, including the electricity industry. However, more supranational
agreements on energy and other areas of commerce, such as those within the
European Union (e.g., the directive on the promotion of electricity from renewable
sources [3.19]), mean changes can be imposed on distribution companies from
41
outside the national sphere. Furthermore, the globalisation of markets in equipment
and services means that new technologies or methods in one country can quickly
have an impact in other countries. Effects can also be felt due to changes in policy
and regulation in other countries. For example, the UK government chose not to
invest in wind power but the governments of other countries, like Denmark and
Germany, did. This resulted in cost effective wind turbines becoming available and
having an impact on the UK, despite no significant push for this from the UK
government, who now have to react to this external technology driver.
3.3. Future Directions in Distribution Networks
The technology, commercial and regulatory drivers discussed above are producing
changes in the way electricity distribution networks are planned, managed and
operated. These changes require the application of new tools and methods, such as
those presented in later Chapters. It is possible to identify some specific concepts
and ideas now coming to the fore in distribution network development. These can be
split into those concerned with network architectures and operation, and those
concerned with institutional and organisational changes.
3.3.1. Network Architectures and Operation
The advances in technology and the changing demands of customers and other
stakeholders are being reflected in changes in distribution network architecture and
operation [3.20, 3.21,3.31]. Most simply, there are ongoing changes in design
practice. For example, previously distribution feeders were tapered to reflect the
lower power flow further down the feeder. But this approach places a constraint on
the possibilities for connecting DG at the end of the feeder. If power flow on
distribution networks can no longer to be assumed to be uni-directional then issues
like circuit utilisation and feeder tapering must be re-assessed. Many changes have
been handled in the past and the knowledge model of the conventional approach
makes explicit the consideration of new products and concepts. However, the
combination of changes in different areas, and in particular the integration of DG, is
42
having an impact on network architectures and operation that is much more
significant than the gradual change of the last half century.
3.3.1.1. Protection and Control
Safety is of primary importance in electricity distribution, and protection systems

must be proven and reliable. Protection is technically complex and can be
commercially controversial when it is concerned with the interface between a
network and a generator or customer. This can place constraints on new
developments and changes in the way things are done. However, there is plenty of
innovation being reported and there is scope for developments in many areas of DG
protection [3.23]. The major issues are concerned with detecting loss of mains and
island conditions. The possibility of exploiting improved communications is one of
the main avenues being explored. To make a significant impact, there must be a
standardisation of approaches and solutions in DG protection.
The connection of dynamic devices, like generators, to distribution networks present

a number of challenges but also offer opportunities to enhance network performance.
Modern control design methodologies mean the numerous and distributed resources
in electricity networks can be effectively controlled to meet local or global
objectives, as discussed in the following section. In addition, there are opportunities
with smaller DG installations for protection and control systems to be integrated in
combined interface units that provide cost-effective control without sacrificing the
safety provided by protection [3.24].
Larger DG installations have a larger impact on the network and, apart from their
own protection and control, will necessitate changes to the DNO’s protection and
control systems. This presents a number of challenges to network operators,
including wider variations in power flows and conditions on the network, which may
require the introduction of more adaptive protection and control [3.25]. These new
technologies can only be implemented if the DNO has a clear understanding of them
and is confident that they will operate as intended. This results in the need for more
analysis and new modelling tools, as discussed in later Chapters.
43
3.3.1.2. Active Distribution Networks
Conventional distribution networks are described as passive because there is very

limited on-line control of the network and almost all connections are for consumers.
The expansion of distributed generation introduces energy sources into distribution
networks, which in turn will require more active control of those networks. Thus,
there will be a transition to active distribution networks [3.27]. Active distribution
networks require monitoring and control akin to that currently used in transmission
networks. Distribution network operators will have to manage generator-network
interaction and a range of system issues like constraints, outage co-ordination,
stability and security, and system recovery and restoration. This will require a new
set of capabilities, in direct network control issues like voltage and power flow, and
in network simulation and analysis like state estimation and real-time studies. All of
these new requirements highlight the need for DNOs to adapt to the new challenges.
In most cases, the transition to active distribution networks is likely to be gradual.

Over time, more capabilities will be introduced as and when they are needed. This is
necessary because of the massive installed asset base and need to maintain service at
a level expected by existing users of the network. Network technologies like fault
current limiters, cancellation current transformers and Statcoms will gradually be
introduced along with greater monitoring of the network [3.26,3.27]. Implementing
changes like this will also require an expansion in the number of people working in
DNOs as well as improvements in their productivity through the use of new methods
for planning and management of information as discussed in later Chapters.
With the prospect of hundreds or even thousands of distributed generators added to a

distribution network, to help keep additional costs to an acceptable level it will be
necessary to utilise automated controllers that can maintain satisfactory network
conditions without the need for human intervention. One such concept under
development is the Power Quality Management System (PoMS) being developed in
the DISPOWER project [3.28]. These devices will monitor the conditions on low
voltage networks and instruct local generators to act in certain ways. For example,
44
some generators might be instructed to mitigate harmonic effects on the network, or
generator outputs may have to be adjusted to operate within network constraints.
3.3.1.3. Semi-Autonomous Networks
As mentioned above, active distribution networks will develop through the gradual
upgrading of existing networks. Where no network exists, as in a new building
development, there is an opportunity to design the energy supply systems from the
ground up. This provides an opportunity to exploit the latest technologies to their
fullest. For example, new residential areas are being built that include community-
run renewable energy sources, with the accompanying distribution network designed
to accommodate the generators.
Some new industrial facilities require a higher standard of power supply than
normally provided. The concept of “custom power parks” has been proposed to
supply these customers with the energy services they need [3.29]. Custom power
parks would utilise a range of technologies like DG, energy storage and power
electronics to provide the high power quality and reliability demanded by some
sensitive industrial processes.
Community-run renewable schemes and custom power parks exhibit some degree of
independence from the public electricity network, taking management and perhaps
ownership of the network away from the local distribution company. Such schemes
may be grouped under the broad heading of semi-autonomous networks. They will
still be connected to the main distribution system but will be operated with some
degree of autonomy. Where generation sources make it possible, such networks may
be operated completely in isolation from the main system.
3.3.1.4. DC Distribution
The improvements in power electronics make possible new ways of using direct
current rather than conventional alternating current [3.21,3.29]. This includes the
conversion of energy from direct-power sources like photovoltaics and fuel cells.
Such conversion is necessary to connect these sources to the main AC system. But if
45
DC sources become widespread then there is the possibility of connecting them
together by DC and even supplying some loads through a DC network. AC-DC-AC
conversion makes it possible to decouple multi-frequency systems, increasing the
independence of semi-autonomous networks.
3.3.2. Institutional and Organisational Changes
There have been many institutional and organisational changes forced on the
electricity industry in the course of restructuring and re-regulation. But given the
pressures on distribution companies now, there are likely to be further changes.
These can be considered in three categories: ownership and responsibility; energy
markets; and regulation and legislation.
3.3.2.1. Ownership and Responsibility
Across modern, industrial economies there are numerous models for the division of
ownership and responsibility within the electricity supply industry [3.11, 3.31].
Electricity distribution is a natural monopoly but distribution companies sometimes
also own generating assets and sometimes also perform the marketing role of selling
electricity to customers. To ensure fair competition in distributed generation,
distribution companies might have the responsibility of providing open access to
their whole network. This might conflict with their generation interests. If
distribution companies are not permitted to own generation and must act as an
independent system operator then some opportunities for integrating the network and
distributed resources might be lost. Governments and regulators must consider these
issues and specify ownership and responsibility as they see fit.
In some countries, distribution companies have used outsourcing and sub-contracting

to shift many aspects of distribution network operation to other companies.
Regulators have encouraged this to free up as many areas as possible to competition.
Distribution companies can then focus on the core, monopoly activities of asset
management, network operation and planning. In the future, it is likely that even
more activities will be outsourced. For example, a separate company under contract
to the distribution company may perform network maintenance. Likewise, external
46
groups could perform network analysis. Network operation need not be tied to asset
management and these activities could be performed by two separate entities within
one geographic area. However, experience in other industries, particularly the
railways in the UK, may make regulators reluctant to divide responsibilities between
too many different bodies.
The possibility of semi-autonomous networks suggests that outsourcing and sub-

contracting might be taken to its fullest degree, i.e. parts of the distribution network
might be divested and run entirely by a separate commercial enterprise or community
group.
In many countries, formerly state-owned electricity industries have been privatised

but great diversity remains in the ownership and responsibilities of distribution
companies. For example, many are owned and operated by a municipal authority.
However, further changes are possible. The pressures of new technologies and
stakeholder requirements might result in distribution companies being taken back
into public-sector ownership.
3.3.2.2. Energy Markets
The drivers impacting on distribution companies, like distributed generation, will

also have an impact on energy markets [3.10]. The development of open market
environments will have to take account of network constraints, especially where
generation is distributed around the network. Energy markets will also need to
consider the participation of renewable sources and energy storage. Some
intermittent renewable sources may want to participate in the market in combination
with others, forming hybrid systems, e.g. wind power combined with gas turbines.
In facilitating a market for distributed resources, even the smallest units should be
given the opportunity to trade.
The semi-autonomous network concept opens the way for new ways of operating and
managing parts of the network. It also offers new ways for managing the trading of
energy within a semi-autonomous network and with the rest of the market. It may be
47
possible to operate a conventional optimisation strategy within an area, which then
trades with other areas, generators and customers in a market structure.
It should be remembered that electricity is just one component of the energy mix that
consumers use to satisfy their needs. Wider use of gas-fired combined heat and
power will result in greater links between the electricity and gas markets. Renewable
sources are already being encouraged in some areas with the development of markets
in renewable credits, which can be traded separately from energy. Thus, the
relationships between different markets are likely to become more complex as a
result of the changes taking place [3.31].
3.3.2.3. Regulation and Legislation
To keep up with the changes taking place in the electricity industry, governments
will have to update regulations and legislation. The accommodation of distributed
and renewable resources will require changes to market rules and to engineering
standards and recommendations. For example, in the UK the government and the
regulator have established industry working groups to address a broad range of issues
[3.30, 3.31].
Legislation and regulations must take account of the different organisation structures
that might emerge, including the possibility of semi-autonomous networks run by
parties separate from the local distribution company.
Change can only happen if there is a political will. Furthermore, the decisions of
politicians to a large extent determine the directions taken in electricity supply.
There are opportunities to move toward a more environmentally friendly electricity
supply, but if the market mechanism discriminates against renewables then this
opportunity will be lost. Thus, political decisions are required on the priorities for
the industry.
It should be noted that changes to regulation and legislation take considerable time.
Technology and commercial drivers may push the industry quickly in directions not
48
favoured by the government and then the introduction of regulation and legislation
would become much more difficult.
This Chapter has explored the drivers and new directions in distribution networks,
both from a technical perspective and a commercial and regulatory perspective.
Distributed generation was identified as the single biggest challenge facing DNOs at
the start of the 21st century. The impact of DG will be wide-ranging, encompassing
technical, economic and environmental issues. Other technologies that will impact
on networks in a variety of ways are energy storage, demand side management,
power electronics and communications and control. A shift in the commercial and
regulatory environment to something more liberal and market-oriented has already
had a profound impact on DNOs.
DNOs will move in new directions both in network architecture and operation, and in
institutional and organisational structures. Integrated protection and control, active
distribution networks, semi-autonomous networks and DC distribution may all
feature in the distribution networks of the future. And there will be ongoing changes
in ownership and responsibility, energy markets and regulation and legislation.
These changes will go some way towards meeting the new challenges. However, as
explained in the next Chapter, there are a number of shortcomings in the
conventional approach to distribution network planning that will have to be
addressed to enable DNOs to successfully meet the new challenges and move in new
directions. Later Chapters present a number of methods that could help DNOs meet
the challenges facing them and will enhance the conventional approach described in
the knowledge model of Chapter 2.
49
3.1. Energy Networks Association; www.energynetworks.org; “Key facts &

figures”
3.2. Jenkins,N., Allan,R., Crossley,P., Kirschen,D., Strbac,G.; “Embedded
Generation”; 2000; The Institution of Electrical Engineers; ISBN 0852967748
3.3. Willis,H.L., Scott,W.G.; “Distributed Power Generation. Planning and
Evaluation”; 2000; Marcel Dekker,Inc.; ISBN 0 8247 0336 7
3.4. Rogers,W.J.S.; “Impact of Embedded Generation on Design, Operation and
Protection of Distribution Networks”; IEE Colloquium on The impact of
embedded generation on distribution networks, London, 15 October 1996;
Digest No:1996/191
3.5. N.Hadjsaid, J.F.Canard, F.Dumas; “Dispersed generation impact on
distribution networks”; IEEE Computer Applications in Power, Vol.12, No.2,
April 1999, p22-28; ISSN 08950156
3.6. Ault,G.,W.; “A Planning and Analysis Framework for Evaluating Distributed
Generation and Utility Strategies;” PhD Thesis, September 2000, Centre for
Electrical Power Engineering, Department of Electronic and Electrical
Engineering, University of Strathclyde
3.7. Electricity Association; “Engineering Recommendation G.59/1 -
Recommendations for the connection of embedded generating plant to the
regional electricity companies’ distribution systems”; 1990
3.8. DTI, Ofgem; “First Annual Report of the Distributed Generation Co-ordinating
Group (2001/2002)”; March 2003; www.distributed-generation.gov.uk
3.9. Alderfer,R.B., Eldridge,M.M., Starrs,T.J.; “Making Connections. Case Studies
of Interconnection Barriers and their Impact on Distributed Power Projects”;
National Renewable Energy Laboratory, NREL/SR-200-28053, May 2000;
www.eren.doe.gov/distributedpower/barriersreport/
3.10. Department of Trade and Industry (United Kingdom); “Energy White Paper:
Our energy future – creating a low carbon economy”; February 2003
3.11. The European Commissions AGORES project (A Global Overview of
Renewable Energy Sources), www.agores.org
50
3.12. The United States Department of Energy, Energy Efficiency and Renewable
Energy site, www.eere.energy.gov
3.13. The Distributed Generation Information Centre run by Resource Dynamics
Corporation in the US, www.distributed-generation.com
3.14. The United Kingdom Department of Trade and Industry Energy Group site,
www.dti.gov.uk/energy
3.15. Ribeiro,P.F., Johnson,B.K., Crow,M.L., Arsoy,A., Liu,Y.; “Energy Storage
Systems for Advanced Power Applications”; Proceedings of the IEEE, vol.89,
no.12, December 2001
3.16. Price, A., Bartley, S., Male, S., Cooley, G.; “A novel approach to utility scale
energy storage”; IEE Power Engineering Journal, June 1999, p.122-9
3.17. Roscoe, A.; “Demand response and embedded storage to facilitate diverse and
renewable power generation portfolios in the UK”; University of Strathclyde,
MSc dissertation, 2004;
http://ftp.strath.ac.uk/Esru_public/documents/MSc_2004/roscoe.pdf
3.18. Foote,C.E.T., Roscoe,A.J., Currie,R.A.F., Ault,G.W., McDonald,J.R.;
“Ubiquitous Energy Storage”; International Conference on Future Power
3.19. Directive 2001/77/EC of the European Parliament and of the Council of 27
September 2001 on the promotion of electricity produced from renewable
energy sources in the internal electricity market; Official Journal L283 ,
27/10/2001 P.0033-0040
3.20. Taylor,T.M., Willis,H.L., Engel,M.V.; “New considerations for distribution
network planning and design”; CIRED 1997; London, UK; IEE Conference
Publication no.438; p.6.1.1-6.1.5; 1997
3.21. Bergman,S.; “Visions of future Distribution Systems”; CIRED 1997; London,
UK; IEE Conference Publication no.438; p.6.4.1-6.4.4; 1997
3.22. Mott MacDonald and BPI for Ofgem; “Innovation in Electricity Distribution
Networks”; Final Report, March 2004
3.23. Ault,G., Booth,C., Dysko,A., McDonald,J., Banks,R., Cooke,R., Sasse,C.,
Stockton,M.; “Opportunities for a New Generation of Protection Devices for
Distributed Generation”; Proceedings of the Second International Symposium
51
on Distributed Generation: Power System and Market Aspects, 2-4 October
2002, Stockholm, Sweden
3.24. Sheaffer,P.; “Overview of Currently Available UIT Systems”; U.S. Department
of Energy Universal Interconnection Technology Workshop, July 25-26, 2002,
Chicago, IL
3.25. Brahma,S.M., Girgis,A.A.; “Development of adaptive protection scheme for
distribution systems with high penetration of distributed generation”; IEEE
Transactions on Power Delivery, January 2004, vol.19, issue.1, p.56-63, ISSN
0885-8977
3.26. Hart,D.G., Uy,D., Northcote-Green,J., LaPlace,C., Novosel,D.; “Automated
Solutions for Distribution Feeders”; IEEE Computer Applications in Power,
October 2000, vol.13, no.4, p.25-30; ISSN 0895-0156
3.27. Collinson,A., Dai,F., Beddoes,A., Crabtree,J.; “Solutions for the Connection
and Operation of Distributed Generation”; July 2003, DTI/Ofgem Technical
Steering Group Workstream 3 – Short-Term Solutions, Distributed Generation
Co-ordinating Group (www.distributed-generation.gov.uk)
3.28. Bertani,A., Bossi,C., Delfino,B., Lewald,N., Massucco,S., Metten,E.,
Meyer,T., Silvestro,F., Wasiak,I.; “Electrical Energy Distribution Networks:
Actual Situation and Perspectives for Distributed Generation”; 17th
International Conference on Electricity Distribution, CIRED 2003, 12-15 May
2003, Barcelona, Spain
3.29. Hingorani,N.G.; “Introducing custom power”; Spectrum, IEEE , Volume 32,
Issue 6, June 1995, p41-48
3.30. Ofgem; “Distributed generation: A review of progress”; January 2003; Ofgem
02/03; (and Open Letter from Callum McCarthy to DNOs, January 2003)
3.31. Botting,D.; “Technical Architecture – A First Report, The Way Ahead”; IEE
Power Systems and Equipment Professional Network, sponsored by the
DTI/Ofgem Distributed Generation Coordination Group, 23 December 2004,
DGCG 3/05
52
4. Shortcomings in the Conventional Approach and the Need for
New Methods
A large number of shortcomings and areas for improvement have been identified in
distribution network planning, by others [4.1, 4.2] and in the previous Chapters.
Some of these are long-standing problems that still pose problems and result in
inefficiencies; for example, inaccuracies in load forecasting and difficulties in
minimising total lifetime costs. Other deficiencies in planning methodologies have
arisen from changes in the electricity industry, as discussed in the previous Chapter,
and planning methods not keeping pace with those changes.
Just as shortcomings can be identified, it is also possible to identify desirable features

of methods for distribution network planning. It has been argued that it is of crucial
importance to evaluate planning methods from diverse fields and adopt “best
practice” for distribution network planning [4.3]. In summary, distribution network
planning methods should satisfy the following [4.2]:
• Handle multiple criteria and be decision focused
• Use appropriate time scales and planning horizons
• Enable consideration of multiple and diverse solutions while being able to
provide whole system solutions
• Enhance planner productivity and provide robustness to limited planning
resources
• Be modular in terms of access to analytical components and provide means of
integrating analytical modules and interfacing with other applications
• Make appropriate use of computer-based tools including simulation,
optimisation, graphical interfaces, bulk data handling facilities and mathematical
decision techniques
• Be automated or interactive as appropriate
• Manage uncertainty and deal explicitly in terms of risk
• Provide insight to the planning problem and solutions
• Produce tractable planning records to facilitate reuse of data, models, solutions
and planning rationale and provide leverage to future activities
53
Following the assessment of the conventional approach to distribution network
planning in Chapter 2, and new drivers and directions in Chapter 3, specific
shortcomings in distribution network planning were identified in five broad areas:
• The use of conventional technologies and methods
• The need for more analysis
• Organisational separation and loss of control
• Difficulties in formulating strategies and making decisions
• The need for knowledge management
These shortcomings and their implications are discussed below.
4.1. The Use of Conventional Technologies and Methods
As a mature industry, electricity distribution is characterised by a relatively slow rate

of change of technology. While great improvements have been made over the years
in overhead lines, underground cables, transformers, switches and circuit breakers,
the fundamental design principles of electricity networks have remained the same.
The need for low costs, high reliability and unquestionable safety has limited the
scope for experimentation. Thus, the conventional approach to planning and design
assumes that conventional technology will be used. However, DG, energy storage
technologies, solid-state devices and expanding capabilities in communications and
control are exerting a growing influence. A special effort is required to consider, let
alone incorporate, these novel technologies. Typically, DNOs will select from a
limited number of technically acceptable designs based on what has been done
before. New technologies present more risk and may be more expensive initially.
DNOs may lack the incentives necessary for a realistic business case to adopt new
technologies to be made. Also, new planning and design methods are required to
examine the impact of new technologies and exploit their use.
This issue will be of particular importance in the near future because in the
developed world, many distribution companies have an ageing asset base that will
largely be replaced over the next decade or two. Cutbacks in investment have
54
exacerbated the problem of ageing assets that will require wholesale replacement.
New technologies will offer some opportunities to extend asset life but in replacing
assets full account must be taken of the new demands that will be made of
distribution networks, such as the integration of DG.
The use of conventional technology is noted in the knowledge model of the

conventional approach where the “Design Network Additions and Alterations” task
assumes the use of conventional solutions. This shortcoming justifies the use of new
approaches to encourage solution-neutral thinking and the incorporation of new
technologies. In particular, the consideration of engineering design theory and the
use of a formal MCDM framework encourages the identification and consideration
of novel solutions, as will be explained further in Chapters 5 and 6.
However, there are considerable barriers to be overcome. DNOs are slow to adopt
new technologies for a range of reasons. Principal among these is cost effectiveness.
Many new technologies offer slight improvements in performance or flexibility but,
until they are widely adopted, remain too expensive. The use of new technologies
will only occur where DNOs feel that they fully understand the technology. This
provides justification for new modelling and analysis, such as that discussed in
Chapter 9, and enhancement of DNOs’ abilities to absorb the outcomes of research
and development. The use of radically different approaches also has to overcome the
inertia of familiarity and corporate procedures as well as a culture built around the
conventional approach to network planning.
4.2. The Need for More Analysis
One of the primary challenges facing distribution companies is the modelling,

simulation and analysis of network performance and behaviour. New technologies,
such as DG and power electronics, require the development and validation of new
models. And with different incentives for stakeholders in the network and new ways
of operating, new types of analysis will have to be performed. DNOs need not be the
ones who develop new models and analysis methods – this task can be undertaken by
academia or other external service providers – but they must have the capacity to
55
absorb the outcomes and understand the results of new analyses. In line with
changes in the industry and regulation, distribution companies must also conduct
more comprehensive economic and financial evaluation. This includes the proper
consideration of externalities like environmental factors. Thus, the analysis that must
be conducted by distribution companies is both more extensive and more complex.
This poses a problem because conventional analysis methods are too expensive, in
terms of time, talent and other resources. High-level planning decisions are
sometimes made without sufficient technical analysis. For example, lower voltage
assets are treated as numerous enough and cheap enough to be replaced based on
statistical analysis. New connections are sometimes agreed without detailed analysis
of the implications. Generally, there is a different approach to major and minor
projects due to the high expense of investigation. It is assumed that the system can
accommodate small changes but limits are eventually reached. If the burden
imposed by technical analysis was smaller then decision-makers could be better
informed about the implications of particular courses of action.
In the past, distribution networks have been over-designed and their components
over-sized. This was due to a number of reasons. The additional capacity and
security provided cover for the uncertainties of demand predictions. The acceptance
of generous margins also made it possible to apply rules of thumb and use prepared
tables and charts in design. However, the excess capacity built into the networks of
the past has been taken up by load growth, and reduced expenditure in recent years
means the excess capacity has not been replaced. Pressures on DNOs now mean that
networks must be designed more precisely with less, potentially useless, spare
capacity. This requires new rigour in the analysis of requirements and design of
solutions.
Power system simulation and analysis is an important activity in the design and
management of modern electricity supply systems. Simulation is used to ensure that
the required standards of security and stability are met and that system design is
optimised. Power system simulation can be a difficult and expensive task, requiring
56
considerable knowledge and experience to perform detailed analyses. There are a
wide variety of simulation tools and huge libraries of models and data available; and
there are many approaches that may be taken to achieve a specific goal. As
discussed, modern electricity systems are undergoing considerable change with the
introduction of new technologies like distributed and renewable generation, and this
is imposing an ever-greater simulation burden on the industry.
“Network modelling and analysis” appears as a method at a number of points in the

knowledge model of the conventional approach. This highlights its importance and
the fundamental role it plays in network planning. The new challenges demand even
more analysis, highlighting a significant shortcoming in the conventional approach.
One way of tackling this shortcoming is through improved information management;
this is discussed below in Chapter 7.
DNOs must be able to accommodate and analyse the new technologies being
connected to their networks, or at least be able to understand and apply the results of
analysis by others. Chief among these is doubly fed induction generators (DFIGs),
the most common generator technology used in new wind farms. Assessing the
impact of DFIGs on the network requires simulation and analysis and this requires
the development of new models that are acceptable to DNOs and others in the
industry. Chapter 9 describes the dynamic modelling of wind farms, in particular the
development of new models of DFIGs and associated systems. This illustrates the
difficulties faced by DNOs and the industry as a whole when trying to introduce
analysis tools for new technologies.
4.3. Organisational Separation and Loss of Control
Across the world there is great diversity in the structure, ownership and management
of electricity supply industries. But where industry restructuring and re-regulation
has taken place there has been a general shift towards organisational separation and a
consequent break down of old lines of control and influence. This is perhaps most
keenly felt at the distribution level with distribution companies expected to maintain
57
the same standards of quality and service while facilitating open access to new
market participants.
With responsibilities changing and private investors keen to exploit new

opportunities, the role of the central planner has diminished, or even disappeared.
DG and private distribution networks interconnected with franchise utility networks
are starting to influence distribution system planning effectiveness. This is one of
the key areas of concern among DNOs in the UK [4.2].
Furthermore, when distribution companies were part of large, vertically integrated

entities or state-owned, they perhaps benefited from some subsidy in research and
development. With the core DNO role isolated from other parts of the industry, and
a constant pressure on cost reduction, distribution companies may sacrifice
investment in research and development – to the long-term detriment of society.
The knowledge model of the conventional approach does include as a level one task:
“Maintain Communication and Exchange Information with Others”. In the new
environment this is becoming increasingly challenging. One way to address this
challenge is through improved information management, with the representation of
models and data being an area where particular enhancements are possible, as
discussed in Chapter in 7.
4.4. Difficulties in Formulating Strategies and Making Decisions
With all the changes in technology, regulation and the commercial environment, one
of the primary challenges faced by distribution companies is the formulation of
strategies. With so many new issues to contend with, within an ever-changing
incentive structure, managers and engineers need new methods to support decision
making.
Conventional approaches to electricity distribution network planning are often

deficient in their accommodation of multiple objectives or multiple criteria. The
multitude of optimisation algorithms that have been proposed [4.4] are typically
58
based on a three-pronged approach addressing load forecasting, substation site and
size, and feeder route and capacity. These algorithms rely on a limited expression of
requirements. They might minimise costs – calculated as a function of substation
and feeder type and capacity – with predicted customer demand profiles as the goal
to be satisfied; statutory and company design standards might be included as
constraints. An optimisation algorithm focusing on a limited number of factors
might be used to produce the first version of a plan then constraints imposed to take
account of externalities like environmental issues. Optimisation methods may also
be used to define design standards, which are then applied in a wide variety of
circumstances.
Planning and design would be improved by explicitly including more factors

amongst the objectives considered in the first place. Attempts have been made to
extend optimisation algorithms to take account of additional issues, reliability being
one example, but it can be difficult to translate continually changing regulatory,
environmental and financial drivers into tangible engineering objectives. In
electricity distribution, there are many stakeholders with different and ill-defined
objectives so the emphasis must be on finding an acceptable solution rather than an
optimum. For this reason, conventional optimisation algorithms are inappropriate
[4.5]. Greater network complexity will necessitate the explicit consideration of even
more issues, reducing further the value of limited optimisation algorithms.
In general, in the conventional approach to distribution network planning and design,

there is a lack of support at the higher decision making level. There have been some
recent advances in the integration of computer-based tools but integration and
interpretation of results is still mainly a manual task. This shortcoming justifies the
application of new methods, including MCDM, as discussed in Chapter 6, and
scenario analysis, as discussed in Chapter 8.
4.5. The Need for Knowledge Management
As with other engineering enterprises, DNOs rely on the knowledge of their staff.
Information systems, including databases and documentation, should be maintained
59
and kept up to date to be of use. But some DNOs have, due to pressures of time and
money, not maintained these systems, some of which date from the days of the
nationalised industry structure. Instead, the knowledge and experience of staff is
relied upon to compensate for missing data or inaccuracies. The loss of knowledge
as people move can be damaging. In particular, centralisation of DNO functions has
resulted in the loss of people with in-depth knowledge of a specific geographical
area. This local knowledge used to compensate for missing or poor quality
information.
Knowledge management is an important issue for DNOs and exposes a shortcoming

in the conventional approach. The methods used in planning and design must take
account of this by facilitating the capture of the reasoning behind decisions as well as
the decisions themselves. The knowledge modelling methodology used in Chapter 2
to describe the conventional approach to network planning demonstrates how
knowledge can be represented in a cohesive and navigable form. The application of
the knowledge management techniques would facilitate the capture of knowledge
from important staff, helping to spread that knowledge and provide insurance against
their departure. This shortcoming is addressed in Chapter 5 below, where methods
for the management of rationale in planning decisions are proposed. In particular,
the use of structured decision making is highlighted, justifying further the application
of MCDM, discussed in Chapter 6.
Following discussion of the conventional approach to distribution network planning

and the identification of drivers and new directions, this Chapter highlighted some
specific shortcomings in the approaches to planning and decision making.
The use of conventional technologies and methods is a shortcoming because it limits

the scope for innovation and adoption of new solutions. New technologies and
methods will be required to meet the new challenges facing distribution planners.
And efforts will be necessary to ensure that the new technologies that are available
are fully exploited by DNOs.
60
The new challenges in network planning result in a need for more analysis that
DNOs are not equipped to perform. This shortcoming in analysis resources must be
addressed through improved methods and greater efficiency. For example, new
models are required for new technologies like DFIGs used in wind farms.
Changes in regulation and business structure have led to organisational separation

and loss of control. This has become a shortcoming in planning because it disrupts
the flow of information and the capability to reach decisions that satisfy the diverse
requirements of different stakeholders.
A combination of new challenges results in difficulties in formulating strategies and

making decisions. This is at the core of electricity network planning so is a
considerable shortcoming for DNOs.
There is a need for knowledge management because it offers a way of improving

performance and efficiency in distribution planning, thereby supporting other
activities that will help to address the other shortcomings.
A number of specific areas have been identified in which the planning process could
be improved to meet the new challenges being faced by DNOs. Addressing these
shortcomings requires methods from other domains to be applied. This provides the
justification for the work presented below, which demonstrates a number of methods
that can help DNOs update their processes and face the new challenges in network
planning. Specifically, the methods proposed are engineering design theory, decision
support, information management, and scenario analysis. The following Chapters
explain how these methods might be applied but it should be noted that the
complexity and diversity of distribution network planning and the range of new
challenges to be faced mean that highly prescriptive and detailed methodologies are
inappropriate because they are inflexible.
61
4.1. Berrie,T.W.; “Electricity Economics and Planning”; 1st edition, London:

Peregrinus, 1992
4.2. Ault,G.W., Foote,C.E.T., McDonald,J.R.; “Distribution system planning in
focus”; IEEE Power Engineering Review; January 2002, Volume 22, Issue 1,
p.60-62; ISSN: 0272-1724
4.3. Ault,G.W., Cruden,A., McDonald,J.R.; “Specification and testing of a
comprehensive strategic analysis framework for distributed generation”
Proceedings IEEE Power Engineering Society Summer Meeting 2000, 2000
4.4. Khator,S.K., Leung,L.C.; “Power Distribution Planning: A Review of Models
and Issues”; IEEE Transactions on Power Systems; vol.12, no.3; p.1151-1159;
August 1997
4.5. Miranda,V., Proença,L.M.; “Why risk analysis outperforms probabilistic
choice as the effective decision support paradigm for power system planning”;
IEEE Transactions on Power Systems; vol.13, no.2; p.643-648; May 1998
62
5. Engineering Design Theory
The application of methods from other domains can help improve the conventional
approach to distribution network planning and address some of the shortcomings
identified above. One source of ideas and methods that can contribute to the
adoption of best practice in distribution network planning is engineering design
theory [5.1, 5.2, 5.3].
Engineering design theory includes descriptive theories, such as the classification of

models and design level, and various prescriptive theories. However, design
theorists recognise that there is often a mismatch between theory and practice.
Prescriptive theories are normally problem focused, while the practical approach
adopted by designers is often solution focused. A purely systematic approach is of
limited value and is unrepresentative of how designers actually go about their work.
Design in practice is an interactive and recursive process that relies on conjecture as
much as problem specification.
Furthermore, design theory tends to be focused on product engineering in the

mechanical engineering domain. An electricity distribution network is a unique and
evolving entity that lasts for decades and undergoes continuous update. All
alterations must take into account the existing network, although each project that
alters the network could be viewed as a single product, albeit a one-off. The
continuous development of the network means decisions must also be made on the
best time to implement plans. In some instances, a problem will arise and plans will
be prepared to deal with it only to be abandoned when conditions on the network
change again and the problem ceases to exist and changes into something quite
different. Nevertheless, engineering design theory still offers some useful ideas on
how to deal with the challenges faced by distribution companies [5.4].
This Chapter examines a number of concepts and methods from engineering design
theory and applies them to distribution network planning in the context of the new
challenges and shortcomings identified above. There is a particular emphasis on the
63
concept of design rationale, which is identified as being of primary importance in
improving distribution network planning.
5.1. Solution-Neutral Problem Definition
The systematic engineering design approach involves clarification of a problem

followed by the development of conceptual designs before forming more detailed
designs. An important stage is expressing problems in an abstract, solution-neutral
way. The use of some modelling methodology is suggested to establish the overall
structure of the problem. This is supposed to lead to the development of novel
solutions, unconstrained by inherent preferences in the problem definition.
Distribution network planners lack the ability to express complex problem structures
in an abstract way, i.e. focus first of all on what customers really need and want
rather than making assumptions about how a problem will be solved. In addition,
there is often no incentive for DNOs to perform such an analysis. The conventional
approach to distribution network planning bypasses the solution-neutral expression
of problems and moves straight on to solution-oriented specifications. This is
illustrated in the knowledge model of the conventional approach where the design of
network additions and alterations is directly linked to the specification of
conventional solutions. From expected load profiles, characterising the requirements
of consumers in conventional terms, planners will immediately specify problems in
terms of solutions, e.g. overhead lines, cables and transformers. Furthermore, with
developments in DG, power electronics and differentiated quality and reliability,
load profiles no longer provide sufficient information to characterise requirements.
A more solution-neutral perspective would consider the overall service requirements
of customers then assess all the different options available for satisfying those
requirements.
Design theory advocates that the conceptual design stage of the systematic approach
should include some abstraction of the specification to promote innovation [5.2].
Distribution design, like many other fields of design, moves into solution terms
quickly, e.g. discussing particular transformer and cable sizes. Although this may
64
restrict innovation, there are good reasons for this approach. Standardisation and
consistency offer economies of scale and confidence in proven performance. These
are of particular value to distribution companies who must manage very large and
expensive networks. However, incorporating novel technologies may require some
abstraction to facilitate innovation. It will also require suitable incentives for DNOs
from a business perspective.
5.2. Level of Risk and Innovation
Engineering design theory uses the concept of design level to reflect the level of risk
and innovation in a design [5.1, 5.2]. Design level can be classified as REPEAT,
VARIANT, ADAPTIVE or ORIGINAL. Moving from one level to the next
represents an increase in product or process innovation and therefore an associated
increase in risk. The review of the conventional approach in Chapter 2 revealed that
electricity distribution network design is largely REPEAT or VARIANT with
gradual changes to the system and its components and low levels of risk and
innovation. However, incorporating new technologies in distribution networks may
require the greater innovation and risk of ADAPTIVE and ORIGINAL design.
Recognising this helps manage the risk in network development. By ensuring that
regulators recognise the level of risk, it is more likely that they will allow the
additional revenues necessary to innovate.
5.3. Decision Classifications
One of the many changes in the electricity industry is that the detailed design of new
additions or alterations to the network is now being done outside of the DNO. This
makes the definition of functional specifications even more important. Engineering
design has been described as a series of decisions classified as FUNDAMENTAL,
INTERMEDIATE and MINOR [5.3]. Distribution design has long been a case of
defining a strategy with fundamental decisions then making intermediate and minor
decisions for each project. This allows the benefits of standardisation and
consistency to be enjoyed. Nowadays, distribution asset managers may make the
fundamental decisions then leave the rest to external contractors. However,
65
increasing complexity may require more fundamental decisions at the project stage.
Distribution companies will either have to accept this devolution of design freedom
and decision making power, or revise their planning and design policies to deal with
the issues being faced. In any case, a rethink of the approaches to organising the
planning and decision making process is required.
5.4. Design Concurrency
The range of issues considered in distribution network planning must be broadened

to include externalities like damage to the environment. But it is also important to
consider the whole life cycle of a project. A powerful theme in modern design
theory is concurrency [5.1, 5.3]. This involves taking into account all aspects of a
design including manufacture, installation, maintenance, operation, failure and
retirement. This can be implemented with only slight modifications to the
conventional approach as detailed in the knowledge model of Chapter 2. The
“Identify Requirements” task already highlights the importance of taking different
factors into account. A more robust assessment of requirements, taking the full life
cycle into account, would provide the concurrency advocated by design theory. The
planning and design of distribution networks can benefit greatly from such changes
to the conventional approach. However, organisational separation, the downsizing of
distribution companies and associated subcontracting of functions, and the near-term
horizon of price control reviews may disrupt attempts to introduce concurrency to
planning and design.
5.5. Alternative Designs
Design theory supports the development of a number of designs at every stage of

development. The distribution network design literature supports this. Alternative
designs may be based on different solution concepts or may address the possibility of
different scenarios emerging. This helps decision-makers manage the risk associated
with predicting the future and offers them real choices in network design. Designers
are advised to develop a number of alternative designs at every stage, either based on
different solution concepts or to address uncertainty and the possibility of different
66
scenarios emerging [5.1]. As the range of distribution network technologies
expands, the spectrum of realistic solution concepts will grow. This will make the
selection task for planners and designers more difficult.
Growing uncertainty in the distribution business requires planning for different

scenarios. It is obvious that the most acceptable approach for large, long-life entities
like distribution networks is to design them to be robust to a range of scenarios,
rather than just to suit the scenario deemed most likely to occur. Modern thought
supports risk analysis approaches for dealing with uncertainty in distribution
planning rather than previously used probabilistic approaches [5.5]. This is
discussed briefly in Chapter 8.
5.6. Design Rationale
Documentation is of crucial importance in all types of engineering design.

Traditionally, this might have represented only the final designs. One aspect of
modern design theory is the capture of the rationale lying behind the final designs.
This should be of particular value in distribution networks where assets remain in use
for decades, long after the original designers have left. Design rationale offers a way
to combat the identified shortcoming of a need for knowledge management.
With a strict regulatory regime pushing revenues down and increasing focus on
performance and the environment, distribution planners and designers must provide
full justification for strategies and expenditure, both internally and externally.
Distribution companies have also been restructuring and reducing their number of
employees. A feature of this downsizing has been the subcontracting of distribution
utility functions, particularly construction, to focus on asset management.
Distribution companies must be prepared for all possible changes, in technology,
government policy, and regulation.
The concept of design rationale [5.6, 5.7, 5.8] is linked to information management
but has the goal of creating a “reasoning trail” not just a “data trail”. Design
rationales can include the argumentation and justification behind a design decision,
67
the alternatives considered, and the trade-offs evaluated. A design rationale system
is a tool for capturing design rationales and making them easily accessible. The
recording of rationale is most important for fundamental decisions that occur early
on and have the greatest leverage, or impact, on a project.
By helping to evaluate the issues and alternatives being examined, design rationales
can provide better design support and improve the formation and comparison of
strategies. This includes tracking dependencies among requirements and the
components that satisfy them. Rationales can help identify past designs that may be
relevant to current problems, or the rationales themselves may be of value.
Collaboration between groups can be improved by providing a common vocabulary
and project memory.
Almost anything, from formal specification documents to informal telephone

conversations, may contribute to rationale. Apart from problems with the volume of
material, anything represented and stored must be accessible so must have some
structure. Informal representations are easy to create but unstructured and difficult to
archive and retrieve. Formalising knowledge is costly and requires all objects and
relations to be defined as formal objects.
The benefits of design rationale will be increasingly valuable as distribution planning

and design becomes more complex. Distribution network assets typically have a life
longer than the careers of the people who plan and design the network. Combined
with modifications to the network made over time and restructuring of distribution
companies, planners and designers can lack valuable insight into earlier decisions.
By explaining design decisions, rationales can support modifications, which are
inevitable in the long life of distribution networks. The recording of rationales can
capture expertise and help tackle problems in the restructuring of DNOs.
Apart from supporting other planners and designers, design rationales can contribute
to documentation for managers and external groups, including the industry regulator.
Rationales may also be useful in supporting patent applications, were distribution
68
companies to seek intellectual property protection for network designs. It is
important that the recording of rationale and other forms of knowledge management
are tailored to the particular domain in which they are applied. One way of formally
recording rationale is to associate it with structured decision making methods, as
proposed below.
Design rationale theory [5.1, 5.3, 5.7] suggests that effective capture requires the
recording of four things:
1. The justification for decisions
2. The argumentation behind decisions
3. The alternatives considered in a decision
4. The trade-offs made in reaching a decision
Unfortunately, capturing rationale effectively and facilitating its reuse is not easy.
There is a conflict between capture and subsequent access. For example, an
engineer’s logbook might contain details of all the thought processes and information
used in reaching design decisions, ideally fully capturing the rationale. But
accessing this knowledge is difficult because of the unstructured and often untidy
nature of a logbook written by one person for his or her own use. The logbook may
be of great value to the engineer who wrote it but of limited value to others.
Different approaches can be adopted to address this conflict. In the theory of design
rationale, three perspectives are identified:
1. The communication perspective, where all naturally occurring discourse in
reaching the decision is captured
2. The documentation perspective, where rationale must be identified and recorded
in documentation
3. The argumentation perspective, where some sort of procedure or system is
implemented that captures rationale in a structured form throughout the decision
process
69
It is important that the time required by the user to document rationale is kept to a
minimum. Forcing the user to spend a great deal of time documenting rationale can
be counter-productive. The user must understand the benefits of capturing rationale.
This is to ensure that the user fully comprehends the type of knowledge that it is
important to capture, permitting them to be as concise as possible when inputting
knowledge. Given the requirements and potential problems in capturing rationale,
different approaches may be necessary for different circumstances and different
decisions.
In this section, three methods are proposed to support the capture and reuse of
rationale in distribution network planning. Firstly, structured decision making
methods encourage the formal identification of alternatives, objectives and decision
criteria. Attaching rationale to the assignment of weightings and scores, which can
be a source of opacity, can enhance these methods further. This combines the
communication and argumentation perspectives by introducing structured support for
the decision process that will change the ways decisions are made while forcing the
recording of decision discourse. Secondly, through an analysis of the planning task,
a number of generic or repeated justifications have been identified, which can form a
basis for the capture of rationale. This might fall under all three perspectives
depending on when and how the generic justification is identified and recorded.
Finally, knowledge modelling methods can be applied to the capture and reuse of
rationale. This is likely to involve posterior analysis of a decision and falls squarely
under the documentation perspective.
5.6.1. Structured Decision Making Methods
Structured decision making methods, such as multiple criteria and multiple objective
decision making, can help capture some of the elements that make up the rationale.
A structured decision making process will normally require the formal identification
of alternatives. It will also require the identification and quantification of decision
criteria or objectives. Even in the most limited methods, there should be
identification of plus-points and minus-points. The weightings and scores used in
structured decision making methods quantify the trade-offs made between
70
alternatives. However, the weightings and scores can appear arbitrary and opaque.
Assigning some rationale directly to the weightings and scores can ameliorate this.
The MCDM example in Chapter 6 illustrates how a decision can be analysed and the
reasons for selecting a particular option made explicit. This approach could be
further enhanced with the capture of rationale associated with each of the weightings
and scores in the decision analysis.
5.6.2. Generic Justifications
Instead of a free-text description of rationale, within structured decision making or

just in rationale capture in general, it should be possible to offer users a pre-set list of
generic justifications. These would be formed from the most frequently used
justifications behind decisions. These could form a basis for the capture of rationale
but would normally require additional information, either as additional free-text
description, or as reference to something providing background knowledge or
guidance.
Examples of generic or repeated justifications might include the following:

• Meeting a particular law or regulatory rule. Reference could be made to the law
or regulation in question.
• Following company strategy or policy, in business or engineering. Reference
could be made to the appropriate policy or strategy document.
• Selection based on cost. The cheapest option will often be chosen but the
rationale should make clear how the overall cost was calculated. Ideally, full
lifetime cost will have been determined.
• Due to the results of analysis and simulation. Studies might determine which
option is the best. Reference could be made to the results, ideally with
information on the tools and methods used.
• Exploitation of a concurrent opportunity. In some instances the best option will
depend on other activities on the system. Often, this will depend on and
influence the timing of actions. Reference could be made to the other
opportunities.
71
• Due to local reasons. This type of justification might be entirely correct but does
not provide much information. This would have to be expanded upon with
explanation of why local circumstances demand a particular solution.
• Due to past experience. Engineers learn to know which solutions are best in
different circumstances and past experience might be the primary reason for
selecting a particular option. Reference could be made to the earlier decision in
question. Ideally, this would have been based on an objective analysis that was
properly recorded.
Each of these generic justifications can provide a starting point for further
explanation of why a decision was made. They could also be used in combination.
As decision makers use the list, it would be tuned to the particular application and
organisation.
5.6.3. Knowledge Modelling Methods
Knowledge modelling methods provide a formal approach to capturing, structuring

and presenting knowledge. A knowledge modelling methodology was used in
Chapter 2 described the conventional approach to distribution network planning.
Such methods could be applied in the capture of rationale and the identification of
knowledge resources associated with particular planning decisions. This would be
conducted after a decision was made and would involve interviewing the decision
maker to identify:
• What decision was taken?
• What factors drove the decision?
• What were the outcomes of the decision?
• What knowledge sources were used in reaching the decision?
• What tools and techniques were applied and what additional knowledge did they
produce?
• Why was the final decision as it was?
The knowledge model of the conventional approach already includes a level one task
where such knowledge modelling methods might be applied: “Review
72
implementation to see that requirements are met”. The application of knowledge
modelling methods would provide scope for very detailed capture of knowledge
associated with a decision. This detail comes at considerable cost in time and effort
required to fully review a planning decision and process it according to the
knowledge management methodology. However, this investment would provide a
return in the form of faster or better decisions in future through exploiting the
captured knowledge.
Engineering design theory offers a range of concepts and methods that can be applied
to electricity distribution network planning to address the shortcomings in the
conventional approach and help meet the new challenges of the 21st century.
A selection of these concepts and methods was considered, with an assessment of

how each of them relates to and influences distribution network planning. The value
of solution-neutral problem definition was highlighted and contrasted with the
conventional approach to network planning, which moves into traditional solution
terms all too quickly. The concept of a level of risk and innovation was explored,
exposing conventional network planning as being mostly low risk whereas the
incorporation of new technologies will require higher-risk planning and design, a
departure from the norm in a risk-adverse industry. The concept of decision
classifications was used to highlight the benefits of standardisation and the potential
dangers in outsourcing. The concept of concurrency was identified as being
extremely useful but is already applied to some degree in conventional network
planning. Likewise, the method of producing alternative designs is already used in
the conventional approach but will have to be expanded to incorporate the possible
use of new technologies. All of these methods are useful only if DNOs are given the
appropriate incentives to change the way they do things.
The capture and effective representation of the rationale behind decisions offers great
scope for adding value to decisions and contributing to the collation of corporate
knowledge. The use of some structure or process for capturing rationale forces
73
decision-makers to make explicit the reasons for their decisions. This should lead to
better decisions but also provides a valuable resource for those facing similar
problems in the future. A novel combination of methods was suggested to assist in
the capture of rationale in distribution network planning: the use of formal MCDM
techniques; making reference to a defined list of generic rationales; and post-decision
analysis with knowledge engineering methods.
Among the concepts and methods considered, the descriptive theories provide a new
perspective on the conventional approach to network planning and the prescriptive
theories offer some ideas on new approaches. The development of the planning
function will benefit from the consideration of engineering design concepts like level
of risk while new procedures will have to be implemented at various stages of the
planning process to effectively capture decision rationale.
5.1. Birmingham,R., Cleland,G., Driver,R., Maffin,D.; “Understanding Engineering

Design: Context, Theory and Practice”; 1997; Prentice Hall Europe; ISBN 0-
13-525650-X; D620.0042UND
5.2. Pahl,G., Beitz,W.; “Engineering Design: a systematic approach”; The Design
Council; 1988
5.3. Starkey,C.V.; “Engineering Design Decisions”; 1992; Edward Arnold; ISBN
0-340-54378-7; D620-0042STA
5.4. Foote,C.E.T., Ault,G.W., Burt,G.M., McDonald,J.R., Green,J.P, “Towards a
structured methodology for distribution network design applications”, 35th
Universities Power Conference, Belfast, September 2000
IEEE Transactions on Power Systems; vol.13, no.2; p.643-648; May 1998
5.6. O’Shaughnessy,K., Sturges,R.H.; “A Systematic Approach to Conceptual
Engineering Design”; Proceedings of Design Theory and Methodology
DTM’92; American Society of Mechanical Engineers, Design Engineering
Division; p.283-291; 1992
74
5.7. Lee,J.; “Design Rationale Systems: Understanding the Issues”; IEEE Expert;
vol.12, no.3; p.78-85; May/June 1997
5.8. Shipman,F.M., McCall,R.J.; “Integrating Different Perspectives on Design
Rationale: Supporting the Emergence of Design Rationale from Design
Communication”; Artificial Intelligence in Engineering Design, Analysis, and
Manufacturing (AIEDAM); vol.11, no.2; p.141-154; April 1997
75
6. Decision Support in Distribution Network Planning
Perhaps the most important element of network planning is the reaching of decisions.
Decisions require information to be gathered and may rely on various forms of
analysis but tools are also available to support the actual decision making process. In
this Chapter, the use of multiple criteria decision making (MCDM) techniques is
discussed, highlighting their value in making explicit the identification,
quantification and analysis of decision criteria. Decision support is crucial to
addressing the difficulties in formulating strategies and making decisions, which was
identified in Chapter 4 as one of the main shortcomings in distribution network
planning.
The second half of this Chapter describes a case study that demonstrates the use of
MCDM techniques to assess the financial viability and technical desirability of a
number of DG options for a large industrial consumer. This includes the
specification and use of alternative ratios within the MCDM framework to provide a
new perspective and enhance the information available to decision makers. The
study demonstrates the assessment of novel solutions based on DG alongside
conventional grid reinforcement solutions. This is the kind of study that planners
will have to undertake to exploit the advantages of new technologies, as noted in
Chapter 4.
6.1. Multiple Criteria Decision Making
The purpose of MCDM techniques is to help a decision-maker to think

systematically about complex decision problems, and improve the quality of the
resulting decisions. Decision problems can involve a number of different criteria and
have several different courses of action or options available to the decision-maker.
When the number of options and criteria becomes large, the need for formal,
structured decision making techniques becomes apparent. It is not unusual to find
some of the criteria to be conflicting in nature, thereby complicating the problem
further. The use of MCDM techniques in such circumstances allows this process to
76
be tackled consistently, in a manner such that the most suitable option – if one exists
– is identified.
MCDM is proposed here as a means of addressing a number of the shortcomings

identified in the conventional approach to distribution network planning. It will be
shown that MCDM can help address the difficulties in formulating strategies and
making decisions, support the fair consideration of new technologies and facilitate
the effective capture of decision rationale and management of knowledge.
The conventional approach to network planning can be enhanced with MCDM by

applying it to the principal decision points in the knowledge model. On level one
these are “Design Network Additions and Alterations” and “Determine Programmes
of Implementation”. The design task can be enhanced by modifying the
conventional three-element, solution-oriented approach to produce something more
like the general structure for MCDM outlined below.
MCDM problems can be classified into two groups [6.1]:

• Multiple objective programming problems – where there is a very large or
infinite number of alternatives, which are described through the use of decision
variables.
• Multiple attribute problems – which have a relatively small number of
alternatives (typically 4-10), represented in terms of attributes.
There are many different multiple objective and multiple attribute techniques
available to a decision-maker. This presents the decision-maker with the problem of
selecting an appropriate MCDM technique. The ultimate decision on which
technique should be selected depends on a number of factors [6.2]:
• Number and type of objectives, criteria, alternatives and constraints
• Type of information the decision-maker wishes to input – some techniques
require more complex and sophisticated information from the user
• Decision making style and problem structure
77
• Time and effort the decision-maker wishes to spend on the problem – some
techniques reach a solution quicker than others and have a lower cognitive
burden
• Accuracy – most techniques cannot handle uncertain outcomes, and some do not
guarantee to produce a non-dominated solution or a complete ranking of all the
alternatives
• Restrictiveness of underlying assumptions
Even after the above factors have been considered the decision-maker may still be
left with several techniques that will produce the required outputs. In this situation,
personal preference or supporting software availability would become the deciding
factors.
6.1.1. General Structure for MCDM
As noted above, a wide range of MCDM techniques is available, each involving a

slightly different methodology. However, a broad assessment of the techniques and
consideration of their application in the distribution network planning domain
suggests a general structure for MCDM as shown in Figure 6.1 and described further
in the sections below. As with design theory, considering distribution network
planning from the perspective of this MCDM structure highlights ways in which the
conventional approach can be improved, as discussed in the sections below.
78
Figure 6.1 – General structure for multiple criteria decision making
DEVELOPMENT DEVELOPMENT
ISSUES OPTIONS
QUANTIFICATION
ANALYSIS
DECISION
MAKING
6.1.1.1. Identify “DEVELOPMENT ISSUES”
The first step in reaching a decision is to properly consider all the relevant issues.
For a technical study, this might be concerned with design specifications and
technical requirements. For more strategic analyses, this might include higher-level
economic, business and social issues. It is at this stage that planners must ensure that
they use appropriate time scales and planning horizons. This will be determined by
the task being undertaken; for example, asset replacement strategies must be assessed
in the long term while the response to an application for connection from a new
generator may demand changes to the network in the near-term. In defining the
development issues, planners can endeavour to incorporate the engineering design
theories of solution-neutral problem definition and concurrency. This should
encourage an objective assessment of the complete life cycle of the development
under consideration.
Having identified high-level issues, these can be translated into specific evaluation
criteria to facilitate the decision analysis. For example, an issue like power quality
can be translated into specific objectives and requirements for particular power
quality measures. This is completed more fully in the quantification stage.
79
6.1.1.2. Identify “DEVELOPMENT OPTIONS”
Options are proposed as a means of addressing the issues faced. In identifying

development options, planners should be cognisant of design theory and consider
alternative designs. Planners should try to introduce new technologies in to the
assessment to help keep abreast of the latest developments and exploit them when
they offer improvements over conventional approaches. In this way, this stage can
enable the consideration of multiple and diverse solutions. This is desirable from the
perspective of ensuring the best solutions are found but also supports decision
making and the recording of rationale by providing alternatives against which
options can be assessed and measured.
The separate identification of development issues and options contrasts with the
solution-oriented conventional approach to the design of network additions and
alterations. The adoption of this general structure for MCDM delivers immediate
benefits through this explicit consideration of issues and options and also supports
the further enhancement of the conventional approach with other methods like the
capture of decision rationale.
6.1.1.3. “QUANTIFICATION” of the Issues and Options
The quantification step involves the acquisition of relevant information, data and
models and selection of appropriate analytical tools to perform the necessary
analysis. The collection of information supports all subsequent stages and is thus
very important. Increasing complexity in network planning and design means a
significant increase in the burden of information processing. This is further
increased by the need to properly manage uncertainty in the information collected.
Various tools and methods can be applied to support planners. These include
methods for the collection, representation and exchange of information either
manually or through computer-based systems, as discussed in Chapter 7. The level
of automation and interactivity will depend on the particular application.
Furthermore, the properly managed exchange of information is crucial in meeting the
challenges of re-organisation and changing responsibilities that have resulted from
changes in the industry’s structure.
80
6.1.1.4. “ANALYSIS”
Analysis is performed with the results of the quantification stage, taking into account
the issues and options identified. This stage will employ the appropriate use of
computers, normally with considerable interaction from the user as they direct the
analysis and make an intellectual contribution based on their knowledge and
experience. Greater complexity in distribution network planning means more
analysis will be required. This might include power systems studies with models of
new technologies or financial studies that evaluate the cost benefits of DG.
Analytical tools to identify patterns and trends in the results might include statistical
analysis or comparison with standards or performance targets. With a greater need
for analysis, but increasing pressure on resources, it is crucial that productivity is
enhanced. This requires the provision of an effective toolkit of analytical
components that are available as modules that interface with one another without
imposing too great a burden of translation and conversion upon the planner. Also,
planners must be able to properly exploit the knowledge available within their
organisation. This requires appropriate knowledge management, which includes the
recording of the rationale behind the choices made by planners to provide a tractable
record of their work that provides some insight into the problems and solutions they
have considered. Drawing on the past experience of others in this way will enhance
the productivity of less-experienced planners.
6.1.1.5. “DECISION MAKING”
This stage takes relevant outputs from the analysis stage and processes the results to
help decision-makers reach a decision. It is important that through the identification
of issues and options, the quantification and analysis, planners remain focused on the
need to ultimately reach a decision. Decision making techniques should handle
multiple criteria, which will be drawn from the identification of development issues.
Tools should support the management of uncertainty and deal explicitly with risk. In
reaching decisions, it can be beneficial to draw on the collective knowledge and
experience of the organisation. Thus, knowledge management has a strong role to
play. The effective capture of rationale associated with decisions can support the
81
development of knowledge management systems. Design theory classifies decisions
and the risk or innovation associated with them into different levels. Consideration
of decisions from these perspectives might help planners reach appropriate decisions.
Or, if it is decided that additional quantification or analysis is required then further
iterations through the process can be undertaken.
6.1.2. Calculation of Alternative Decision Ratios
In terms of applications, MCDM techniques have been applied to the following

electricity industry decision problems [6.3 – 6.8]:
• Electricity generation expansion planning
• Electricity distribution system planning
• Electric utility resource allocation
• Assessing renewable generation technology options
The above applications of MCDM techniques to electricity industry decision

problems have improved the modelling and representation of the various factors
influencing the decision process. However, these applications of MCDM techniques
are not without criticism. Of particular concern is the inclusion of financial and
technical criteria together within a multiple criteria environment. Research suggests
[6.9] that while there is no doubt that such widely varying criteria are required to
accurately model and represent the decision making process, in many decision
problems the inclusion of financial criteria within a multiple criteria environment
may lead to substandard decision outcomes. This is because options that have a low
desirability and provide little benefit may be rewarded because of their low cost
values, while an option that provides significant benefits and is very desirable will be
penalised because of its high cost values. The inclusion of financial criteria in such
circumstances effectively masks the technical benefits that each option can provide,
with concomitant loss of value to the decision-maker because the most desirable
solution from both a technical and financial point of view cannot be readily
identified.
82
Extending this idea, it may be desirable to separate out criteria other than just
financial cost. In particular, environmental criteria could be assessed separately to
produce an environmental score or cost for each option. Ratios could then be
calculated between technical benefit, environmental cost and financial cost. These
different ratios would reveal the relationships between technical, environmental and
financial performance for each option and provide decision makers with additional
insight into the trade-offs to be made. This enhancement of the information and
understanding available to decision makers should lead to better decisions and
therefore helps to address one of the principal shortcomings and represents an
improvement to the conventional approach.
6.2. MCDM Case Study
This section describes the use of the MCDM decision support methodology
described above to an electricity supply planning task.
Given that financial criteria normally conflict with technical criteria (i.e. the need to
minimise overall costs while maximising technical desirability), a conventional
MCDM technique was modified to separate financial and technical desirability
values determined for each option. This ensures that the most economically efficient
option, providing the greatest technical benefit per unit cost, can be identified as well
as the option with the lowest overall cost. Even in situations where the difference in
cost between options is fairly small, the ability to determine the most financially
efficient investment option is a significant benefit that cannot be achieved by
simultaneously assessing all criteria (whether financial or technical), or by
assessment using only cost minimisation procedures. This is in line with the
conventional approach to distribution network planning, which has a number of
development drivers that may not always be consistent and complementary.
However, the use of decision support techniques like this goes some way to
addressing the identified shortcoming in formulating strategies and making
decisions.
83
The decision problem considered here concerns a large industrial company that is
upgrading its facilities and requires additional electrical capacity. The maximum
energy demand of the company will be 6MW while the existing utility connection
has a capacity of 4MW. Given the fairly remote location of the company’s
manufacturing site in relation to the nearest grid supply point (GSP), the company
has decided to consider investing in a DG scheme as a possible alternative to
upgrading their existing grid connection. Whichever investment option is selected
must be capable of delivering high quality power and have a high reliability.
Minimising environmental impacts is also an important objective.
6.2.1. Development Issues
The first stage is to identify all relevant development issues that will influence the
decision problem. From there, an appropriate set of evaluation criteria can be
produced. Given that this decision problem is concerned with the evaluation of
options to provide additional electrical capacity at a high power quality and
reliability, the following development issues were identified:
1. Financial cost – a key criterion in any investment decision
2. Reliability – compare the reliability of the various DG schemes with that
achieved by the distribution utility connection
3. Electrical performance – determine how the options affect the power quality or
dynamic performance of the manufacturing facilities
4. Environmental impact – consider the visual, noise or emissions effects likely to
be produced by each option
5. Space available at the industrial site
Using these development issues as a foundation, ten specific evaluation criteria were
identified.
1. Net present cost per kWh (c/kWh) – includes initial capital and installation,
annual operation and maintenance, energy costs, and pollution taxes (calculations
are based on a project economic lifetime of 15 years, although the effect of
varying this is tested with sensitivity analysis)
2. Reliability – number of interruptions per year
84
3. Reliability – annual interruption time (hours)
4. Power quality
5. Dynamic performance – impact on voltage and frequency
6. Standalone operation capability (if grid supply is lost)
7. Local environmental impact – emissions
8. Local environmental impact – noise
9. Local environmental impact – visual impact
10. On-site space requirement
6.2.2. Development Options
The next stage is to identify the set of development options to be considered. In this
case study, seven DG schemes will be considered alongside the upgrading of the
existing distribution network connection. It should be noted that while this case
study considers a relatively small number of options for the sake of simplicity, a
larger number of options could be assessed using the same approach. The eight
options considered are:
• Single Gas Turbine
• Multiple Gas Turbines
• Diesel Engine
• Fuel Cell
• Energy Storage
• Wind Power and Storage
• Photovoltaics and Storage
• Grid Reinforcement
The single gas turbine and diesel engine are conventional choices. They provide a
useful benchmark to gauge the performance of the fuel cell, energy storage, and two
renewable technologies. A multiple gas turbine combination is considered to assess
whether there is any benefit in adopting multiple, dispersed generator installations.
The two renewable schemes include an energy storage system to compensate for the
stochastic output of these technologies. The storage system included with the two
schemes is essentially the same as the energy storage option itself. The storage
85
system is not defined in any detail and assumes some technology able to deliver the
required amount of power and energy. With current technology, this is likely to
require standby generation coupled with some use of batteries in uninterruptible
power supplies.
6.2.3. Quantification
Once the development options and issues have been identified, it is necessary to
provide some quantification of their impact on one another. It is necessary to
identify the appropriate analytical tools, models and data to quantify and analyse the
issues and options. The primary sources of data for this case study were the
California Energy Commission [6.10] for cost data and DG literature [6.11] for cost
and reliability data.
Cost data was collected to enable the calculation of net present cost for each of the
investment options. To determine the net present cost in cents per kilowatt-hour
(c/kWh) for each option, four cost components were considered:
1. Initial construction costs
2. Annual operation and maintenance (O&M) costs
3. Annual energy costs
4. Annual pollution tax costs
Initial construction costs were specified as a cost per kW installed capacity for
generators and the grid reinforcement, and a cost per MW-hour for energy storage.
The required energy storage capacity was calculated to be 24 MWh. This assumes a
peak power capacity of 2MW – the new load on the industrial site – and a daily load
factor of 0.5. For most of the options it was assumed that they could supply the full
2MW at all times if necessary. For the wind and photovoltaic options, whose output
is not always at full capacity, capacity factors of 0.3 and 0.2 respectively were
assumed. This requires the installation of more capacity to ensure that sufficient
energy is produced to meet the expected demand. Thus, the installed capacity of
wind and photovoltaics was a function of peak load, load factor and capacity factor.
The initial construction cost data is summarised in Table 6.1.
86
Table 6.1 – Initial Construction Cost Data
Technology Cost / kW Power Cost / MWh Energy Total ($)

power capacity energy capacity
capacity (kW) capacity (MWh)
($/kW) ($/MWh)
Single GT 510 2000 N/a N/a 1020000
Multiple GT 510 2000 N/a N/a 1020000
Diesel Engine 400 2000 N/a N/a 800000
Fuel Cell 3500 2000 N/a N/a 7000000
Energy Storage N/a N/a 100000 24 2400000
Wind & Storage 800 3333 100000 24 5066667
PV & Storage 4500 5000 100000 24 24900000
Grid Reinforcement 750 2000 N/a N/a 1500000
Annual operation and maintenance costs were specified as a cost per kW installed
capacity. Energy costs in c/kWh for the fossil fuel options were dependent on a heat
rate in BTU/kWh and the cost of the fuel. For the renewable options, energy costs
were set at zero. The energy storage option would use grid electricity but its energy
cost increased because its efficiency was assumed to be 89%. Pollution tax costs
were set as a percentage of the energy costs, benefiting renewable technologies. The
ongoing cost data is summarised in Table 6.2.
Table 6.2 – Ongoing Cost Data
Technology Annual Heat Rate Fuel Cost Energy Pollution

O&M Costs (BTU/kWh) ($/MBTU) Costs Tax Costs
($/kW) (c/kWh) (c/kWh)
Single GT 35 12400 2 2.48 0.248
Multiple GT 35 13400 2 2.68 0.268
Diesel Engine 66 14000 4 5.6 0.56
Fuel Cell 66 9000 2 1.8 0.18
Energy Storage 175 N/a N/a 3.37 0.337
Wind & Storage 202 N/a N/a 0 0
PV & Storage 400 N/a N/a 0 0
Grid Reinforcement 20 N/a N/a 3 0.3
Reliability data was collected to estimate the number of interruptions each year and
the expected annual interruption time. These values were calculated for each of the
options by using values obtained from literature for interruptions per year and
average length of interruptions for different components. For example, the single gas
87
turbine option is calculated from the figures for natural gas supply and the single gas
turbine itself. The reliability data is summarised in Table 6.3.
Table 6.3 – Reliability Data
Reliability Components Interruptions per year Average length of

interruption (hours)
Electric grid supply 1.8 1.6
Natural gas supply 0.3 1.6
Diesel supply 0.01 12
Single Gas Turbine 1.7 3
Multiple Gas Turbines 1.4 1.7
Diesel Engine 1.7 3
Fuel Cell 1.7 3
Energy Storage 1.7 3
Wind Turbines 1.5 3
Photovoltaic Arrays 0.5 2
For the purpose of this case study, subjective assessments based on engineering
knowledge and judgement have been used to provide data for the power quality,
dynamic performance, standalone operation, environmental impact, and space
requirement criteria.
• Power quality for each option was assessed in terms of whether it was “Better”,
“Worse” or the “Same” as the existing grid connection.
• The dynamic performance of each option was assigned an ordinal ranking, with
“1” representing the best and higher numbers representing poorer performance.
Some of the options were ranked equally.
• Standalone operation capability of each option was specified as either “Yes” or
“No”.
• The local environmental impact on emissions for each option was quantified
using a symbolic scale from “Zero” (representing the most desirable), through
“Low” and “Medium” to “High”.
• The local noise and visual environmental impacts were assessed using an ordinal
ranking, with “1” representing the most desirable. Some of the options were
ranked equally.
88
• The space requirement criterion was evaluated using four symbolic values, from
“None” at the top of the scale, through “Negligible” and “Some” to
“Significant”.
In a real-life study of DG options, it is likely that a detailed analysis would be

performed for these criteria, rather than adopting subjective measures. For example,
simulation studies might be run to determine the power quality impact of each
option. This would require models of the different technologies and simulations to
determine their performance. Results from these simulations could then be used
within the decision framework.
The use of subjective data and simplification of the quantification and analysis is
considered acceptable in this case study because the aim is to demonstrate the
MCDM method. This reduced the analysis burden for this case study but also
demonstrates the flexibility of the MCDM framework in accepting different types of
criteria.
6.2.4. Analysis
Having collated the data, analysis was performed to produce a net present cost in
c/kWh and the required reliability figures for each of the options. The calculation of
net present cost used a discount rate of 10% and assessed costs over 15 years. For
the purposes of this study, it was assumed that all the options had the same lifetime.
In reality, different options are likely to have different lifetimes requiring a slightly
more complicated analysis. An assessment period of 15 years is typical for
investments of this type. To determine a net present cost in c/kWh, the net present
cost in purely monetary terms must be divided by an energy value. This energy
value is a discounted total equivalent value, which takes account of the discount rate
over the assessment lifetime. This is necessary to accurately reflect the present costs
of future energy use. An alternative approach would have been to calculate the costs
in terms of an equivalent annuity and divide that value by the annual energy
production.
89
As mentioned above, the study assumes a new load of 2MW peak with a daily load
factor of 0.5, resulting in an annual energy requirement of 8760MWh. The other
non-financial criteria were assigned their values through subjective assessment. The
analysis performed for each of the eight options is illustrated in Table 6.4, which
shows the analysis for the Single Gas Turbine. A summary of the results for all eight
options is shown in Table 6.5. The analysis was performed using a standard
spreadsheet tool.
Table 6.4 – Analysis for the Single Gas Turbine Option
Initial Construction Costs

Cost per kW installed capacity ($/kW) 510
Installed Capacity (MW) 2
$1,020,000
Annual Operation & Maintenance Costs
Cost per MW installed capacity ($/kW) 35.04
Installed Capacity (MW) 2
$70,080
Annual Energy Costs
Heat rate (BTU/kWh) 12.4
Natural Gas cost ($/MBTU) 2
Energy cost (c/kWh) 2.48
Annual MWh 8760
$217,248
Annual Pollution Tax Costs
Tax (c/kWh) 0.248
Annual MWh 8760
$21,725
Net Present Cost ($) $3,370,680

Discounted total equivalent MWh 66629
Net Present Cost per kWh (c/kWh) 5.06
Non-Financial Criteria
Interruptions per year 2
Annual Interruption Time (hrs) 5.58
Power Quality Better
Dynamic Performance 3
Standalone Operation Capability Yes
Local Environmental Impact - Emissions Medium
Local Environmental Impact - Noise 6
Local Environmental Impact - Visual 2
New space required on site Significant
90
Table 6.5 – Summary of Performance Results for all Criteria and all Options
Option Local
Environmental
New space required on site

Annual Interruption Time
Net Present Cost (c/kWh)
Impact
Dynamic Performance
Interruptions per year
Standalone Capability
Power Quality
Emissions
(hours)
Visual
Noise
Single GT 5.06 2 5.58 Better 3 Yes Medium 6 2 Significant
Multiple GT 5.28 1.7 2.86 Better 3 Yes Medium 4 2 Some
Diesel Engine 8.86 1.71 5.22 Better 3 Yes High 5 2 Significant
Fuel Cell 13.99 2 5.58 Worse 4 Yes Low 3 2 Significant
Energy Storage 11.31 3.5 7.98 Worse 2 No Zero 1 2 Significant
Wind Power and Storage 15.29 3.2 9.6 Worse 2 Yes Zero 5 4 Significant
Photovoltaics and Storage 60.21 2.2 6.1 Worse 2 Yes Zero 1 3 Negligible
Grid Reinforcement 6.01 1.8 2.88 Same 1 No Zero 2 1 None
6.2.5. Decision Making
Once the results describing the performance of each option for each of the evaluation
criteria have been compiled, a MCDM technique can be applied to assess the
desirability of each investment option. For this case study, the MCDM technique
that will be applied is a modified version of the Simple Multi-Attribute Rating
Technique (SMART) [6.1]. Other MCDM techniques are applicable and may be
used depending on the complexity and functionality required. With SMART, the
values for each criterion are converted to an equivalent, normalised value between
zero and one. The value one represents the most desirable value and zero represents
the least desirable value in each criterion. A linear relationship was adopted to scale
values between the most and least desirable values in each criterion. The results of
this normalisation are shown in Table 6.6.
91
Table 6.6 – Normalised Results for all Criteria and all Options
Option Local
Environmental
New space required on site

Annual Interruption Time
Net Present Cost (c/kWh)
Impact
Dynamic Performance
Interruptions per year
Standalone Capability
Power Quality
Emissions
(hours)
Visual
Noise
Single GT 1.000 0.833 0.596 0.900 0.333 1.000 0.300 0.000 0.667 0.000
Multiple GT 0.996 1.000 1.000 0.900 0.333 1.000 0.300 0.400 0.667 0.300
Diesel Engine 0.931 0.994 0.650 0.900 0.333 1.000 0.000 0.200 0.667 0.000
Fuel Cell 0.838 0.833 0.596 0.100 0.000 1.000 0.600 0.600 0.667 0.000
Energy Storage 0.887 0.000 0.240 0.100 0.667 0.000 1.000 1.000 0.667 0.000
Wind Power and Storage 0.815 0.167 0.000 0.100 0.667 1.000 1.000 0.200 0.000 0.000
Photovoltaics and Storage 0.000 0.722 0.519 0.100 0.667 1.000 1.000 1.000 0.333 0.600
Grid Reinforcement 0.983 0.944 0.997 0.500 1.000 0.000 1.000 0.800 1.000 1.000
6.2.5.1. Criteria Weightings
Almost all MCDM techniques, including SMART, require the weighting of each
evaluation criterion to indicate its importance in the decision analysis. The weight
values are then used with the normalised evaluation data to obtain an overall value
between zero and one indicating the desirability of each of the eight investment
options. The elicitation of criteria weight values is one of the most contentious
issues related to MCDM techniques as the chosen weight values have a significant
impact on the overall results obtained.
As discussed in Chapter 5, the structured decision making process can be used to

capture the rationale behind the various decisions and assumptions. This includes the
rationale behind the selection of criterion weight values. The capture of rationale
behind criteria weight values is an area of ongoing research but can be supported by
the use of specific methods and techniques in the elicitation of those values [6.12].
Further insight into the effect of different weightings can be gained through
sensitivity analysis. This reveals how the different criteria affect the decision
outcome. In this case study, sensitivity analysis was used to examine criteria
weightings, as discussed later.
92
In the modified MCDM approach used here, the net present cost criterion is excluded
from the evaluation of the overall score. Thus, it is not assigned a weighting. The
weight values selected for the nine normalised evaluation criteria are shown in Table
6.7. The sum of these weight values equals one. The weight values reflect the
importance attached to different criteria in the decision analysis. The weights for all
nine criteria are relatively close together, reflecting the view that all are considered
important. However, some distinction was drawn between the criteria and different
weightings assigned, and some rationale behind the values is provided in the
following paragraph.
Table 6.7 – Evaluation Criteria Weight Values
Criterion Weight
Interruptions per year 0.120
Annual Interruption Time 0.120
Power Quality 0.140
Dynamic Performance 0.085
Standalone Operation Capability 0.120
Local Environmental Emissions 0.110
Impact Noise 0.110
Visual 0.110
New space required on site 0.085
Power quality is of primary importance in this case study because the industrial load
being considered was deemed very sensitive to fluctuations in power quality. The
reliability and standalone capability criteria were assigned the next highest ratings
because of the direct effect of these criteria on output from the industrial process.
The local environmental impact criteria were considered the next most important as
they affect the conditions on site. Space requirements on site and dynamic
performance were considered to be of the lowest importance because of their
relatively lower impact on the industrial process.
6.2.5.2. Results of Benefit/Cost Evaluation
Using the performance results outlined in Table 6.5 and the weight values shown in
Table 6.7, an overall desirability score can be calculated for each investment option
93
using the SMART approach. The net present cost has not been included in Table 6.7
because in the modified version of SMART being used it is not combined with the
other criteria at this stage. Espie [6.9] advocated this approach in his use of MCDM
in distribution network planning. The separation of cost from other aspects of the
assessment reinforces the point that it is not necessary for all criteria to be converted
to an equivalent financial value. It also makes explicit the relationship between
alternatives in terms of their costs and benefits, which could be obscured by the use
of a single overall score for each alternative. The net present cost is combined with
the outcome of the MCDM analysis to calculate a benefit / cost ratio of each of the
eight options. This ratio is simply the total benefit score, from the weighted
combination of individual criteria scores, divided by the calculated cost, which in
this case is expressed as c/kWh. The overall desirability scores for each of the eight
options as well as the net present costs and the benefit / cost ratios are shown in
Table 6.8.
Table 6.8 – Results of Decision Analysis
Option Overall Score Net Present Cost Benefit / Cost

(c/kWh) Ratio (Score /
c/kWh)
Single GT 0.552 5.06 0.109
Multiple GT 0.690 5.28 0.131
Diesel Engine 0.567 8.86 0.064
Fuel Cell 0.511 13.99 0.037
Energy Storage 0.393 11.31 0.035
Wind Power and Storage 0.343 15.29 0.022
Photovoltaics and Storage 0.647 60.21 0.011
Grid Reinforcement 0.781 6.01 0.130
From the results shown in Table 6.8, it is evident that with the chosen criteria weight
values, the Grid Reinforcement option obtains the highest desirability score and the
Wind Power and Storage option obtains the lowest. The option with the lowest net
present cost is the Single Gas Turbine option while the Photovoltaic and Storage
option yields the highest net present cost. In terms of the benefit obtained for a given
investment, the option with the best benefit / cost ratio is the Multiple Gas Turbine
option while the Photovoltaic and Storage option yields the poorest.
94
The results demonstrate how neither the option with the highest score or the option
with the lowest cost is actually the best in terms of the benefit / cost ratio. This is the
reason why the MCDM approach was modified to use a separate cost and desirability
score for each option. This supports the decision maker in the assessment of the
various options as three investment choices are now available:
• the option with the lowest cost value
• the option with the highest desirability score
• the option with the best benefit / cost ratio
By considering the three investment choices a decision maker has more useful
information than is normally available from the outcome of a MCDM technique
where all criteria, including costs, are directly considered within the decision making
process.
Note that this holistic evaluation of all the options, including grid reinforcement and
distributed resources, assumes a single decision maker seeking the best overall
solution. In reality, regulatory frameworks in the electricity industry often break up
the supply chain to foster competition where possible and regulate the monopolies
that remain. The separation of networks from generation can present problems in
performing integrated assessment of alternatives. This was highlighted in the chapter
on new challenges.
6.2.5.3. Evaluation of Alternative Ratios
An MCDM framework provides flexibility for decision-makers to view the problem

and possible solutions from different perspectives. The calculation of three different
scores or performance measures, as described above, is one example of that
flexibility. An MCDM analysis like that described here might also be used to
examine more closely the environmental costs and benefits of different options. The
environmental impact of planning decisions has risen in profile in recent years such
that in some cases it becomes the primary consideration.
95
Table 6.9 shows overall scores for the eight options excluding both financial costs
and environmental costs and assuming equal weightings for the remaining criteria.
The table also shows a value of overall “environmental cost” for each option. This is
calculated by dividing one by the average of the normalised scores for the three
Local Environmental Impact criteria. The values show that the Grid Reinforcement
option has the lowest environmental cost, as assessed here, and the Diesel Engine
option has the highest environmental cost. The final column in the table shows the
ratio of the score to the environmental cost, i.e the new total benefit score (excluding
environmental criteria) divided by the new environmental “cost”, which is specified
as an inverse score rather than in financial terms. This analysis shows that the Grid
Reinforcement option offers the highest ratio of benefit to local environmental
impact.
Table 6.9 – Results of Environmental Cost Analysis
Score excluding
Overall Benefit /
financial cost and
Option Environmental Environmental
environmental
Cost Cost
impact
Single GT 0.611 3.103 0.197
Multiple GT 0.756 2.195 0.344
Diesel Engine 0.646 3.462 0.187
Fuel Cell 0.422 1.607 0.262
Energy Storage 0.168 1.125 0.149
Wind Power and Storage 0.322 2.500 0.129
Photovoltaics and Storage 0.601 1.286 0.468
Grid Reinforcement 0.740 1.071 0.691
The use of an MCDM technique such as that demonstrated here provides flexibility
to calculate alternative scores and ratios, which can provide an alternative
perspective on decisions. This is useful to decision-makers, and those they report to,
as they seek to explore problems and provide justification for the decisions that they
reach.
96
6.2.6. Sensitivity Analysis
The results obtained from the MCDM analysis are highly dependent on the input data
describing the performance of each investment option in each criterion as well as the
chosen criteria weight values. The sensitivity of the calculated benefit / cost ratio of
the eight options to variations in these data values was assessed. This contributes to
an understanding of how the input data affects the results and, among other benefits,
can help decision-makers arrive at suitable criteria weight values.
The figures below highlight how the benefit / cost ratio of each option varies
depending on changes to some of the input data. In each figure, the variable on the
X-axis was changed in small steps and the MCDM results re-calculated for each step.
97
Figure 6.2 shows the effect of varying equipment costs. This uses an “Equipment
Cost Multiplier”, which is used to scale the cost of initial construction by a factor
from 0.1 to 2. Thus, when the multiplier is equal to 1, the values on the plot match
the results as calculated above.
Figure 6.2 – Impact of Varying Equipment Cost on Benefit / Cost Ratio
98
Figure 6.3 shows the effect of varying the assessment lifetime used in the net present
cost calculations. The value used in the calculations above was 15 years. The results
were re-calculated for values between 1 and 30 years. This has the effect of
changing the relative impact of initial and ongoing costs although it can be seen that
the position of the different options relative to each other does not change very much.
Figure 6.3 – Impact of Assessment Lifetime on Benefit / Cost Ratio
99
Figure 6.4 shows the effect of scaling the cost of gas. This used a multiplier with
values between 0.1 and 2. With a multiplier value of 1 the results are as calculated in
the original results described above. As the multiplier changes, only the cost of gas
is changed in the calculations. Thus, only those options where the cost of gas has an
influence – particularly the GT options – show changes in their overall benefit score /
cost ratio.
Figure 6.4 – Impact of Varying Gas Cost on Benefit / Cost Ratio
100
Figure 6.5 shows the effect of scaling the cost of diesel. As with the gas cost
multiplier, the results for a multiplier value of 1 match those in the original
calculation while changes in the cost of diesel does not affect most of the options. It
is clear that as diesel becomes more expensive, the total cost of that option rises and
the benefit / cost ratio falls. Conversely, if diesel were very cheap then this option
would be very attractive.
Figure 6.5 – Impact of Varying Diesel Cost on Benefit / Cost Ratio
101
It is clear from the plots that different options are influenced differently by changes
to input parameters. Of greatest interest are the instances where the lines for
different options cross, particularly if they are the options with the highest benefit /
cost ratios. For example, Figure 6.4 shows how the gas turbine options drop below
grid reinforcement as the gas cost increases.
The criteria weight values are important factors in determining the outcomes
obtained from the decision making process. To help decision-makers arrive at
suitable weightings and to ensure that the outcomes are robust and justifiable, the
sensitivity of the benefit / cost ratio of each option across the nine criteria was
evaluated. This involved re-calculating all the results for changes in the criteria
weight values. The weight values used in the original calculation were put aside.
For the sensitivity analysis a more mechanistic approach was used. For each
criterion, its weight was varied from 0.1 to 1. The weights of the other eight criteria
were set equal to one another such that the sum of all nine criteria weights was one.
For example, if a weight value of 0.4 is assigned to the criterion in question, then
weight values of (1-0.4)/8 = 0.075 were assigned to all other criteria.
The sensitivities of the benefit / cost ratio to changes in the criteria weights for four
of the criteria are shown in the figures below.
102
Figure 6.6 shows how the results change if the weight assigned to the interruptions
per year criterion is varied between 0.1 and 1. As described above, the weights for
the other criteria also change so that the total of all weights is 1. The cost of each
option remains the same. As the weight assigned to interruptions per year increases,
the options that score highly in this criterion achieve higher overall benefit scores
and higher benefit / cost ratios.
Figure 6.6 – Impact of Varying Interruptions / Year Criterion Weight on Benefit / Cost Ratio
Figure 6.7 shows the effect of changing the weight assigned to the power quality
criterion. As the weight assigned to this criterion is increased, the weights assigned
to all other criteria are reduced and the overall benefit score for each option changes.
The cost of each option remains the same. With a weight value for this criterion of
103
0.1, the weights assigned to all nine criteria are approximately equal and the results
are very close to those of the original decision analysis in Table 6.8, with the grid
reinforcement option coming out top. In the assessment of performance, summarised
in Table 6.5 then quantified in Table 6.6, the grid reinforcement option was scored
less highly for this criterion than the gas turbine and diesel engine options. Thus, as
the weight assigned to this criterion increases, the overall score of these other options
increases. As the weights assigned to the other criteria must fall as a consequence,
the overall score for the grid reinforcement option drops.
Figure 6.7 – Impact of Varying Power Quality Criterion Weight on Benefit / Cost Ratio
0.30
0.25
0.20 Single GT
Multiple GT
Diesel Engine
Fuel Cell
0.15 Energy Storage
Wind Power and Storage
Photovoltaics and Storage
0.10 Grid Reinforcement
0.05
0.00
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Weight Value for Power Quality
Criterion
104
Figure 6.8 shows the effect of increasing the weight assigned to the standalone
capability criterion. As in Figure 6.7, with the weights all approximately equal (on
the left hand side of the plot), the grid reinforcement option has the highest benefit /
cost ratio. As the weight for this criterion increases and the weights for all the other
criteria fall, the total benefit of the grid reinforcement option decreases because it
scored poorly in this criterion.
Figure 6.8 – Impact of Varying Standalone Capability Criterion Weight on Benefit / Cost Ratio
105
Figure 6.9 shows the impact of increasing the weight assigned to the visual
environmental impact criterion. The changes in the relative positions of the different
options are not as great as when other criteria weights are increased because for this
criterion the scores assigned to each option were approximately reflective of the
overall total scores calculated with all criteria. The financial cost of each option
remains the same and so when the benefit / cost ratio is calculated the outcomes are
roughly the same whether all criteria are weighted evenly (on the left of the plot) or
only the visual environmental impact criterion is taken into account (on the right of
the plot).
Figure 6.9 – Impact of Varying Visual Environmental Impact Criterion Weight on Benefit /
Cost Ratio
106
It is evident from examination of the figures above that the benefit / cost ratio of an
option can change significantly if the weight value for a criterion is increased. For
example, Figure 6.8 shows how increases in the Standalone Capability criterion
weight causes the benefit / cost ratio for the Grid Reinforcement Option to fall
dramatically, cutting across the lines for other options. This indicates that this
criterion could have a significant influence on the decision outcome. In contrast,
Figure 6.9 shows that the order of options from best to worst changes little as the
Visual Environmental Impact criterion weight changes. This suggests that the
weighting for the Visual Environmental Impact criterion could be reduced without
having a great effect on the decision outcome. This kind of analysis can help
decision-makers arrive at appropriate criteria weight values and provide some
rationale for why some weights are increased and others reduced.
Difficulties in formulating strategies and making decisions was identified as one of

the main shortcomings in electricity distribution network planning. Ultimately,
distribution network planning is all about making decisions. Decision makers rely on
the results of various forms of analysis, such as power system simulation and cost-
benefit studies, but tools and approaches like MCDM can support the decision
making process itself and help address this shortcoming.
Formal MCDM methods support objective analysis and the reaching of justifiable
decisions by making explicit the identification, quantification and analysis of
decision criteria. Thus, the use of MCDM also contributes to the capture and
management of knowledge, another shortcoming identified in the conventional
approach and the primary lesson drawn from the assessment of engineering design
theory. It also assists with the incorporation of new technologies by requiring the
solution-neutral identification of all issues and options.
There are a wide variety of MCDM methods available and planners must chose those
that are most appropriate for particular tasks. However, a general structure for
MCDM-based planning was presented and a case study used to demonstrate its
107
application. A suggested novel variation on established methods is to calculate ratios
of benefit to environmental or other costs, where the benefit or “cost” need not be
expressed in purely financial terms but as a score or some measure that still allows
objective comparison of different options. Such an approach helps decision makers
to assess the options from different perspectives. This type of calculation might also
be of interest to regulators, where they want to make sure that distribution companies
are properly considering the environmental impact of their planning decisions.
The case study demonstrated how MCDM might be used in a power system planning
problem. The analysis of alternative options for expanding the supply to a remote
industrial facility found that grid reinforcement had the highest overall desirability
score. This suggests that despite the excitement over DG and its growth,
conventional grid reinforcement is still an attractive option. Grid reinforcement also
scored highest in the assessment of benefit against environmental costs. However,
the option with the lowest net present cost was found to be the single gas turbine
option and the option with the highest benefit / cost ratio was the multiple gas turbine
option. This shows that DG options are valid alternatives that may be selected
depending on the specification of decision criteria. The fair evaluation of DG is
subject to the regulatory framework, which may forbid network operators from
owning DG and require complex commercial agreements to be reached between
parties, thereby obscuring the opportunities to find and implement the least cost or
best value solution. Such complex commercial agreements are another new
challenge for network operators but will facilitate the exploitation of new resources
like DG and should therefore be fully explored.
6.1. Mollaghasemi, M., and Pet-Edwards, J.: “Making Multiple Objective

Decisions (IEEE Computer Society Technical Briefing)”, 1st Ed., IEEE
Computer Society Press, Los Alamitos, 1997.
6.2. Espie, P., Ault, G.W., and McDonald, J.R.: “Multiple criteria decision making
in distribution utility investment planning”, Proceedings of the International
108
Conference on Electric Utility Deregulation and Restructuring and Power
Technologies 2000, DRPT2000, London, UK, April 4 – 7, 2000, pp. 576 – 581.
6.3. Koritarov, V.S., Conzelmann, G., Veselka, T.D., Buehring, W.A., Cirillo, R.R.,
and Podinovski, V.V.: “Incorporating environmental concerns into electric
system expansion planning using a multi-criteria decision support system”, Int.
J. of Global Energy Issues, Vol. 12, No. 1 – 6, pp. 60 – 67, 1999.
6.4. Xiangjun, S., Xuefeng, M., Qiang, F., and Jianwen, L.: “MOPCA: a multi-
criteria decision model for power system expansion planning – methodology
and case study”, Int. J. of Global Energy Issues, Vol. 12, No. 1 – 6, pp. 68 – 80,
1999.
6.5. Pan, J., and Rahman, S.: “Multiattribute utility analysis with imprecise
information: an enhanced decision support technique for the evaluation of
electric generation expansion strategies”, Electric Power Systems Research,
Vol. 46, No. 2, pp. 101 – 109, 1998.
6.6. El-Hami, M.: “Application of Decision Support Systems to Power Distribution
System Planning”, IEEE PowerTech International: Symposium on Electrical
Engineering, pp. 756 – 760, 1995.
6.7. Wenstop, F.E., and Carlsen, A.J.: “Ranking Hydroelectric Power Projects with
Multicriteria Decision Analysis”, INTERFACES, Vol. 18, No. 4, pp. 36 – 48,
1988.
6.8. Georgopoulou, E., Lasa, D., and Papagiannakis, L.: “A Multicriteria Decision
Aid approach for energy planning problems: The case of renewable energy
option”, European J. of Operational Research, Vol. 103, No. 1, pp. 38 – 54,
1997.
6.9. Espie,P., Ault,G.W., Burt,G.M., McDonald,J.R.; “Multiple criteria decision
making techniques applied to electricity distribution system planning”; IEE
Proceedings Generation Transmission and Distribution, vol.150, no.5,
September 2003, p.527-535
6.10. California Energy Commission, “Distributed Energy Resource Guide:
Economics of Owning and Operating DER technologies”,
www.energy.ca.gov/distgen/index.html, 2002.
109
6.11. Willis, H.L., and Scott., W.G.: “Distributed Power Generation: Planning and
Evaluation”, Marcel Dekker Inc., New York, 2000.
6.12. Watson, S.R., and Buede, D.M.: “Decision synthesis”, Cambridge University
Press, UK, 1987.
110
7. Information Management in Distribution Network Planning
Planning and design, whether for electricity distribution networks or for other
systems, involve the management and processing of information. The manual
conversion of one form of information to another is where the principal intellectual
contributions are made and value is added. This is clear in the model of the
conventional approach to distribution network planning, where all tasks are
concerned with the processing of information.
However, the management of information can impose a huge burden on DNO

analysts and decision makers. Apart from the difficulties inherent in processing
information to create knowledge and make decisions about network development,
this burden includes less intellectually intensive tasks like the collection,
representation and exchange of information. The methods proposed in this Chapter
aim to reduce the burden of information processing and thereby enhance the
productivity of analysts and decision makers, which is necessary if distribution
companies are going to handle the new challenges that they face. Support can be
provided in a number of ways by introducing new approaches or methods and
utilising appropriate technologies.
The collection of information – to identify and quantify development issues and

options – can consume considerable resources. As distribution network planning
becomes more complex, this burden will grow. In particular, the expansion of DG
will require the collection and processing of significantly more information. In the
first part of this Chapter, a novel combination of three methods is presented that
could assist planners in the collection of information. Adoption of these methods
would improve productivity and enhance the ability of DNOs to deal with the
increasing burden of information management.
To keep up with the demands of managing distribution networks that are growing in
complexity, distribution companies will have to better exploit modelling and
simulation, as discussed in previous chapters and demonstrated in Chapter 9. This
111
includes the modelling and analysis of new technologies and also the greater
exploitation of resources through improved representation and exchange of power
system models and data. This can be addressed through effective use of information
and communication technology and is discussed in the second part of this Chapter.
In the final part of this Chapter, by way of demonstrating the ideas presented and
further examining the principal challenge in network planning, there is discussion of
information management for DG. A comprehensive set of information requirements
for DG has been defined to produce a structured approach for collecting information.
The methods proposed in this Chapter are illustrated with reference to DG, showing
how their use by DNOs would improve their capabilities and make possible new
approaches to network planning and operation.
7.1. Novel Combination of Methods to Assist in Collecting Information
In the past, the collection of information to support network planning and design has
been ad-hoc and specific to the task in hand. This was acceptable because expert
engineers were involved directly with the processing of information and because the
number of new connections to an existing network was manageable in this way. It
has been argued in previous chapters that the expansion of DG and other challenges
will make the collection of information more burdensome for distribution companies.
To help provide assistance in this area, a novel combination of methods is proposed.
Each of the individual methods is well known and has been applied in different ways
within electricity network planning but the combined approach provides a new
perspective. Each of the three methods (a structured approach, standard formats,
generic device types), and their role in the combined approach, is explained further in
the sections below.
DNOs have not adopted all three methods in combination before because the need
was not pressing and the incentive not clear. As DG has emerged as a bigger and
bigger challenge, DNOs have sought new ways of dealing with it through
information collection [7.1,7.8,7.9]. However, these have been hindered by inertia in
planning methods, rapid advances in DG technologies and difficulties in forming
112
consistent approaches across DNOs and across technology types. The time has come
for DNOs to implement new approaches in information management, particularly
with respect to DG. Doing so would speed up the planning process, benefiting both
DNOs and those that seek to connect to their networks.
7.1.1. Structured Approach
A structured approach with standard categories of information reduces the costs of

collecting information by providing consistency and formalism. Outlining the
categories and sub-categories of information allows information collectors to clearly
communicate their requirements to information providers. This is recognised in
many countries as an element of industry rules and regulations. For example, the UK
Distribution Code [7.1] defines the information that must be provided for new
connections to the DNO.
A structured approach for collecting information is necessary for and supports the
application of MCDM and the recording of rationale. These concepts are presented
elsewhere in this document. It is clear that a structured approach to information
management is central to distribution network planning, underpinning many other
methods and techniques.
The final part of this Chapter outlines a structured approach for the collection of
information on DG. This comprehensive set of information requirements provides
further insight into the potential impact of DG on distribution networks and
illustrates how a structured approach can assist in collecting information – see the
structure of information requirements for DG in Table 7.1 in section 7.3.2.
7.1.2. Standard Formats
A structured set of information requirements specifies the types of information

required but does not specify the exact format of that information. This type of
approach, where content is specified but format is not, is common in the electricity
supply sector. For example, the Distribution Code [7.1] takes this approach.
113
However, when information is expressed in non-standard formats, there is a need for
translation into the formats required for particular applications.
Without standard formats, the burden of information translation lies with the
information collector, in this case the DNO. If the provision of information had to
follow a pre-determined, standard format then the burden of translation would lie
with the information providers. This information could then be transferred directly
into the analysis and management systems used by DNOs. Standards have already
been established for the representation and exchange of data associated with energy
management systems. The Common Information Model (CIM) defines the format of
data and uses the eXtensible Markup Language (XML) as a vehicle for data
exchange [7.2]. These principles are being extended to data associated with network
planning and operation [7.3].
It seems inevitable that software vendors – if not the utilities and manufacturers –
will gradually converge on standard formats. An important aspect of most modern
data formats is their extensibility and interoperability so the traditional problems
associated with lock-in to standard formats is mostly overcome.
7.1.3. Generic Data
Where information is being collected on large numbers of small installations – e.g.

future domestic CHP systems – the concept of generic device types or generic
category data may be useful.
For DG or other network development drivers, generic types could be established,

with particular properties and fully specified information in all the required
categories of information. The description of the generic type might cover a range of
acceptable values, which are certified by DNOs as being acceptable. If specific
devices or projects fell within range for all information categories then they could
immediately be labelled as being a certified generic type. In many cases, that would
provide the DNO with all the information they need to conduct analysis, specify
connection conditions and manage their network. Only if there were something
114
particularly unusual about a specific device or network development would
additional information need to be collected.
Applying the same principal in a more flexible way, generic data could be specified
separately in each of the information requirements categories or sub-categories. In
providing information for a specific device, generic data could be selected in some
categories and specific data supplied for other categories. For example, for a new
DG installation there may be generic power and energy capabilities or generic types
of electrical connection. It may be possible to fully specify a project in terms of
generic data in each of the categories. Only if the project were particularly unusual
would non-generic information need to be collected.
The use of generic data and a structured approach to information requirements is

demonstrated in Table 7.1 in section 7.3.2. The fourth column of the table gives
examples of the data categories that might be used to describe a DG installation in
terms that are specific enough to be useful to planners but general enough to not be
too intrusive to the DG owners.
7.2. Representation and Exchange of Power System Models and Data
As discussed in previous chapters, the new challenges faced by DNOs will

necessitate an expansion of modelling, simulation and analysis of networks and their
components. To successfully meet this increased burden, the capabilities of DNOs
must be improved and models and data must be exploited more effectively. This will
require more frequent transfers of models and data between software tools and
between organisations. In addition, the making available of data publicly is being
encouraged, or required, by regulators and this will require the use of flexible and
translatable data formats. Thus, a significant aspect of power system simulation and
planning that must be addressed is the representation and exchange of models and
data.
115
7.2.1. Models and Data in Power Systems
Discussion of models and data can engender considerable confusion as the terms can
have multiple meanings. This is especially true in the field of power systems
engineering. Thus, to help establish meanings and to facilitate further discussion, it
is worthwhile considering the dictionary definitions.
• A model is: “a schematic description of a system, theory, or phenomenon that

accounts for its known or inferred properties and may be used for further study
of its characteristics” [7.4]; or “a description of observed behaviour, simplified
by ignoring certain details. Models allow complex systems to be understood and
their behaviour predicted within the scope of the model, but may give incorrect
descriptions and predictions for situations outside the realm of their intended use.
A model may be used as the basis for simulation.” [7.5]
• Data is: “1. Factual information, especially information organized for analysis or
used to reason or make decisions. 2. Numerical or other information represented
in a form suitable for processing by computer. 3. Values derived from scientific
experiments.” [7.4].
Within the power systems domain, a multitude of models are used, from models of
the electromechanical behaviour of generating sets to models of the degradation of
cable insulation. However, some of these models have become so widely used that
they are accepted as standard so that assumptions may be made or the model may be
implied by the data provided. For example, the pi-equivalent circuit of a
transmission line with the three-phase system represented by symmetrical
components translated to per-unit values, and with the negative and zero sequences
neglected, is a standard model used in power system simulation. When presented
with values of resistance, reactance and admittance, a power engineer may
confidently assume that the standard model of a transmission line is being used. The
same power engineer might consider the term “model” to refer to a model of a
complete distribution network (made up of multiple models of buses, branches,
generators, loads, etc.) or a model of a specific item of plant, such the generator,
116
exciter and governor models associated with a generating set. Thus, care must be
taken when discussing models and data in the power systems domain.
The nature and complexity of a power system model depends on the simulation and
analysis being performed. Models may have to be expanded to meet the demands of
a particular study; for example, assumptions implicit in the model may be invalid or
additional dynamic terms may have to be introduced. Alternatively, it may be
sensible to simplify a given model because the analysis being performed does not
warrant the additional complexity. The software environment in which models are to
be used will also influence the complexity required. This in turn determines the data
associated with a model.
Data by itself is meaningless, typically being a set of numbers or symbolic

parameters. As discussed above, when presented with data a certain model may
sometimes be assumed. But generally, data must be associated with a model of the
system or phenomenon being represented. In most cases, a model will have data
associated with it where the data influences the behaviour of the model and any
results produced from simulation.
7.2.2. Horizontal and Vertical Exchange of Power System Models and Data
The exchange of models and data can be considered as being “horizontal” or

“vertical”. The exchange of models and data between organisations is likely to be
horizontal in the sense that the models and data will be used for the same purpose but
perhaps in different software environments. For example, different organisations
may use different network analysis packages to perform load flow calculations. This
type of exchange may also be necessary within a single organisation. Models and
data may also be passed vertically between applications. This is most likely to be
within a single organisation. An example would be the exchange of topology data
between a load flow program and a reliability assessment tool.
Horizontal exchange of models and data will require some translation in the
representation or format but all the data should be available. In contrast, vertical
117
exchange of models and data will normally require some additions or reductions to
the actual content. For example, the exchange from steady state analysis to dynamic
simulation will require the addition of dynamics data. This variation in content and
representation is illustrated in Figure 7.1.
Figure 7.1 – Varying content and representation in power system models and data
Asset
management
database
Reliability
assessment tool
Varying Content
Load flow Load flow

program 1 program 2
Dynamic
simulation
program
Electromagnetic
transients
program
Varying Representation
7.2.3. Representation and Exchange of Power System Data
For data to be put to any use, methods of representation and exchange must be used.
Most simply, this might be a list of numbers written on a piece of paper. However,
interpretation of data normally requires it to be represented in some sort of structure
or format, often associated with a particular model of the system or component under
consideration. Computers provide a range of options for improving the
representation of data and its exchange between users and between formats.
Over the years, a number of formats for power systems data in simulation software
have been proposed and adopted to varying degrees. These include common formats
agreed on by industry groups or standards bodies, and vendor formats developed by
118
the makers of simulation software. These formats are based on models of power
systems that are bus/branch oriented, which is suitable for basic network planning
functions. Software vendors normally provide conversion tools to convert from the
most popular formats to their own proprietary formats, thereby facilitating data
exchange. The scope of these formats is necessarily limited and they are not
designed for extension.
A more recent, and ongoing, development is the Common Information Model (CIM),
which has been developed by the Electric Power Research Institute (EPRI) in
collaboration with others [7.2]. CIM has been developed to represent power system
modelling information for control centres and energy management systems, and
through data exchange facilitates the interoperation of electric utility software in
these application areas. CIM is node/breaker oriented, as is required for its target
applications, rather than bus/branch oriented, as in traditional planning and analysis
models.
XML is now being used as a means of holding and transferring data in the CIM
format [7.2]. XML provides a flexible and well-supported means of data exchange.
It offers the advantage of being able to hold meta-knowledge and semantics about the
data within a document. It can serve as a generic form of data representation, which
can then be translated to whatever format is required by particular software packages.
7.2.4. Representation and Exchange of Power System Models
The representation and exchange of models is much more complex than for data.
Models can be composed of various entities or modules, each described by their
behaviour and the relationships with one another and the outside world. The means
of representation and exchange will depend very much on the application.
For example, in presenting the common pi-equivalent transmission line model to a

human user, it would normally be represented by a textual description, diagrams and
various equations. Within power system simulation software, the same model is
represented in computer code that is difficult for most people to understand.
119
Alternatively, a system block diagram with transfer functions expressed in the S-
domain might be used to represent a generator excitation system. Humans might
readily understand this same representation, albeit with some background
knowledge, and it might serve as an input to software through a graphical model-
description interface.
In some cases it is desirable to use the same model in multiple simulation

environments. However, the generic representation of models is difficult and
previous efforts in model translation between formats have not been very successful.
This has prompted the development of a framework that retains models in their
native formats but allows the linking of multiple simulation tools, thereby gaining the
benefits of integration without having to transfer models between formats and tools
[7.6]. XML and intelligent agents are used to facilitate the inter-operation of the
multiple simulation environments. XML documents define the system structure in
terms of the models, the links between them, and the software in which each of them
run. The information held on each model includes commentary on modelling
assumptions. Information on initialisation and set-up of models is also required.
7.3. Information Management for Distributed Generation
Technical developments in distribution systems, particularly the connection of DG,

and changes in the commercial and regulatory environment pose a number of new
challenges for DNOs [7.7]. DNOs will need to perform new types of analysis to plan
and design their network, develop management and control strategies, and implement
appropriate commercial arrangements to ensure a reliable and high-quality electricity
supply is maintained. But these new studies will require appropriate information to
be collected. This might be information on the electrical network itself, new DG
installations or new technologies associated with network developments.
The task of collecting and processing information on DG could represent a

significant new burden for DNOs. There are now a wide variety of DG technologies
available, and at first sight, it seems that each requires slightly different information
for characterisation. For example, the UK Engineering Recommendation G83 for the
120
connection of small generators to low voltage networks has separate annexes for
each generation technology [7.8]. Furthermore, in low voltage networks, where DG
installations may be small but numerous, the cost of collecting information on every
installation may outweigh the benefits.
Earlier in this Chapter, it was proposed that the burden of collecting information
could be reduced through the use of a number of methods. These include a
structured approach, the use of standard formats, and the identification of generic
device types or categories. These methods will help DNOs to collect all the
information they need to evaluate DG in their networks. Most DNOs have not yet
started collecting in-depth information on DG like this because the numbers have
been small enough for conventional approaches to be maintained. As the new
challenges become greater, DNOs will benefit from work such as that described here,
which will assist them in managing the large amounts of information resulting from
DG and other changes to the network.
Section 7.3.1 discusses the need for information on DG. A structured approach for
collecting information on DG is outlined in section 7.3.2. Section 7.3.3 presents
some examples of the types of analysis relating to DG that DNOs must perform, and
how information on DG would be used to improve planning and operation.
7.3.1. The Need for Information on DG
There have always been some generators connected at distribution level but the
expansion of DG poses new challenges for DNOs. When a new generator applies for
connection, analysis must be conducted to determine the impact on network planning
and design, management and operation, and commercial arrangements. This is
existing practice, but the expansion of DG will mean DNOs having to deal with
many more connection requests and adapting their methods appropriately. This will
be true especially when connecting small DG units at low voltage becomes
commonplace. So while DNOs have not needed to collect so much information on
DG before, they can expect to have to collect much more in future.
121
7.3.1.1. Information for Planning and Design
Previously in most DNOs, DG would be considered in the planning and design

process only in response to a new DG connection application. With DG becoming
widespread, the ongoing planning and design of networks must consider it explicitly
in a more forward-looking manner. The treatment of DG within network
development must become proactive rather than reactive. Thus, information must be
collected on DG to gain an understanding of the different types being connected.
To ensure effective simulation and analysis of DG, both prior to connection and on
an ongoing basis, there is a need to characterise DG. Information must be collected
on all aspects relevant to DNO activities. The expansion of DG is due in part to the
emergence of new generation technologies. Devices like micro gas turbines and
interfaces utilising power electronics are not part of the standard libraries used by
DNOs for modelling their network. Thus, there is a need to characterise DG to
facilitate modelling and simulation.
7.3.1.2. Information for Management and Operation
In the past, the lower voltage levels of distribution networks have often been
installed on the basis of “fit-and-forget”. Network management has been passive,
with the network delivering power in one direction from substations fed by high-
voltage transmission systems to consumer loads. The introduction of distributed
energy resources, perhaps even at low voltage, will require more active management
of networks [7.9].
The degree of active management necessary must be determined. In some cases,

probably with very small generators, DG may be installed and left to operate
autonomously, being treated just like most loads have in the past. While in other
cases, probably with larger generators, extensive communication and control systems
may have to be implemented to provide integrated management of the network and
generator. Operating constraints for DG may have to be specified and then
implemented. There may be some trade-off to be evaluated between operating
constraints and reinforcing the network [7.10].
122
Ongoing management of the network and its assets requires knowledge of what is
connected and the influence it might have. DG could be particularly influential,
affecting maintenance programmes and asset replacement decisions. The
incorporation of DG within network management and operation will require analysis
methods to be revised; and this will require new information to be collected.
7.3.1.3. Information for Commercial Arrangements
DNOs must also consider commercial issues. A charging structure for DG must be
determined and applied to each new installation. There has been much discussion of
“deep” and “shallow” charging for the connection of DG to a network [7.11]. It may
also be necessary to establish pricing structures for the provision of network support
services from DG installations; e.g. voltage and frequency support. The charging
systems adopted by a DNO will have to be agreed with the appropriate regulator.
Within the bounds of any agreed system, there will probably be scope for DNOs to
negotiate with DG to tailor charging structures to suit particular circumstances.
These negotiations will have to be supported by analysis of a type that DNOs may
not have performed in the past. This analysis will require DNOs to gain an
understanding of the commercial issues associated with DG and be able to collect
and process appropriate information.
7.3.2. A Structured Approach for Collecting Information on DG
Although different DG technologies have different operating principles and different

associated data, it is possible to identify a basic set of information requirements that
facilitate a structured approach to information collection. A comprehensive set of
structured information requirements is proposed and discussed below. The
information requirements are summarised in the two left-most columns of Table 7.1.
Each component is included because it will support analysis that DNOs may have to
perform. Column 3 of Table 7.1 contains an example of specific device information
for a micro gas turbine.
123
Table 7.1 illustrates two of the three methods proposed at the start of this Chapter as
being of use in information management. The first two columns define the
structured approach, specifying all the information to be collected. The fourth
column gives examples of generic data, with examples of the categories that might
be used for each element of the information structure. The third method, standard
formats, requires more formality in the way information is expressed. This could be
achieved in a number of ways including on-line or electronic forms or prescribed
data formats.
Other parties have undertaken the task of outlining basic information requirements
for DG for different purposes. For example, in the UK, the Distribution Code [7.1]
defines the information that must be provided for new connections to the DNO
including new DG. In the USA, EPRI and the Electricity Innovation Institute (E2I)
are developing object models that define the information exchange necessary for
monitoring and controlling DG [7.12].
Table 7.1 – Information Requirements for Characterisation of Distributed Generation and

Examples of Specific Device Information and Generic Category Data
Category Sub-Category Example of Specific Device Example of Generic

Information Category Data
1. General Description 1.1. Technology Type Micro Gas Turbine. Generic Micro Gas Turbine
1.2. Network Location Connected to 11kV bus N/A
serving Customer X.
1.3. Ownership and Owned and operated by N/A
Responsibility Customer X facilities
management.
2. Power and Energy 2.1. Normal Operating 0 – 28 kW (maximum Maximum between 20 and
Capabilities Range reduces with higher ambient 40 kW
temperatures – refer to data
sheets)
2.2. Overload Capabilities None Assume none
2.3. Response Rates 10 kW/second up and down Between 5 and 20 kW/sec.
2.4. Start-up and Shutdown 1 minute for start-up. Less than 3 minutes for start-
Times 1 minute for shutdown. up or shutdown
2.5. Power and Energy Over Continuous operation Flexible depending on
Time possible subject to gas application
supply but operated in heat-
led load-following mode
(typical profiles available).
3. Variable Costs 3.1. Cost of Fuel Variable but currently Less than 5c/kWh for input
3c/kWh fuel
3.2. Fuel Efficiency or 25% ± 2 electrical efficiency Between 20% and 40%
Losses 70% ± 5 total efficiency electrical efficiency
3.3. Operational Unknown Less than 2c/kWh
Degradation Costs
3.4. Start-up and Shutdown Negligible Less than $5 to start-up or
Costs shutdown
124
3.5. Environmental Costs Emissions certificates costs Less than 3c/kWh
of 1.2c/kWh
4. Fixed Costs 4.1. Installation Costs $25,000 $20,000 - $40,000
4.2. Ongoing Fixed Costs $500 p.a. Less than $2000 p.a.
4.3. Technology Maturity Commercially available Commercially available
5. Electrical Connection 5.1. Generator or Interface Standard PWM switched, Standard inverter and
Parameters three-phase inverter and transformer
standard three-phase
400/llkV transformer
5.2. Transformer Parameters Rating = 30kVA Rating 20 to 50 kVA
Impedance = 10% Impedance less than 20%
5.3. Operating Limits 360 – 528 VAC, 10 – 60 Hz 390 – 410 VAC, 45 – 55 Hz
and THD<20% at inverter and THD<15%
terminals
5.4. Start-up and Shutdown Soft-start capability No significant effect on
Behaviour produces smooth current network
increase on starting and
decrease on stopping
5.5. Power Quality Effects Negligible THD<15%, PSD<1, PLT<1
5.6. Short Circuit Current Up to 65A Between 50A and 100A
6. Protection 6.1. Protection Types Built-in protection: According to standard
under/over voltage and engineering
frequency, loss of mains recommendation.
detection
6.2. Means of Isolation Inverter blocking plus Full isolation must be
isolator switch on grid side possible
of transformer
7. Control 7.1. Pre-programmed Fully programmable through Must have constant power
Control Strategies computer interface. Set to factor and voltage control
follow supermarket heat options
load.
7.2. Control Changes Implemented immediately Manually within 24 hours of
through computer control notice being received
interface.
7.3. Island Operation Not possible in current Not required
configuration.
8. Communications 8.1. Interface and Protocol USB connection to PC Direct link to PC
running proprietary software
8.2. Inputs Heat and power schedules, Heat, power and reactive
voltage control options power settings at least
8.3. Outputs All system parameters for Heat, power and reactive
monitoring and control power values at least
In some cases of information collection, it may be appropriate to just make reference

to existing sources of information, such as manufacturer data sheets. Some of the
information may be provided from monitoring or simulation so reference could be
made to the appropriate results. However, for some generators, the information
simply will not be available. In addition, generators may not wish to reveal
information relating to cost or to innovative designs or operating methods that they
regard as valuable commercial or intellectual property. The sharing of information
between generators and the DNO must be acceptable to all parties and governed by
appropriate confidentiality agreements.
125
7.3.2.1. General Description
A general description of the technology should be provided, covering technology

type, network location, and ownership and responsibility. This provides DNOs with
the basic information to identify a DG installation.
Technology Type
A brief description of the technology should be given in terms of its energy source
and means of energy conversion, e.g. “a natural gas fired micro turbine with high-
speed asynchronous generator and AC-DC-AC grid interface”.
Network Location
The location of the unit within the DNO network should be noted. This will identify
the unit uniquely and determine the electrical environment in which it operates.
Ownership and Responsibility
Information about the owners and those responsible for operating and maintaining it
should be noted alongside the more technical information.
7.3.2.2. Power and Energy Capabilities
The power and energy capabilities should describe the basic operational performance
of the unit under consideration. This information is required by DNOs to conduct
power flow and other studies of network performance.
The performance of a generation or storage system may be limited by the energy

source, the means of energy conversion, or the grid interface. For example, the ramp
rate of a CHP unit might be limited by its boiler dynamics but the limit on total
power output might depend on its generator or on the power electronics in the grid
interface. The exact source of any limits should be noted.
126
Normal Operating Range
The range of operation in normal, continuous operation should be described in terms

of limits on real and reactive power. This may be described by limits, equations or
with an operating chart, e.g. as shown in Figure 7.2.
Figure 7.2 – Simple example of a possible operating chart for a distributed generator
Power (kW)
10
Reactive power
(kVAr)
-5 0 5
Overload Capabilities
The ability to operate outside of the normal range for short periods, i.e. overload
capabilities, should be noted. This may be described by extensions to the limits, new
equations or an extended operating chart and associated time limits.
Response Rates
The rate of change possible in increasing and decreasing both real and reactive power
should be noted. This may be described as rates, W/s or VA/s, or in whatever form
is necessary to describe the capabilities. Response rates may also vary depending on
operating conditions, e.g. as shown in Figure 7.3.
127
Figure 7.3 – Simple example of a possible response rate chart for a distributed generator
dP/dt(kW/s)
1
0 2 6.5 10 12 P(kW)
-1
Start-up and Shutdown Times
There may be minimum, and perhaps maximum, limits on the time required for start-
up and shutdown of a unit. There may be links to the response rates category.
Information should be provided on the specific time limitations of start-up or
shutdown in whatever form is appropriate to the technology.
Power and Energy Resource Availability
The amount of power and energy available over time should be described. This will
depend very much on the technology so should be described in whatever form is
most suitable. However, the description should include information on the source of
energy and its availability over time, or the probability of its availability. In simplest
terms, this could be represented by a single capacity factor. Preferably, the despatch-
ability or stochastic nature of the unit should be noted. For example, power output
from a CHP unit might depend on heat loads. The power from PV panels will
depend on expected patterns of solar irradiation. Energy storage units will have
finite capacity, further complicated by constraints on charge and discharge.
7.3.2.3. Variable Costs
Variable costs are those costs that depend on how a unit is operated and the amount
of electrical energy delivered. In the fullest detail, they could be specified in terms
of fuel costs, fuel efficiency, equipment degradation, and start-up and shutdown
costs. It may also be useful to specify environmental costs separately so they can be
128
evaluated explicitly. Information on variable costs would be useful to DNOs in
optimising operational strategies.
However, owners and operators of DG may wish to keep their operating costs secret.
The information disclosed to the DNO will depend on the market and network
management structure. For example, the distributed generator may present the DNO
with a simple cost curve like that shown in Figure 7.4. A simple single-cost
approach is most straightforward but it may be necessary to specify separate “bid”
and “offer” prices within a market dispatch structure.
Figure 7.4 – Simple example of a possible daily cost curve for a distributed generator
Cost per energy unit

($/kWh)
0.100
0.075
0.050
00:00 06:00 10:00 16:00 20:00 00:00

Time of day (hh:mm)
Cost of Fuel
The cost of fuel may vary over time depending on the contract for supply. The cost
of fuel may internalise external costs such as pollution taxes and this should be
noted. The exact format of the description will depend on the technology. Some
technologies, such as photovoltaics, may not have a cost of fuel.
Fuel Efficiency or Losses
To determine operating cost, the cost of fuel must be combined with fuel efficiency
or losses. Fuel efficiency and losses may depend on the operating point and dynamic
operation of the unit. For example, efficiency may be different for operation at
different capacities, and efficiency may vary if the unit is ramping up or down at
different rates. This may be described by values, equations or additional information
129
on the operating chart. For some technologies, this characteristic will not be
applicable.
Operational Degradation Costs
The degradation of equipment, and thus the operating cost, may depend on its
operation. For example, equipment may degrade more quickly when operated close
to its capacity. Operating for short periods outside of the normal limits may be
possible but may degrade the equipment, imposing additional costs. These
operational degradation costs should be described in whatever form is most suited to
the technology under consideration, but preferably in a manner similar to the cost of
fuel.
Start-up and Shutdown Costs
Start-up and shutdown of a unit may incur specific costs. It may be possible to
describe these costs under fuel efficiency or operational degradation costs. Where
this is not possible, information should be provided on the specific costs incurred as a
result of start-up or shutdown in whatever form is appropriate to the technology.
Environmental Costs
It may be possible to internalise environmental costs within the cost of fuel.

However, it may be valuable to keep environmental costs separate. Sensitivity to
environmental effects may depend on conditions and the time of day. For example,
the “cost” of emissions from fossil-fuel-fired generation may be greater at times
when emissions from transport are already high.
7.3.2.4. Fixed Costs
Fixed costs are those costs that do not depend on how a unit is operated and are of
less relevance to the DNO. In fact, owners and operators of DG will probably wish
to keep this cost information secret. However, if available, this information is
valuable for analysis such as long-term planning.
130
Installation Costs
Ideally, installation costs should be described in terms of cost/kW rating and

cost/kWh capacity. In addition, the costs of additional components and systems
should be noted, e.g. communication and control systems. Different technologies
will be composed of different components so the description will depend on the
technology.
Ongoing Fixed Costs
Ongoing fixed costs not related to operation should be noted. For example, some
maintenance and replacement of parts may be necessary whether the unit produces
power or not. Some fixed costs, such as property taxes and staff costs, are
independent of the technology but vary with location.
Technology Maturity
Related to the cost of installation is the maturity of the technology. In simplest

terms, it should be noted whether the technology is widely available commercially or
whether it is a test or demonstration project. This information helps DNOs stay
abreast of technology trends.
7.3.2.5. Electrical Connection
Information should be provided on the electrical connection. This will strongly

influence the operational capabilities of the generator or storage device, and thus is
of great interest to DNOs. In particular, information on the electrical connection is
required for accurate modelling and simulation.
Generator or Interface Parameters
The type of electrical generator and/or grid interface should be noted. If this is of a
standard type that can be represented by a standard model then parameter values
should be provided. If it is not of a standard type then some description should be
provided to inform work on the creation of models. This should include details of
the voltage level and whether it is a single or three-phase connection.
131
Transformer Parameters
If a transformer is used to connect to the grid then details necessary for modelling
should be provided. This should include information on in-rush currents if these will
affect the performance of the unit.
Operating Limits
The operating limits for the electrical connection should be described. This should
include information on maximum and minimum grid voltages, maximum and
minimum grid frequency, and total harmonic distortion that the technology can
tolerate. If any of these conditions vary with operating point, e.g. different on start-
up or shutdown, then details should be provided.
Power Quality Effects
When running normally, the technology may have an effect on power quality on the
grid. For example, it may inject harmonic currents. Alternatively, it may be able to
mitigate against some undesirable power quality phenomenon. Any behaviour that
may affect grid power quality, for better or worse, should be noted.
Start-up and Shutdown Behaviour
The requirements and behaviour of the technology on start-up and shutdown should
be described. The primary concern is the effect on the grid and power quality, e.g.
waveform distortion. In particular, any behaviour that may cause further damage to
the grid when it is already in a vulnerable state should be noted.
Short Circuit Current
The ability of the device to supply short circuit fault current should be described.
This may be in the form of sub-transient and transient reactances, and time constants
for decay of any DC-offset components of fault current. Fault levels have been
identified as the primary network issue of concern for DG in urban areas in the UK
132
and Engineering Recommendation G74 [7.13] describes the procedure to be used to
calculate short-circuit currents and the data required from generators.
7.3.2.6. Protection
Information should be provided on the types of protection associated with the unit
and what makes them operate. Protection may be intrinsically linked to the
technology type or it may be standard protection that is used with all DG resources.
Often, the protection required will be specified by the DNO, subject to the
appropriate standards and regulations. In addition, the detailed design of protection
must be coordinated with the DNO network protection.
Protection Types
For each type of protection installed, some information should be provided on the
means of measurement, relay operation and the limits of operation.
Means of Isolation
The means of isolating the unit from the grid should be described. In particular, the
use of conventional circuit breakers or solid-state switches should be noted.
Reference should be made to the conditions that trigger isolation, the time it takes for
isolation to be achieved, and any grid-visible effects that can arise.
7.3.2.7. Control
Information should be provided on the control of the unit. This information will be
required if a DNO implements active management of DG resources.
Pre-programmed Control Strategies
An outline of the pre-programmed control strategies available should be provided.

These might include “maximise energy output”, “maintain bus voltage”, “fixed
power factor”, or some combination. If no pre-programmed control strategies are
available then this should be noted.
133
Control Changes
Any restrictions on the changing of control strategies should be noted. For example,
must the unit be shutdown before a new control strategy can be implemented? Can
new control strategies be implemented remotely or must changes be made at the
unit? Response times should also be noted. Some reference to Communications, see
below, may be necessary.
Island Operation
Where possible, the conditions necessary for the unit to continue operating as part of
a power island, independently from the grid supply, should be described. This may
refer to the operating limits described above, to some other aspects of the grid
conditions, or to some aspects of the unit itself. In the future, and subject to legal
and regulatory approval, DNOs may be interested in operating power islands because
of the potential improvements in overall supply reliability and availability.
7.3.2.8. Communications
Details of the communications interfaces should be provided. This may include

reference to the control information. If DNOs are going to exert any active control
over DG, then communication links will be required.
Interface and Protocol
Details of the communications interface and protocol should be provided, e.g. DNP3,
RS232, Ethernet, modem. This should include details of existing capabilities for
remote control and monitoring.
Inputs
Details should be provided of the input parameters to the generator or storage unit.
This will determine what information must be provided to the unit to control
operation.
134
Outputs
Details should be provided of the parameters that can be monitored or extracted from
the unit, including the sampling rates. This will determine whether additional
measurements must be taken to satisfy all monitoring requirements.
7.3.3. Examples of Distributed Generation Analysis
DNOs require information on DG to perform the analysis necessary to manage their

networks. The following examples illustrate how the information provided might be
used by DNOs. These examples show how the collection of comprehensive
information on DG in an effective way, using methods such as those outlined above,
will facilitate the analysis and network planning and operation that will be necessary
to accommodate DG on existing and future distribution networks. This includes
scenario analysis to evaluate different possible outcomes, as discussed in the next
Chapter. Without collecting suitable information – as is current practice in many
DNOs – the necessary analysis will not be possible and network planning and
operation will be held back.
7.3.3.1. Modelling of New Technologies
The expansion of DG is driven in part by the development of new generation

technologies. A DNO may have to develop new models to perform computer
simulation and determine the effect of new technologies on their network. For basic
power flow studies, new models may be developed using information provided on
power and energy capabilities for a DG installation.
For example, the expected production profile of a CHP unit might be used to study
the variation of network voltage profiles. If provided in the correct format, this
information might be inserted directly into generic models that the DNO has
available. However, if information is provided in another format then the DNO will
have to translate it before it can be used.
135
Models may also be required to simulate the DG response to network disturbances.
Information provided on the electrical connection and protection systems would help
the DNO establish suitable models.
7.3.3.2. Designing Active Management Systems
Over time, it is expected that DNOs will have to shift from passive to more active
management of their networks to accommodate DG. This will require an
understanding of the DG resources, which the information requirements described
above should provide. In particular, the design of active management systems will
require information on the communications and control systems associated with a
DG installation.
For example, a new generator may have a number of control set points that determine
its operation. These set points may be controllable remotely using a modem
connection. If the DNO is to exert control over this unit, they will have to establish a
modem connection and understand the effects of the different control set points.
Software might be developed to control the DG unit and integrate it with the DNO
network management systems. In addition, the nature of the underlying operating
costs of DG units might be required to assess the most economic way to manage the
DG resources.
7.3.3.3. Contracting for Voltage Support Services
Some DG has the potential to provide ancillary services to the network, such as
voltage support. However, the provision of such services may be subject to
commercial arrangements. A DNO might use the information provided on operating
capabilities and variable costs to design a suitable contract. Given the importance of
maintaining network voltages, the engineering solution using DG will have to be
closely examined. Equally, the contract agreed between the DNO and DG will have
to cover all eventualities and provide the correct incentives to maintain network
performance.
136
The growth of DG and other developments in distribution network planning impose

new challenges on DNOs, including new burdens in analysis and the collection of
information. Three methods were suggested to support information collection: a
structured approach to information requirements; the use of standard formats for
representing that information; and the use of generic device types or device
information categories. DNOs do not currently use these methods in combination
but doing so would improve their information management capabilities and put them
in a better position to tackle the new challenges they face, particularly the expansion
of DG.
The representation and exchange of power system models and data was discussed.
The potential confusion associated with the terms “model” and “data” was noted and
the concept of horizontal and vertical exchange was presented. Further discussion
highlighted how the representation and exchange of models and data are being
improved through the development of new formats and the exploitation of
information technology. DNOs will have to utilise new technologies in their
information processing and analysis as well as on the networks themselves.
A comprehensive set of information requirements was presented for characterising

DG. This covered a general description, power and energy capabilities, variable and
fixed costs, the electrical connection, protection, control and communications.
Collecting this information would allow DNOs to perform the analysis necessary to
manage the growth of DG in their low voltage networks, including scenario analysis
or other ways of dealing with risk and uncertainty as discussed in the next Chapter.
However, it is accepted that some information may not be available or generators
may not wish to provide it, particularly when it comes to costs or innovative designs.
It was noted that in the requirements described above, the exact format of the
information is left to the provider. This leaves the collector of the information with
the task of translating information to the format required. The approach would be
even more structured, and the burden of translation would be shifted from DNO to
137
DG, if all information had to conform to an agreed pre-defined format and structure,
as outlined in this Chapter. Reaching agreement on this issue may require the
intervention of government or industry regulators.
The concepts and approaches described in this section have been applied in the
development of management systems for DG in LV networks. The concept of pre-
defined data formats and structures was explored further in the development of
markup languages for power system planning and associated platforms for web
services, including on-line submission and automated checking of DG-related data
[7.3]. These applications demonstrate the importance and value of formal methods
for the characterisation of DG.
7.1. Distribution Code Review Panel, “The Distribution Code of Licensed

Distribution Network Operators of Great Britain”, Issue 2, March 2003
7.2. McMorran,A.W., Ault,G.W., Elders,I.M., Foote,C.E.T., Burt,G.M.,
McDonald,J.R., “Translating CIM XML Power System Data to a Proprietary
Format for System Simulation”, IEEE Transactions on Power Systems, vol.19,
issue 1, February 2004, p.229-235
7.3. McMorran,A.W., Ault,G.W., Foote,C.E.T., Burt,G.M., McDonald,J.R., “Web
Services Platform For Power System Development Planning”, Proceedings of
the 38th International Universities Power Engineering Conference,
Thessaloniki, Greece, September 2003
7.4. The American Heritage® Dictionary of the English Language, Fourth Edition,
Copyright © 2000 by Houghton Mifflin Company, through dictionary.com
7.5. The Free On-line Dictionary of Computing, through dict.org
7.6. McArthur,S.D.J., Davidson,E.M., Dudgeon,G.J.W., McDonald,J.R.; “Toward a
model integration methodology for advanced applications in power
engineering” IEEE Transactions on Power Systems, Volume 18, Issue 3,
August 2003, pages 1205-1206
7.7. Jenkins,N., Allan,R., Crossley,P., Kirschen,D., Strbac,G., Embedded
Generation, 2000, The Institution of Electrical Engineers, ISBN 0 85296 774 8
138
7.8. Energy Networks Association, “Engineering Recommendation G83: The
connection of small-scale embedded generators (up to 16 a per phase) in
parallel with public low voltage distribution networks”, UK, 2002.
7.9. First Annual Report of the Distributed Generation Coordinating Group
(2001/2002), March 2003, Department of trade and Industry, Ofgem
7.10. Foote,C.E.T., Ault,G.W., Burt,G.M., McDonald,J.R., “Enhancing flexibility
and transparency in the connection of dispersed generation”, Proceedings of
CIRED 2001, Amsterdam, June 2001
7.11. Watson,J., “The Regulation of UK Distribution Networks: Pathways to
Reform”, Proceedings of the Second International Symposium on Distributed
Generation, Stockholm, Sweden, 2-4 October 2002
7.12. Electricity Innovation Institute, “Open Communication Architecture for
Distributed Energy Resources in Advanced Distribution Automation
(DER/ADA)”, November 2003, Project description available from
http://www.e2i.org/e2i/ceids/technical/DER.html
7.13. Energy Networks Association (originally produced by the Electricity
Association), “Engineering Recommendation G74: Procedure to Meet the
Requirements on IEC 909 for the Calculation of Short-Circuit Currents in
Three-Phase AC Power Systems”, UK, 1992
139
8. Scenario Analysis
Distribution network planners have always faced uncertainty, for example in load
forecasting. However, trends like industry restructuring and DG, as outlined in the
Chapter on drivers and new directions, are increasing the factors over which DNOs
cannot exert control and increasing the level of uncertainty and risk.
A number of approaches are widely used by power system planning engineers to deal
with uncertainty and risk. The most common approaches used are scenario analysis,
decision trees and sensitivity analysis. These provide planners with a means to
quantitatively assess an uncertain situation and explore the boundaries of what might
happen. This is essential to fully understand the full range of issues and options in a
decision problem. In addition, the use of more sophisticated risk analysis techniques
have been proposed for power network planning [8.1, 8.2]. Where uncertainty
proves difficult to characterise and there is the prospect of radical change at some
unknown point in the future, it is important to ensure that systems and approaches are
flexible and adaptable to change.
This Chapter discusses scenario analysis and some of the other methods available to
manage uncertainty and risk in distribution network planning. It is proposed that
flexibility is the only way to effectively deal with uncertainty. A novel scenario
development approach is demonstrated with an assessment of the expected growth in
DG in low voltage grids. This spotlights the primary challenge currently facing
planners and shows how gathered information can be used to generate scenarios to
support further analysis. The scenarios developed here are useful in providing base-
line forecasts of DG in low voltage grids. This provides DNO planners and other
researchers with something they can work with to explore the impact of DG and start
to devise the flexible solutions that will be robust to a range of uncertain futures.
140
8.1. Methods for Managing Uncertainty and Risk
This section reviews some of the methods available for managing uncertainty and
risk in electricity distribution network planning, and proposes that flexibility is the
most important feature of planning tools and processes.
8.1.1. Scenario Analysis
Scenario analysis allows planners to forecast a range of possible outcomes and then
quantitatively assess the boundaries of their decision problems. With reference to the
decision analysis structure described in Chapter 6, it can be presented as involving
four main steps [8.3]. The first step is to select the network development options or
strategies to be evaluated. Following this step a number of scenarios must be
constructed by considering each of the development issues and assigning plausible
values to uncertain parameters. For each combination of scenario and option,
technical analysis must be performed to evaluate the outcome in terms of a score or
other overall value. The most suitable option will then be selected according to a
given decision criterion. Commonly used decision criteria include the following
[8.3, 8.4]:
• With the Expected-Score (or Expected-Cost) criterion, a probability value is
assigned to each scenario and the weighted average of the total scores of a
particular option under the different scenarios yields an Expected Score for each
option. The option with the maximum Expected Score is then selected for
implementation.
• The Laplace criterion is similar to the Expected Score criterion except that all
scenarios are assumed to have an equal probability of occurrence.
• The Minimise-Maximum-Regret criterion represents a risk analysis approach in
contrast to the probabilistic approach of the Expected-Score criterion. An ideal
option is identified in each scenario and the regret for each of the other options
calculated – regret is the difference between the option score and the score of the
ideal option. The option with the lowest regret value over all scenarios is
selected as the most desirable choice for implementation.
141
• In the Von-Neumann-Morgenstern criterion, it is assumed that one of the
extreme scenarios will occur, either the most pessimist or optimistic viewpoint.
The option that performs best under this scenario is selected for implementation.
• The Hurwicz criterion is a variation of this that allows the explicit consideration
of the decision-maker’s attitude to risk using a parameter , which is given a
value between 0 and 1. If = 0 or 1 the results correspond with the pessimistic
and optimistic scenarios respectively considered under the Von-Neumann-
Morgenstern criterion.
Although each of these decision criteria has its own merits and collectively can allow
uncertainty to be dealt within in a number of innovative ways the key disadvantage
with such approaches is related to the number of scenarios. In order to readily apply
scenario analysis techniques an appropriate number of scenarios must be considered.
However, if the number of scenarios being considered becomes too large, which can
very easily be the case when multiple sources of uncertainty are included and all
combinations of events considered, the application of such approaches becomes
highly impractical because of the large analysis burden and difficulty in interpreting
results.
8.1.2. Decision Trees
Decision trees are also commonly used to deal with planning uncertainty [8.3, 8.4].
With this technique, an event tree describing each possible course of action is
developed with each future represented by a separate branch of the tree. By
developing a path through the tree to an end-level future or course of action for each
uncertainty, the effects of time-related decisions can be assessed. The structure of
the decision tree provides a useful graphical display of each event and the path of a
particular decision is useful in providing insight into the problem itself. While the
main advantage of the decision tree technique is the graphical display and associated
visual representation, as with scenario approaches, including a large number of
events and all possible courses of action for these events may quickly lead to a
decision tree with an impractical number of end-level nodes.
142
8.1.3. Sensitivity Analysis
Sensitivity analysis is also commonly used to assess the sensitivity of outcomes to

changes in parameters. In power system planning for example, it is unlikely that the
exact value of a particular variable (e.g. expected load at a specific time of day) will
be known with absolute certainty. Statistical distributions can however be used to
quantify the expected bounds in simulation results [8.5] and when incorporated
within a sensitivity analysis will bring increased mathematical rigour to what can
otherwise be a fairly ad hoc assessment. The key issue with sensitivity analysis is
ensuring that a sufficient number of parameter changes are performed and also that
they adequately represent the various types of uncertainties and possible events. As
with scenario analysis and decision trees, assessing the impact of uncertainty
associated with multiple sources each with a wide range of disparate values or events
may lead to an excessive number of parameter changes being considered within the
sensitivity analysis. Sensitivity analysis was used in the MCDM example in Chapter
6.
8.1.4. Probabilistic Choice Versus Risk Analysis
In managing uncertainty and risk, there is a choice to be made between the

probabilistic choice and risk analysis paradigms [8.1]. For power system planning,
the risk analysis approach is preferable because it focuses on decisions rather than
the possible solutions, as is the case with probabilistic choice. Probabilistic choice
relies on the law of large numbers and the specification of probabilities. A drawback
associated with the three approaches described above is that events or outcomes that
are considered to have a low probability of occurrence are often disregarded or not
included in the assessment, regardless of the consequences of the event actually
occurring. Many power system planning decisions are not frequent enough to rely
on estimated probabilities, and the possibility of catastrophic situations disrupts the
stochastic process. Nevertheless, power system planning still relies on forecasts of
the future and an assessment of probabilities in areas like load forecasting or failure
rates for low-value but numerous assets, such as those at low voltage.
143
8.1.5. Flexibility as a Means of Dealing with Uncertainty
Considering the new challenges facing planners, as outlined in Chapter 3, and the
review of methods for managing uncertainty described above, it is clear that in the
management of uncertainty and risk, the only certainty is that there will be
uncertainty. When faced with an unpredictable future, the best way to reduce risk is
to maintain flexibility. Thus, the electricity sector in general must be made more
flexible, adaptable and responsive so that they are able to handle any unforeseen or
sudden changes. This applies both to the network extensions that are designed and
built, and to the tools and methodologies used by planners. It was already noted in
Chapter 4 that the complexity and diversity of distribution network planning and the
range of new challenges to be faced mean that highly prescriptive and detailed
methodologies are inappropriate because they are inflexible
Flexibility in tools and methodologies could be achieved in a number of ways. For

example, processes and methods should not be based on very narrow assumptions
because they will ultimately have to be modified to accommodate changed
circumstances. Software selected for activities such as power system simulation
should have scope for extending the type of analysis performed and incorporating
models of new technologies. Perhaps most importantly, people must be trained to be
able to assess changing circumstances and adapt their own behaviour as unforeseen
events unfold. However, care must be taken to ensure that flexibility is not used as
an excuse to avoid investing in tools and systems that improve efficiency.
These initial proposals are limited in their application but the provision of flexibility
in planning tools and methods is probably an area deserving of further research.
8.2. Distributed Generation Penetration Scenarios
The first level one task in the knowledge model of the conventional approach is
“Identify Requirements”. Managing the expansion of DG requires an understanding
of the expected levels of DG and relative capacities of different DG technologies. To
this end, and as part of the DISPOWER project [8.6], an assessment of the likely
144
penetration of DG in different low voltage (LV) grids was conducted. This allows
the specification of realistic scenarios to enable further analysis and help planners to
assess the problems to be addressed in their networks.
8.2.1. Methodology for Penetration Assessment
This assessment employed an original methodology in which the future penetration

of DG was analysed from two perspectives, termed “top-down” and “bottom-up”.
• The top-down approach involved the collation of forecasts made at European,
national or regional levels by various governments and other institutions. The
top-down approach provides a high-level perspective of the total expected levels
of DG.
• The bottom-up approach involved a survey with a questionnaire in which
respondents were asked to rank different DG technologies in terms of their
suitability to different LV grid scenarios. The bottom-up approach identifies
which DG technologies are most likely to be installed in different LV grid
circumstances.
The assignment of DG to LV grid scenarios involves combining the results of the

top-down and bottom-up approaches, while also taking into account the objectives of
any analysis being performed. Three different strategies for performing this
assignment are proposed.
This methodology contrasts with earlier approaches [8.7] firstly by interpreting high-
level national forecasts for DG or renewable power in terms of the likely impact on
LV grids, secondly by producing quantitative technology rankings based on a new
survey of expert opinion, and thirdly by offering three strategies for combining the
two perspectives to produce scenarios of interest. This combined approach provides
results that are robust from the perspective of high-level forecasts of DG penetration
and from the perspective of how suitable each technology is to a particular LV grid.
The new methodology is described in more detail and demonstrated in the sections
below.
145
8.2.2. Top-Down Approach
A number of governments and other organisations have produced forecasts of the

uptake of DG and related technologies. Many forecasts have been produced for
renewable energy technologies but these include large-scale technologies that will
not be installed at LV and often exclude small-scale non-renewable technologies that
will form a part of the LV DG mix. However, no specialised forecasts were
identified covering the specific topic of DG penetration in LV networks. The data
from forecasts had to be processed to take account of these issues and a number of
assumptions had to be made.
EU directive 2001/77/EC commits countries in the EU to setting targets for

renewable energy in 2010 [8.8]. Meeting these targets will require an expansion of
both large-scale installations and renewable DG, including DG in LV grids. Thus,
the targets – shown in Table 8.1 – provide some indication of how DG might grow in
EU member countries (the EU 15 pre 2004).
Table 8.1 Electricity from renewable sources in 1997 and targets for 2010 for EU countries
Electricity from Electricity from Target for electricity

renewable sources in renewable sources in from renewable
1997 (TWh) 1997 (% of total) sources for 2010 (%
of total)
Belgium 0.86 1.1 6
Denmark 3.21 8.7 29
Germany 24.91 4.5 12.5
Greece 3.94 8.6 20.1
Spain 37.15 19.9 29.4
France 66 15 21
Ireland 0.84 3.6 13.2
Italy 46.46 16 25
Luxembourg 0.14 2.1 5.7
Netherlands 3.45 3.5 9
Austria 39.05 70 78.1
Portugal 14.3 38.5 39
Finland 19.03 24.7 31.5
Sweden 72.03 49.1 60
United Kingdom 7.04 1.7 10
Total 338.41 13.9 22
146
More detailed forecast data was obtained for Italy, Poland, Germany and the UK.
The processed data is presented below and the results drawn together in a summary.
8.2.2.1. Italy
It is estimated that by 2010 Italy [8.9] will have a total installed capacity of
generation of between 95 and 110GW. Table 8.2 shows the estimated installed
capacities of various DG technologies in 2010 in MW and as a percentage of
102.5GW – the middle of the range of estimated total capacity.
Table 8.2 – DG Penetration Forecast for Italy
DG Technology Estimated capacities in Percentage of estimated

2010 (MW) total installed capacity
Thermal plants (<25MW) 5500 – 9000 5.37 – 8.78
Small Hydro (<10MW) 2500 – 2700 2.44 – 2.63
Wind 2500 – 3500 2.44 – 3.41
Photovoltaic 100 – 200 0.10 – 0.20
Biomass and Waste 1500 – 1800 1.46 – 1.76
Total DG 12100 – 17200 11.80 – 16.78
8.2.2.2. Poland
In Poland [8.10], the contribution of distributed sources (unconventional and

renewable) to total electric energy production is to be increased to 7.5% by 2010.
The total from distributed sources is expected to be 14082GWh, suggesting a total
electrical energy production from all sources of 187760GWh. If an average
production factor of 45% is assumed, then the total system capacity in Poland in
2010 can be estimated to be approximately 48GW.
Table 8.3 shows the expected energy contributions from different DG technologies.
By assuming production factors for each technology, the expected installed capacity
can be estimated and given as a percentage of total system installed capacity of
48GW.
147
Table 8.3 – DG Penetration Forecast for Poland
DG Technology Expected Assumed Estimated Percentage of

energy production capacity in estimated total
production in factor 2010 (MW) installed
2010 (GWh) capacity
Small 800 0.42 217 0.45
hydroelectric
power plants
Wind power 1200 0.18 761 1.59
plants
Photovoltaic 2 0.11 2 0.004
systems
Biogas power 2480 0.32 885 1.84
Biomass fired 9600 0.32 3425 7.14
heat and power
Total DG 14082 - 5290 11.02
8.2.2.3. Germany
In Germany [8.11], renewable sources are to provide 12.5% of electrical energy by

2010. The expected energy production from renewable sources in 2010 is 77TWh.
This includes energy from onshore wind, offshore wind, hydro, biomass,
photovoltaic, geothermal and other renewable sources. Many of these installations
can be expected to be DG (connected to either medium voltage or LV networks) but
some will be connected to the high voltage transmission network and should be
removed from the DG calculation. The total for renewable sources can be adjusted
by removing all the offshore wind energy and 50% of the onshore wind energy. This
leaves 55TWh as a rough approximation of the energy from renewable DG in
Germany in 2010. If a production factor of 0.26 is assumed for renewable energy
technologies, this corresponds to an installed capacity of renewable DG of
approximately 24GW. Lacking a forecast for non-renewable DG, this is assumed to
make a negligible contribution to the total amount of DG.
In Germany, the total energy production in 2001 was 477.5TWh and total
consumption was around 500TWh, the difference being made up with imports. In
2001, the total installed capacity of conventional generation greater than 1MW was
approximately 112GW. The 2001 capacity of DG less than 1MW was negligible.
The capacity of conventional generation is expected to decrease as DG becomes
more widespread but overall energy demand is expected to increase at around 0.6%
148
per annum [8.12, 8.13]. If it is assumed that this increase in demand is translated
into an increase in total capacity, whether it is conventional or DG, then the total
installed capacity of all generation sources in Germany in 2010 can be estimated to
be approximately 118GW.
Thus, DG capacity in Germany in 2010 is expected to be approximately 20% of total

installed capacity.
8.2.2.4. United Kingdom
The total installed capacity of electrical generation in the UK [8.14] in 2010 is

expected to be around 74GW. Table 8.4 shows estimates of the installed capacities
of different DG technologies in MW and as a percentage of total installed capacity.
The figure for wind power represents 50% of the total expected onshore wind power;
offshore wind was ignored for this DG penetration study, as was CHP above
50MWe.
Table 8.4 – DG Penetration Forecast for the UK
DG Technology Estimated capacities in Percentage of estimated

2010 (MW) total installed capacity
Wind (adjusted) 1500 – 4000 2.03 – 5.41
Biomass including energy crops 300 – 1000 0.41 – 1.35
Landfill gas 1100 – 1100 1.49 – 1.49
Waste incineration 500 – 500 0.68 – 0.68
Domestic (Micro CHP + PV) 0 – 2400 0.00 – 3.24
Mini CHP (5-500kWe) 200 – 300 0.27 – 0.41
Small CHP (500kWe-5MWe) 500 – 800 0.68 – 1.08
Medium CHP (5-50MWe) 2000 – 2700 2.70 – 3.65
Total DG 6100 – 12800 8.24 – 17.30
8.2.2.5. Summary of Results
The DG penetration forecasts for Italy, Poland, Germany and the UK are summarised
in Table 8.5. The results were calculated as percentages of the total installed
capacity of electrical generation. By assuming a plant margin (peak load / installed
capacity) of 85% and an average load factor (average demand / peak demand) of
65%, the DG penetration levels can be expressed as percentages of peak load and
average load.
149
Table 8.5 – Summary of DG penetration forecasts
Forecast 2010 DG capacity as 2010 DG capacity as 2010 DG capacity as

percentage of total percentage of peak percentage of average
capacity load load
Italy 11.80 – 16.78 13.88 – 19.74 21.36 – 30.37
Poland 11.02 12.96 19.95
Germany 20 23.5 36.2
UK 8.24 – 17.30 9.69 – 20.35 14.91 – 31.31
Minimum 8.24 9.69 14.91
Average 14.14 16.63 25.58
Maximum 20.00 23.50 36.20
These forecasts for 2010 represent a significant increase in DG but they should be
compared with existing DG penetration levels in Denmark and the Netherlands. In
Denmark, DG already represents more then 30% of the total installed capacity of
generation. In the Netherlands, about 25% of the total installed generation capacity
can be considered as DG [8.15].
Thus, it is predicted that by 2010, DG (defined in general terms as being generation

connected to the sub-transmission or distribution networks) capacity will represent
between 8% and 20% of total installed generating capacity in Europe. This capacity
will represent between 10% and 23% of maximum demand and between 15% and
36% of average demand.
It is likely that much of this DG will consist of medium-sized generators and will be
installed at voltage levels above LV. It is difficult to predict what fraction of DG
will be connected at LV – defined in EN50160 as being voltages below 1kV [8.16].
However, rough figures for Germany and Denmark suggest that LV-connected DG
might represent between 3% and 30% of total DG [8.15].
Table 8.6 summarises the results of the analysis of penetration forecasts. Taking the
full range of values into account, it is forecast that LV connected DG could amount
to between 0.45% and 10.86% of average load. The mid-range prediction is for the
capacity of LV Connected DG to be approximately 4.22% of average load.
150
Table 8.6 – Penetration forecasts for LV connected DG
2010 LV connected 2010 LV connected 2010 LV connected

DG capacity as DG capacity as DG capacity as
percentage of total percentage of peak percentage of
capacity load average load
3% of DG Min.: 0.25 Min.: 0.29 Min.: 0.45
connected at LV Ave.: 0.42 Ave.: 0.50 Ave.: 0.77
Max.: 0.60 Max.: 0.71 Max.: 1.09
16.5% of DG Min.: 1.36 Min.: 1.60 Min.: 2.46
Max.: 3.30 Max.: 3.88 Max.: 5.97
30% of DG Min.: 2.47 Min.: 2.91 Min.: 4.47
Max.: 6.00 Max.: 7.05 Max.: 10.86
8.2.3. Bottom-Up Approach
The bottom-up approach to assessing DG penetration in LV grids was based on a

survey of the opinions of experts working in the DG domain. A questionnaire was
distributed with a description of 15 LV grid scenarios – as described in Table 8.7.
For each scenario, respondents were asked to rank the DG technologies from a list
provided. Rankings were requested for both 2010 and 2020.
The list of DG technologies was as follows:

• Reciprocating engines
• Micro gas turbines
• Fuel cells
• Photovoltaics
• Wind turbines
• Micro hydro
• Biogas engines
• Other DG (Please specify)
• No distributed generation
8.2.3.1. LV Grid Scenarios
The fifteen LV grid scenarios were specified firstly in terms of structure, operation,
area type and network type. Three additional criteria were identified as being
151
particularly relevant to the assessment of DG penetration: types of loads,
meteorological climate, and the social and political environment. However, it was
assumed in all scenarios that the social and political environment was positive and
supportive towards DG. A less supportive social and political environment would
simply reduce the overall level of DG and not differentiate between different
technologies.
The scenarios were specified to provide a range of types of loads and the
meteorological climates found in Europe. The 15 LV grid scenarios are described in
Table 8.7.
Table 8.7 – Summary of LV Grid Scenarios
Scenario 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Commercial Mesh, Germany
Residential Ring, Germany
Rural South Coast, Spain
Rural Link, Netherlands

Mixed Radial, Germany
Urban Ring, Denmark

Urban Radial, France
Rural Ring, Belgium

Label
Urban Link, Greece

Urban Link, Poland
Urban Meshed, UK
Rural Ring, Austria

Urban Radial, Italy
Rural, Poland
Rural, Italy
Structure Radial X X X X X X
Link X X X
Ring X X X X
Meshable X X
Radial X X X X X X X X X X X X X
Operation
Meshed X X
Rural X X X X X X
Area Type
Urban X X X X X X X X X
Overhead X X X X X X
Network Type
Underground X X X X X X X X X
Residential X X X X X X X X X X X
Retail X X X X X X X
Load Type Farms X X X X X
Offices X X X X X X X
Industrial (Small) X X X
1100
1100
1000
1050
1050
1200
1500
1500
1500
1750
1100
950
950
950
950
Meteorological Solar irradiation

Climate
Average wind speed 5 5 5 6 6 5 3 4 3 3 6 7 6 6 4
152
8.2.3.2. DG Penetration Survey Results
For each of the LV grid scenarios, respondents to the questionnaire ranked the DG
technologies in terms of which are most likely to be installed in the scenario
conditions by 2010 and by 2020. A ranking of “1” indicated the technology is most
likely to be installed in the associated grid conditions. Rankings of “2”, “3”, “4”, etc.
indicated decreasing likelihood of technologies being installed. The rankings from
different respondents were combined to calculate average rankings or scores for 2010
and 2020 for each scenario. These scores, which give an indication of which
technologies are considered most suited to each grid scenario and are most likely to
be installed, are shown in the results in Appendix A.
8.2.3.3. Summary of Questionnaire Results
Table 8.8 and Table 8.9 summarise the top three ranked technologies in each of the
LV grid scenarios for 2010 and 2020. It is apparent that photovoltaics are the
dominant technology in 2010 with reciprocating engines and wind turbines also
appearing frequently. In 2020, photovoltaics retain their dominance and
reciprocating engines and wind turbines feature strongly. However, new
technologies like micro gas turbines, biogas engines and fuel cells start to feature
more prominently.
153
Scenario Highest ranking Second highest Third highest

ranking ranking
Residential Ring, Germany Photovoltaics Reciprocating Stirling engines
engines
Commercial Mesh, Germany Photovoltaics Reciprocating Fuel cells
engines
Mixed Radial, Germany Photovoltaics Reciprocating No distributed
engines generation
Urban Meshed, UK Reciprocating Photovoltaics No distributed
engines generation
Rural, Poland Wind turbines Photovoltaics Micro hydro
Urban Link, Poland Reciprocating Photovoltaics Micro gas turbines
engines
Urban Radial, France Photovoltaics Reciprocating No distributed
engines generation
Rural, Italy Photovoltaics No distributed Reciprocating
generation engines
Urban Radial, Italy Photovoltaics Reciprocating No distributed
engines generation
Urban Link, Greece Photovoltaics Reciprocating No distributed
engines generation
Rural South Coast, Spain Photovoltaics Wind turbines Biogas engines
Rural Link, Netherlands Wind turbines Photovoltaics Biogas engines
Rural Ring, Belgium Wind turbines Photovoltaics Biogas engines
Urban Ring, Denmark Reciprocating Micro gas turbines Photovoltaics
engines
Rural Ring, Austria Photovoltaics Micro hydro Biogas engines
154
Scenario Highest ranking Second highest Third highest

ranking ranking
Residential Ring, Germany Photovoltaics Reciprocating Fuel cells
engines
Commercial Mesh, Germany Photovoltaics Reciprocating Fuel cells
engines
Mixed Radial, Germany Photovoltaics Micro gas turbines Fuel cells
Urban Meshed, UK Photovoltaics Micro gas turbines Reciprocating
engines
Rural, Poland Wind turbines Photovoltaics Micro hydro
Urban Link, Poland Photovoltaics Micro gas turbines CHP / Stirling
engines
Urban Radial, France Photovoltaics Micro gas turbines Fuel cells
Rural, Italy Photovoltaics No distributed Reciprocating
generation engines
Urban Radial, Italy Photovoltaics Micro gas turbines Fuel cells
Urban Link, Greece Photovoltaics Micro gas turbines Fuel cells
Rural South Coast, Spain Photovoltaics Wind turbines Biogas engines
Rural Link, Netherlands Wind turbines Photovoltaics Biogas engines
Rural Ring, Belgium Wind turbines Photovoltaics Biogas engines
Urban Ring, Denmark Micro gas turbines Photovoltaics Reciprocating
engines
Rural Ring, Austria Photovoltaics Micro hydro Biogas engines
8.2.4. Combining Top-down and Bottom-up Results
The top-down and bottom-up approaches provided two different perspectives on the
penetration of DG in LV grids. To propose specific DG penetrations in each of the
LV grid scenarios, the two approaches must be combined.
The top-down forecasts were calculated from system-wide, national perspectives and
suggest overall levels of DG that are quite low: between 0.45% and 10.86% of
average load. In reality, some LV grids will contain much more DG than others. In
fact it is recognised that in some instances, the amount of DG in an LV grid may
even exceed 100% of average load. Research and development in this area is
primarily concerned with grids where there is a high penetration of DG. Therefore,
while the top-down analysis provides a guide to expected overall DG levels, the
definition of scenarios for further analysis requires much larger values of DG
penetration.
155
The results of the questionnaire indicate which DG technologies are most likely to be
installed in each of the LV grid scenarios. The translation of these rankings into
actual DG penetration levels requires some decisions to be made about the overall
level of DG and the sharing of that total between technologies.
The overall level of DG in scenarios used for analysis will be influenced by the
results of the top-down approach but will be modified to produce scenarios with
higher penetrations of DG. The sharing of total DG between technologies might
follow one of three strategies, as outlined below. Any of the strategies outlined
below could be used to produce useful scenarios. The choice of strategy for
combining the top-down and bottom-up results will depend on the precise
requirements of the analysis being performed.
8.2.4.1. Strategy One – Dominant Technology
It might be assumed that in a given LV grid, one DG technology will be dominant.

This may be because the local conditions favour one type of DG significantly or DG
installation may be a result of specific initiatives. For example, the developer of a
new housing scheme may decide to install roof-integrated PV panels. This would
result in a high penetration of PV and probably exclude other DG. Thus, scenarios
may be specified with only the DG technology ranked highest by the questionnaire
results.
For example, for scenario 15 (Rural Ring, Austria) in 2010 scenarios could be
defined with only photovoltaic DG connected at LV. If it were assumed that this LV
grid contained levels of DG equivalent to the upper range of average expected levels
then DG capacity would be 11% of average load. If average load were 150kW, then
this LV grid scenario would have 16.5kW of photovoltaic DG.
8.2.4.2. Strategy Two – Prioritised but Arbitrary Assignment
The total DG level in an LV grid could be divided between technologies by assigning

larger shares to the technologies ranked more highly. The lower ranked technologies
could be neglected. The total amount of DG could be split between the top three
156
ranked technologies only. The split could be based on an arbitrary division, with the
top-ranked technology taking half, the second ranked technology taking a third, and
the third ranked technology taking the remaining sixth. This arbitrary division
between technologies would produce useable scenarios.
For example, applying this approach to scenario 8 (Rural, Poland) in 2010 would
produce LV-connected DG in the following proportion: 50% Wind turbines; 33%
Photovoltaics; and 17% Micro hydro. If this grid hosted above average DG levels
equivalent to 50% of average load, and if the average load was 100kW, then the
assignment of DG to this scenario would be 25kW Wind turbines; 16.5kW
Photovoltaics; and 8.5kW Micro hydro.
8.2.4.3. Strategy Three – Function of Score Assignment
The total level of DG in an LV grid could be divided between the top three ranked
technologies based on a mathematical function of the scores produced by the
questionnaire results. The scores are the result of combining rankings from different
respondents and provide some measure of how likely each DG technology is to be
installed. The scores could be used in a mathematical formula to weight the share
taken by each DG technology.
For example, the process described in Table 8.10 for scenario 1 (Residential Ring,
Germany) in 2010 might be followed.
Table 8.10 – Possible strategy for defining DG penetrations in Scenario 1
DG Score Reciprocal Fraction of DG Total Percentage of

Technology DG Total
Photovoltaics 1.3 1 / 1.3 = 0.77 0.77 / (0.77+0.43+0.20) 55%
Reciprocating 2.3 1 / 2.3 = 0.43 0.43 / (0.77+0.43+0.20) 31%
engines
Stirling 5.0 1 / 5.0 = 0.20 0.20 / (0.77+0.43+0.20) 14%
engines
If a very high penetration of DG were assumed with capacity equal to 100% of an

assumed 200kW average load in this LV grid, then the analysis scenario would
157
contain 110kW of photovoltaics, 62kW of reciprocating engines and 28kW of
Stirling engines.
One of the greatest challenges for distribution network planners is dealing with
uncertainty and risk. This Chapter discussed scenario analysis and some of the other
methods available to manage uncertainty and risk in distribution network planning.
It was proposed that flexibility is the only way to effectively deal with uncertainty
and some initial proposals were made as to how flexibility might be achieved in tools
and methodologies: processes and methods should not be based on very narrow
assumptions; software should have scope for extending the type of analysis
performed and incorporating models of new technologies; and people must be trained
to be able to assess changing circumstances and adapt their own behaviour. These
initial proposals are limited in their application but the provision of flexibility in
planning tools and methods is probably an area deserving of further research.
An original methodology was employed to evaluate expected penetrations of DG,

both in terms of overall capacities and in terms of the relative share of different
technologies, and therefore provide support in understanding its impact. The
methodology for analysis of penetration scenarios combined two perspectives. The
top-down approach collated forecasts at a national level to produce expected overall
levels of DG in European networks in 2010. The bottom-up approach collated
opinions on the suitability of different DG technologies to different LV grid
conditions.
The top-down forecasts suggest overall levels of DG that are quite low. Taking the
full range of values into account, it is forecast that LV connected DG could amount
to between 0.45% and 10.86% of average load. The mid-range prediction is for the
capacity of LV Connected DG to be approximately 4.22% of average load. In
reality, some LV grids will contain much more DG than others; in some instances,
the amount of DG in an LV grid may even exceed 100% of average load. Therefore,
while the top-down analysis provides a guide to expected overall DG levels, the
158
definition of interesting scenarios for further analysis may require much larger values
of DG penetration.
A questionnaire was designed to collect opinions on the suitability of different DG

technologies to different LV grid conditions. The results of the questionnaire
identified the top three ranked technologies in each of the LV grid scenarios for 2010
and 2020. Photovoltaics emerged as the dominant technology in 2010 with
reciprocating engines and wind turbines also appearing frequently. In 2020,
photovoltaics retain their dominance and reciprocating engines and wind turbines
again feature strongly. However, new technologies like micro gas turbines, biogas
engines and fuel cells also start to feature more prominently in 2020.
The DG penetration studies presented here provide useful guidance on the expected
overall levels of DG and the DG technologies expected to feature most strongly in
different LV grid conditions. To support the translation of overall DG levels and
technology rankings into actual technology penetration values, three different
strategies were proposed. The application of these strategies and the detailed
definition of scenarios for further analysis will depend on the precise objectives of
those analyses but the results and methodologies produced from this work provide a
valuable resource for scenario development.

IEEE Trans on Power Systems, vol.13, no.2, May 1998, p.643-8
8.2. Miranda,V.; “Distribution System Planning Based on a Risk Analysis
Approach”; CIRED 12th International Conference on Electricity Distribution,
1993, Subject Area 6: Design and Planning of Public Supply Systems; IEE
Conference Publication No.373, p.6.11.1-6.11.5
8.3. Van Geert,E.; “Increased Uncertainty a New Challenge for Power System
Planners”; IEE Colloquium, 1998
159
8.4. Methods for Planning under Uncertainty: “Towards Flexibility in Power
System Development”; Paper prepared by Working Group 37.10 of CIGRE
(Convenor: E. Van Geert)
8.5. Hiskens,I.A., Pai,M.A., Nguyen,T.B.; “Bounding Uncertainty in Power System
Dynamic Simulations”; IEEE Power Engineering Society Winter Meeting, 23-
27 January 2000, Singapore
8.6. Bertani,A., Bossi,C., Delfino,B., Lewald,N., Massucco,S., Metten,E.,
Meyer,T., Silvestro,F., Wasiak,I.; “Electrical Energy Distribution Networks:
Actual Situation and Perspectives for Distributed Generation”; 17th
International Conference on Electricity Distribution, CIRED 2003, 12-15 May
2003, Barcelona, Spain
8.7. Watson,A.S.; “Modelling the performance of distribution networks with
distributed generation”; Submitted for the Degree of Master of Philosophy,
University of Strathclyde; August 2003
8.8. Directive 2001/77/EC of the European Parliament and of the Council of 27
September 2001 on the promotion of electricity produced from renewable
energy sources in the internal electricity market; Official Journal L283 ,
27/10/2001 P.0033-0040
8.9. Centro Elettrotecnico Sperimentale Italiano; 2003
8.10. Polish Ministry of Environment; “Strategy for Development of Renewable
Electric Power”; 2003
8.11. German Department of Environment, Conservation and Nuclear Safety;
Forecasts under the “Erneuerbare Energien Gesetz (EEG)” (Renewable
Energies Law); 2003
8.12. Verband der Elektrizitätswirtschaft – VDEW – e.V. (German association of
power supply companies), May 2002
8.13. RWE Power AG, Germany, 29th of April 2003
8.14. Distributed Generation Co-ordinating Group, DTI and Ofgem; “Scenarios of
Distributed Generation Development”; WS1 P06-D01 V1.2; www.distributed-
generation.org.uk
8.15. DTI New and Renewable Energy Programme; “Survey Study of Status and
Penetration Levels of Distributed Generation (DG) in Europe and the US
160
(Stage One)”; Report Number: K/EL/00306/02/REP one – URN 03/896;
KEMA Limited – Tuncay Tuerkucar, David Gailey
8.16. European Committee for Electrotechnical Standardisation (CENELEC);
European Standard EN50160; “Voltage Characteristics of Electricity Supplied
by Public Distribution Systems”; November 1994
161
9. Dynamic Modelling of Wind Farms
Previous chapters have considered the application of various methods and techniques
to address the shortcomings identified in the conventional approach to distribution
network planning. Some of these methods are new to this domain while others have
been applied to varying degrees by some DNOs. All DNOs currently perform power
system analysis in some form. This is necessary to understand how the network
operates and make plans for future developments. In Chapter 4, the need for more
analysis was identified as one of the principal shortcomings in the conventional
approach. DNOs will have to extend their analysis capabilities to deal with new
technologies and take full advantage of them. In particular, at this time power
system operators are concerned with the expansion of wind power and the effect it
will have on their networks. This requires appropriate modelling and analysis of
wind farms within power system simulation software.
Although the number of wind farms has been growing strongly for a decade or more,
there continues to be a lack of models and data to facilitate the power system analysis
that DNOs and others wish to perform. The challenge of modelling wind turbines
and wind farms has been undertaken by a number of different groups, including
researchers, network operators, manufacturers and software providers. While there
have been some efforts at collaboration there has been a resistance on the part of
manufacturers to provide the information required to model the technology fully.
Many of the models that have been developed remain proprietary, specific to
particular wind turbines, and not widely available. This Chapter examines the
dynamic modelling of wind farms as a demonstration of the type of analysis DNOs
may have to do in future and an illustration of one way in which the new challenges
in network planning can be met. The findings of this research as presented here
provide a useful – and publicly available – resource for those wishing to develop
models and will hopefully advance the general understanding of wind farms and their
dynamic performance from a power system’s perspective.
162
Wind power differs from conventional generation sources in a number of ways. This
highlights how the assumption that conventional technology will be used is a
shortcoming of the conventional approach. Firstly, the power output from wind
farms is dependent on the wind conditions so is not controllable or predictable in the
same way as conventional generation. Secondly, wind turbines have, until recently,
been specified such that they disconnect from the network if there is a disturbance.
When there was very little wind power, this approach was acceptable. With a
significant proportion of system load supplied from wind farms, it is unacceptable for
wind farms to disconnect when there is a disturbance. The ability of wind farms to
“ride-through” disturbances was an important subject for investigation a few years
ago and prompted some of the early modelling efforts. Finally, modern wind farms
do not use conventional synchronous and induction generators. To facilitate
variable-speed operation, most modern wind turbines use power electronics to some
degree. The most commonly-used type of machine in new wind turbines is the
doubly-fed induction generator, which uses power electronics in its rotor circuit.
Thus, modelling new wind farms requires the development of new models and the
adaptation of simulation tools to accommodate these new generation technologies.
The precise modelling requirements are dependent on what issues are being
considered. To assess power system stability in response to network disturbances, a
simulation tool such as PSS/E [9.1] is appropriate and models must be developed that
represent the behaviour of wind farms and their interaction with the electricity
network in the frequency range of about 1 to 10 Hz, which captures the
electromechanical response of power systems.
Wind farm models for the PSS/E simulation environment were developed. This
Chapter discusses a dynamic model of a wind farm for PSS/E, including the
assumptions and simplifications that were made. The model is demonstrated with
studies that assess the equivalence of single and aggregate models of wind farms and
also the effects of a high penetration of wind farms on distribution networks.
163
9.1. List of Symbols
The following symbols are used in this Chapter. All are introduced more fully at the
appropriate point in the discussion but are listed here for reference.
• ρ is air density and can normally be assumed to be 1.225 kg/m3

• A is the area swept by the rotor in m2
• v is velocity in m/s
• Cp is the power coefficient
• θ is either the blade pitch angle or shaft twist
• λ is the tip-speed ratio
• H is an inertia constant (Ws/VA)
• T is torque, either mechanical or electromagnetic (pu)
• ω is rotational velocity of the windmill or generator (equivalent electrical rad/s)
• ω base is base frequency of the system ( = 2πf where f is grid frequency)
• K is shaft stiffness (pu torque / electrical rad)
• E is energy in joules or watt-seconds
• J is the moment of inertia (kg.m2)
• M is mass (kg)
• L is length (m)
• D is the diameter of the rotor (m)
In the generator equations, the subscripts d and q indicate direct and quadrature axis
components respectively and the subscripts s and r indicate stator and rotor
components respectively. All variables are in per unit.
• v is voltage (et is the terminal voltage)
• i is current
• R is resistance
• L is inductance (Lm is the mutual inductance)
• Ψ is flux
• iqr or IQR is the quadrature component of the rotor current
• idr or IDR is the direct component of the rotor current
164
• vqr or VQR is the quadrature component of rotor voltage
• vdr or VDR is the direct component of the rotor voltage
• P is real or active power
• Q is reactive power
• s is slip
• p is the derivative operator
9.2. PSS/E Requirements for Generator Modelling
The models were developed and tested using PSS/E version 26.2 and Digital Visual
Fortran version 5.0D on a Windows 2000 system.
PSS/E was designed for analysis of large power systems, and makes simplifications
in the representation of system components to reduce the computational burden.
Models designed for use with PSS/E need only be as complicated as the PSS/E
environment itself. This allows some simplifying assumptions to be made. PSS/E
was designed to have a bandwidth of approximately 1 to 10 Hz. The default
simulation time step is 0.01 seconds. This can be reduced but there are limits on the
effective bandwidth. For example, the transmission network representation assumes
that all transients in lines and transformers die away within each time step, so the
time step can only be reduced to a level at which this assumption remains valid. All
these requirements of the simulation environment influence the design of the models.
9.3. A Dynamic Model of a Wind Farm
To produce a complete dynamic model of a wind farm, it is sensible to split it into a

number of components (see Figure 9.1), each of which makes a different contribution
to the overall model. For some of the model components, different wind turbine
technologies require different models. For all of the components, there are choices to
be made in the approach to modelling. Each of the components is discussed below.
165
Figure 9.1 – Components of a complete wind farm dynamic model
Wind Speed Model

Aerodynamic Model
Mechanical Systems
Generator and Grid Interface
Control Systems
Ancillary Systems
Protection
Wind Farm Site Network
Public Electricity Network
9.3.1. Wind Speed Model
Wind speed is a random variable. The Rayleigh or Weibull distributions can be used
to produce a typical pattern of ten-minute average wind speeds. However, a wind
speed signal for use in dynamic simulations with sub-second time steps requires
other methods, such as those based on power spectral densities. Analysis of wind
speed spectra reveals little energy at frequencies above 3Hz [9.2]. Furthermore, in
applying a wind speed sequence to a wind turbine, it is necessary to take into account
the variation in the wind across the area of the blades. This has the effect of filtering
out higher frequencies. Filter approaches have been used to produce single-point
wind speed sequences [9.2].
Models were produced that control the mechanical power input to the standard
PSS/E induction generator models, CIMTR1 and CIMTR3 [9.1]. These can be used
to assess the impact on the network of varying wind speed.
However, the stochastic variation of wind speed is not considered a primary issue for
power system stability analysis because it is thought that in the time scales involved,
constant wind speed may be assumed. Models must be valid over the range of
possible wind speeds and the dynamic response of a turbine may vary with its power
166
output. Wind speed distributions will help determine the probability of the most
serious conditions.
9.3.2. Aerodynamic Model
The energy in the wind is collected by the wind turbine blades, which experience
aerodynamic effects. These effects can influence machine degradation and expected
lifetime, and can raise power quality issues on the electrical network. A full analysis
of the aerodynamics would involve the use of blade element theory but the various
effects can be more simply represented by introducing variations at different
frequencies in the mechanical power transmitted through the turbine. However, in
terms of the immediate dynamic response to disturbances on the electrical network, it
is thought that the aerodynamic effects can largely be neglected.
For the purposes of dynamic analysis in PSS/E, it is considered sufficient to

represent the turbine aerodynamics just by the power collected from the wind, using
Equation 9.1 [9.2].
1 ρ is air density and can normally be assumed to be 1.225 kg/m3

P = C p ρAv 3
2 A is the area swept by the rotor
Equation 9.1 v is the wind speed in m/s
Cp is the power coefficient, a measure of how much of the energy in the wind is
actually extracted, and is a function of the relative speed of the rotor and the wind.
Even if constant wind speed is assumed, changes in rotor speed in response to
network disturbances will affect the power coefficient and hence the power extracted
from the wind. The power coefficient is also affected by blade pitch angle so the
pitch control system may have to be modelled. Thus, it may be necessary to include
a power coefficient and input power calculation in a dynamic model. Heier [9.2]
gives a function for approximating the power coefficient (Equation 9.2 and Equation
9.3). The coefficients should be adjusted to reflect the characteristics of particular
turbines.
167
116
−21
1 1 0.035
C p = 0 .5 − 0.4θ − 5 e λ ′ = − 3
λ′ where λ ′ λ + 0.08θ θ + 1
Equation 9.2 Equation 9.3
vrotortip
θ is the blade pitch angle. λ is the tip-speed ratio ( λ = )
vwind
The power curve produced by the equations above must be amended to take account
of cut-in speed and cut-out speed. Imposing these limits produces a realistic
representation of power versus wind speed, such as that shown in Figure 9.2.
Figure 9.2 – Typical power versus wind speed characteristics for an 800kW wind turbine
Power (kW)
900
800
700
600
500
400
300
200
100
0
0 5 10 15 20 25 30
Wind speed, U (m/s)
9.3.3. Mechanical Systems
Very complex drive train models have been produced in the past for the purposes of
wind turbine design, but for power systems analysis simpler models suffice. Many
studies have used a single lumped mass for the rotating system but a number of
authors [9.3, 9.4, 9.5] have argued that a multi-mass model of the drive train is
required to produce accurate simulations with conventional wind turbines. It has
been shown that fluctuations in variables like voltage and power are greater with
multi-mass models than with a single lumped mass.
168
However, in modelling a wind turbine shaft it is necessary to distinguish between
constant speed and variable speed wind turbines. In constant speed wind turbines,
the shaft should be taken into account in the following situations:
• Analysis of faults, because the potential energy stored in the shaft influences the
rotor acceleration and after clearing the fault the shaft causes an oscillation
• Power quality studies, because the dynamics associated with the shaft cause
oscillations which can be observed in the power spectral density of the output
power
In variable speed wind turbines, it has been argued that variable speed achieved with
power electronics means that the mechanical and electrical systems are de-coupled,
making a multi-mass drive train model unnecessary.
A two-mass representation, neglecting damping torques, should be sufficient for

dynamic analysis with PSS/E, as shown in Figure 9.3. The diagram and associated
equations are shown below.
Figure 9.3 - Two-mass representation of a wind turbine shaft
ωM ωE
K
TM TE
HM θ HE
HM - inertia constant of the windmill (Ws/VA)
HE - inertia constant of the generator rotor (Ws/VA)
TM - mechanical torque of the windmill (pu)
TE - electromagnetic torque of the generator (pu)
ω M - windmill speed (equivalent electrical rad/s)
ω E - generator rotor speed (electrical rad/s)
ω base - base frequency of the system (electrical rad/s) ( = 2πf where f is grid
frequency)
169
K - shaft stiffness (pu torque / electrical rad)
θ - shaft twist (electrical rad)
The system is described by three differential equations:

2 H M dω M
= TM − Kθ Equation 9.4
ω base dt
2 H E dω E
= Kθ − TE Equation 9.5
ω base dt
dθ
= ωM − ωE Equation 9.6
dt
If a single-mass model, with a single inertia constant and single rotational speed, is
deemed sufficient then only one differential equation is required:
2 H dω
= TM − TE Equation 9.7
ω base dt
9.3.3.1. Determining the Windmill Inertia Constant
The inertia constant (H) is defined as the stored energy at synchronous speed per
volt-ampere of machine rating. The units of H are watt-seconds per volt-ampere (or
just seconds).
Estored Equation
H=
VAbase 9.8
For a rotating system, stored energy (joules or watt-seconds) is given by:
1 Equation
Estored = Jω 2
2 9.9
where J is the moment of inertia (kg.m2) and ω is the rotational velocity (rad/s).
The mass, and hence the moment of inertia, of a windmill hub is split between the
blades and the shaft. To determine the moment of inertia of a single blade, it can be
170
considered as a slender rod with the axis of rotation at one end. For this type of
structure the moment of inertia is given by:
1
J = ML2 Equation 9.10
3
where M is the mass of the blade (kg) and L is its length (m).
If the total blade mass of a three-bladed wind turbine rotor, Mblades (kg), is divided
equally between three blades, and if the diameter of the rotor is Dblades (m), then the
total moment of inertia of the blades is given by:
2
1 M blades Dblades
J blades = 3 Equation 9.11
3 3 2
The shaft can be considered a cylinder, for which the moment of inertia is given by:
2
1 Dshaft
J shaft = M shaft Equation 9.12
2 2
where Mshaft is the mass of the shaft (kg) and Dshaft is its diameter (m).
Substituting Equation 9.11 and Equation 9.12 into Equation 9.9 gives:
2
2
1 M blades Dblades M shaft Dshaft 2
E stored = + ω Equation 9.13
2 12 8
Converting the rotational velocity from rad/s to RPM gives:

2 2
2
1 M blades Dblades M shaft Dshaft 2π
E stored = + ω RPM Equation 9.14
2 12 8 60
Substituting Equation 9.14 into Equation 9.8 and tidying up gives:

2
π 2 ω RPM
2 2
M blades Dblades M shaft Dshaft
H= + Equation 9.15
1800 VAbase 12 8
171
9.3.3.2. The Link Between Eigen-frequency and Shaft Stiffness
The system of differential equations describing the two-mass shaft model has an
eigen-value that is the square of the natural or modal frequency of the shaft. It can
be shown that the eigen-frequency is linked to the shaft stiffness and inertias by the
following:
ω base K 1 1 2ω eigen
2
HEHM
ω 2
eigen = + or K= Equation 9.16
2 HE HM ω base HE + HM
ω eigen - natural/modal/eigen-frequency of the shaft (rad/s)
ω base - base frequency of the system (electrical rad/s)

( = 2πf where f is grid frequency)
K - shaft stiffness (pu torque / electrical rad)
HM - inertia constant of the mill (Ws/VA)
HE - inertia constant of the generator rotor (Ws/VA)
9.3.4. Generator and Grid Interface
It is in the type of generator, and associated grid interface and converter systems,
used that the most important differences lie from the perspective of dynamic stability
analysis of wind farms in power systems. The four main types of generator are
summarised in Table 9.1 and their modelling requirements described in the following
sections.
Table 9.1 – The four main types of wind turbine generator
Type Turbine Speed Generator Converter

1 Fixed Asynchronous None
2 Variable Asynchronous Partial (rotor circuit)
3 Variable Asynchronous Full
4 Variable Synchronous Full
A fully comprehensive analysis capability would include full models of all turbine
types connected to a power system. However, with some types being used only in
small numbers it may be considered unnecessary to model them fully. In the
immediate future, it is expected that the largest wind farms, and the ones that must be
suitably represented in power systems studies, will utilise machines of type 2 –
172
doubly fed induction generators with partial power converters. The modelling
options for this type of generator and converter are covered in detail below.
All the models presented here assume that the power electronic components are ideal
and act fast enough to be considered instantaneous. This assumption is fair given the
purpose of the models and the frequency bandwidth of interest (1 to 10 Hz).
9.3.4.1. Fixed Speed Turbine - Asynchronous Generator - No Converter
Most wind turbines installed up until the end of the 1990s operate at a fixed speed
and have an asynchronous, induction generator. Some turbines operate at two fixed
speeds and drive asynchronous wound-rotor generators that can switch between 4
and 6 poles. Many designs also include thyristor controlled soft-start features. As
standard induction generators, dynamic models of these machines should be readily
developed although, as mentioned above, it will be necessary to include multi-mass
drive train models.
There is also some debate about the representation of wind-powered induction

generators in dynamic stability programs. Knudsen and Akhmatov [9.6] argue that
an EMTP-type transient model of the generator including representation of stator
transients must be used. However, this is inconsistent with the neglect of stator
transients in dynamic stability programs like PSS/E, as explained by Kundur [9.7].
This issue is discussed in greater detail in the next section.
9.3.4.2. Variable Speed Turbine - Asynchronous Generator - Partial Power

Converter
The type of turbine currently being installed in the majority of large wind farms is
variable speed with a doubly fed wound rotor induction generator. The variable
speed capability, provided by a power electronic converter controlling the rotor
voltage, is limited and only a small proportion of the total output power is fed
through the converter. This allows the power electronics rating to be approximately
25% of the overall machine rating. This makes the machine cheaper than equivalent
machines that require power electronics with a 100% rating.
173
The models described below use the d-q transformation, which was introduced in the
past to simplify analysis and compensate for the lack of computational power. With
computing power much greater than it was, some modern approaches to machine
modelling dispense with transformation methods. In this case, the models need to
interface with the power system representation used in PSS/E so the d-q
transformation was used.
The use of the d-q transformation illustrates the need for new models like this to
remain consistent with existing modelling platforms even where advances in
computing and simulation technology may have superseded them. Existing system
models and their various components have been established over a number of years,
decades in some cases. It would be expensive to replace these legacy models and so
it is preferred if new models are compatible with the existing resources. With some
new technologies this will not be possible because the assumptions underlying the
existing models may not be applicable.
Standard Induction Machine Representation
The standard set of equations for induction machines is given below [9.7]. The
subscripts d and q indicate direct and quadrature axis components respectively. The
subscripts s and r indicate stator and rotor components respectively. The values are
per unit and currents are positive when the machine is generating. The variable s
represents slip and p is the derivative operator.
vds = − Rs ids − ω sψ qs + pψ ds Equation 9.17
vqs = − Rs iqs + ω sψ ds + pψ qs Equation 9.18
vdr = − Rr idr − sω sψ qr + pψ dr Equation 9.19
vqr = − Rr iqr + sω sψ dr + pψ qr Equation 9.20
The per unit flux linkage equations are as follows.
174
ψ ds = −(Ls + Lm )ids − Lmidr Equation 9.21
ψ qs = −(Ls + Lm )iqs − Lmiqr Equation 9.22
ψ dr = −(Lr + Lm )idr − Lmids Equation 9.23
ψ qr = −(Lr + Lm )iqr − Lmiqs Equation 9.24
The electromagnetic torque is given by the following.

Te = ψ qr idr − ψ dr iqr Equation 9.25
There are three different approaches to the modelling of induction machines for
stability analysis:
• Neglect of both stator and rotor transients
• Neglect of stator transients and retention of rotor transients
• Retention of both stator and rotor transients
Each of these approaches is discussed below along with the modifications required to
describe the doubly fed wound rotor induction machine.
Neglect of Both Stator and Rotor Transients
This approach was espoused by Slootweg et al [9.8]; and was implemented in PSS/E
in a slightly modified form. Stator transients are neglected to make the model
compatible with the models of other system components in stability studies,
particularly the transmission network. Rotor transients are neglected on the basis of
the argument that their dynamics are fast enough to be outside the standard PSS/E
bandwidth.
These assumptions reduce the generator model to a set of four algebraic equations.
The stator voltages are taken from the network solution at each time step and rotor
voltages are determined by the power electronics. This gives a set of four equations
with four unknowns (Equation 9.26 to Equation 9.29) that can be solved to give the
175
machine currents, and from there, the values of torque and power required for the
simulation.
vds = − Rs ids + ω s (Lss iqs + Lm iqr ) Equation 9.26
vqs = − Rs iqs − ω s (Lss ids + Lm idr ) Equation 9.27
vdr = − Rr idr + sω s (Lrr iqr + Lm iqs ) Equation 9.28
vqr = − Rr iqr − sω s (Lrr idr + Lm ids ) Equation 9.29
Neglect of Stator Transients and Retention of Rotor Transients
The standard approach to induction machine modelling in stability studies is to

neglect stator transients to ensure compatibility with the models of other system
components [9.7]. A further assumption normally made is that the rotor is short-
circuited and hence that the rotor voltages are zero. This second assumption is not
valid for the doubly fed machine so the standard second-order model must be
modified to retain the rotor voltages. This approach has been suggested by Usaola
and Ledesma [9.9,9.10] and by CIGRE [9.11]. The resultant equations are as
follows.
vds − v′d = − Rs ids + X s′iiqs Equation 9.30
vqs − vq′ = − Rs iqs − X s′iids Equation 9.31
p (v′d ) = sω s vq′ −
1
T0′
[ ]
v′d − ( X s − X s′ )iqs − ω s
Lm
Lr + Lm
vqr Equation 9.32
p (v′q ) = − sω s vd′ −
1
T0′
[ ]
v′q − ( X s − X s′ )ids − ω s
Lm
Lr + Lm
vdr Equation 9.33
Where:
ω s Lm
v′d = ψ qr Equation 9.34
Lr + Lm
ω s Lm
vq′ = − ψ dr Equation 9.35
Lr + Lm
176
Lr + Lm
T0′ = Equation 9.36
Rr
X s = ω s (Ls + Lm ) Equation 9.37
L2m
X s′ = ω s Ls + Lm − Equation 9.38
Lr + Lm
Electromagnetic torque is given by:
Te = vd′ ids + vq′ iqs Equation 9.39
In a submission to the British Isles Wind Technical Panel, Efthymiadis [9.12] has
proposed a slightly different approach where instead of retaining vdr and vqr from the
standard induction machine equations, vd' and vq' are replaced by terms incorporating
rotor voltage components.
Retention of Both Stator and Rotor Transients
Akhmatov et al [9.3,9.4,9.6] have suggested that the neglect of stator transients is not
applicable to induction machines because the effect of the DC offset on speed must
be taken into account. It is argued that when interfacing with the network, stator
transients must be ignored but they can be retained internally within the machine
model. Saenz et al [9.13] propose a similar fourth-order model. If this approach is
adopted then the equations describing the machine are as follows.
pψ ds = vds + Rs ids + ω sψ qs Equation 9.40
pψ qs = vqs + Rs iqs − ω sψ ds Equation 9.41
pψ dr = vdr + Rr idr + sω sψ qr Equation 9.42
pψ qr = vqr + Rr iqr − sω sψ dr Equation 9.43
The flux linkages are states in the dynamic model, updated by the differential
equations shown above. The stator voltages are determined by the network solution.
The rotor voltages are determined by the power electronic converter, in line with the
177
control strategies. The currents can be found by solving the system of linear
equations, Equation 9.17 to Equation 9.20, shown above.
9.3.4.3. Variable Speed Turbine - Asynchronous Generator - Full Power

Converter
The third type of wind turbine is variable-speed with a cage-rotor induction generator
connected to the grid fully through a power electronic converter. With all power
flowing through the converter, the overall dynamic response will be determined by
the power electronics. The response will depend on how the converters are
programmed. The power electronics themselves act almost instantaneously in
comparison to conventional power system components. It may be possible to
represent the converters simply by the limits of current and voltage that cause the
converters to switch off and isolate the wind turbine and generator. However, it may
be necessary to model the generator to determine when the limits of the power
electronics are reached. The standard PSS/E induction generator models may suffice
in this instance.
9.3.4.4. Variable Speed Turbine - Synchronous Generator - Full Power

Converter
The more common type of wind turbine that passes all power through a power
electronic converter uses a directly driven synchronous generator. The overall
dynamic response will be determined by the power electronics but it may also be
necessary to model the generator. The range of PSS/E standard synchronous
machine models may include something suitable to represent these machines but
these was not examined in detail.
9.3.5. Control Systems
9.3.5.1. Control Objectives
Older wind turbine designs, using conventional induction generators and installed in
relatively small numbers, had limited control objectives. Modern variable speed
wind turbines with power electronics are able to pursue additional objectives.
178
Protect Equipment
Wind turbine control is closely linked with protection. Older wind farms were
designed to disconnect from the power system at the first sign of a disturbance. This
ensured that the wind farm equipment was protected from potentially damaging
transient effects and reduced the possible negative effects on the power system.
With more wind power connected to the system, fault ride-through capability
becomes important. There is the danger that large amounts of wind power will
disconnect when there is a system disturbance. Thus, the objective of protecting
equipment must be traded against supporting the power system.
The majority of large wind turbines now being installed are variable speed turbines
that use power electronics in their interface with the grid. An important factor in
these machines is that the power electronics must be protected from excessive
voltages and currents. The limits of the power electronics can be quite low, with
higher limits incurring a higher cost in power electronic hardware. This protection is
an important objective of the control/protection system and is implemented with so-
called “crowbar” protection. This applies a short-circuit across the terminals of the
power electronics in the rotor circuit, disconnecting them and effectively turning the
generator into a conventional induction machine. Crowbar protection has not been
modelled explicitly in the models developed in this work, although it could be added.
Rotor current is accessible as a variable and its value can be used to trigger the
disconnection of the machine or some other action. When models of a doubly fed
induction generator were tested with disturbances, the current in the rotor was
identified as the first variable to trigger protection and disconnection of the machine.
In normal operation, wind turbines must be protected from mechanical stresses that
can arise from variations in wind speed and aerodynamic effects such as blade-tower
interaction. In variable speed machines, torque pulsations in the shaft can be reduced
by adjusting the electromagnetic torque of the generator. However, this results in
pulsations in the output power. Thus, there is a trade-off to be made between torque
pulsations in the shaft and power pulsations at the output.
179
Maximise Energy Capture
An important control objective of wind turbines is to maximise the energy captured

from the wind. All wind turbines turn to face the wind as its direction changes, but
energy capture is also dependent on the speed of rotation of the blades. Older wind
turbines operate at a speed that is approximately fixed. Variable speed wind turbines
can maximise the energy captured from the wind by rotating at the optimum speed.
This variation in speed can be achieved through pitch control and through control of
the generator’s electromagnetic torque using power electronics.
The power coefficient is a measure of the efficiency of energy capture. It is normally

presented as a function of tip speed ratio and pitch angle. The tip speed ratio is
found by dividing the speed of the tip of the turbine blades (proportional to rotational
speed) by the wind speed. Figure 9.4 shows how at any given tip speed ratio, the
power coefficient can be maximised by adjusting the pitch angle.
Figure 9.4 – Performance coefficient as a function of tip speed ratio with pitch angle as a
parameter
0.5
0.45
0.4
0.35
Performance coefficient, Cp
0 deg
0.3 2 deg
5 deg
0.25
10 deg
0.2 15 deg
25 deg
0.15
0.1
0.05
0
0 2 4 6 8 10 12 14 16
Tip speed ratio, lambda
180
System Support
Given the prospect of wind power representing a significant share of total generation,
system operators would like wind farms to comply with additional control
requirements [9.14]. These would be implemented at different times depending on
system conditions.
Power Control
Wind farms should be able to limit the power they are producing when commanded
to do so by system operators. This may be to alleviate constraints on the network or
to respond to a high frequency. Wind farms may also be asked to participate in more
general frequency control by operating at 90% of their potential output, thereby
providing scope for increasing and decreasing output power as required. In addition,
the rate of change of power from a wind farm should not exceed a specified rate.
This may require switching off turbines in anticipation of high wind speeds that will
force turbines to switch off anyway.
Voltage Control
The power electronic converters used on variable speed turbines have the capability
to control reactive power supplied to the grid. Large wind farms will be expected to
contribute to network voltage control. This will require being able to operate across
a range of power factors.
In comparison with conventional generators, wind farms present additional problems

because they consist of a number of small machines rather than a few large
machines. Each wind turbine has its own transformer to step the voltage up to
connect to the wind farm site network. The site will then be connected to the wider
power network through more transformers. The additional transformer on each wind
turbine will absorb reactive power, reducing the effectiveness of any reactive power
contribution from the generator.
181
Power Quality
The effect of wind farms on power quality must be considered and the mitigation of
power quality problems can be a control objective.
The fluctuating wind speed combined with blade transients and blade-tower
interaction mean that the torque from a wind turbine is not steady. This can be
translated as flicker on to the power system. Absorbing torque fluctuations in the
mechanical systems of the wind turbine, by allowing speed to vary, may reduce
flicker but this imposes stress on the turbine shaft.
The power electronics used in variable speed machines may introduce harmonics on
to the network. Care must be taken to ensure this does not affect other grid users.
Farm-Wide Control
Large wind farms being installed today consist of dozens of turbines, each with their
own control systems. Power system operators are concerned primarily with the wind
farm as a whole. Control must be exerted over the whole wind farm, either to share
the burden of control requirements, or to shutdown and start-up individual turbines as
required.
9.3.5.2. Control Options
Yaw Control
The yaw control systems of wind turbines turn the nacelle and rotor to face into the
wind. In large turbines, this involves moving a very large object so the dynamics of
yaw control are too long-term to have an impact on normal power system stability
studies. It will be necessary to model longer-term dynamic effects such as yaw
control if simulations are to cover a period of minutes, and if changes in wind
direction are considered.
182
Blade Pitch Angle Control
In older and smaller turbines, the aerodynamic stalling effect is used to limit the
speed of rotation and the power from the generator. Most manufacturers now use
blade pitch angle control for large turbines, sometimes in combination with stall
control.
As shown in Figure 9.4, the pitch of the blades influences the energy captured from
the wind. Pitch angle control can be used to maximise the energy captured up to the
limit of the generator, and to limit the rotor speed in strong winds. It may be used to
smooth the flow of power from the machine, or to reduce output in response to a
system disturbance. Pitch control may also be used to operate the turbine below its
full potential then provide a rapid power increase on demand. This kind of response
may be a requirement of system operators.
Although modern wind turbines are very large with blades that might be 40m or
more, pitch control can act very quickly. It is possible to move the blade pitch angle
through its whole range in a matter of seconds. In normal operation, the blade pitch
will change at around 5-10 degrees/second. In emergencies, the blade pitch can be
made to change at 10-20 degrees/second. Thus, blade pitch angle control should be
considered in power system stability studies. In addition, the blade positioning
system consumes approximately 1% of a turbine’s rated power in normal use and
more in extreme situations [9.2].
The wind turbine aerodynamics form a complex, non-linear system. Publications

suggest that from a measured wind speed, a non-linear function or look-up table is
used to determine the optimum rotor speed and blade pitch angle. Pitch angle control
is used in combination with electromagnetic torque control, described below, to bring
the turbine to the desired state.
Publications suggest that once the desired pitch angle has been determined, a
Proportional-Integral (PI) controller is used in a feedback control loop, taking into
account the dynamics of the actuator and the blades themselves.
183
Power Electronic Converters
The doubly fed induction generator uses power electronic converters to control
currents and voltages in the generator and to control the interface with the grid. The
main task of the back-to-back converters in the rotor circuit is to regulate the active
and reactive power of the induction machine. The converters use pulse width
modulation (PWM) to inject the desired voltages into the rotor circuit.
The generator stator is directly coupled to the grid and this circuit carries most of the
power, typically around 90%. The converter closest to the rotor can inject voltages
to the rotor circuit to control the speed (or slip) and reactive power of the machine.
The converter closest to the grid can act like a STATCOM, controlling reactive
power exchange with the grid. The converters are bi-directional so power can flow
both into and out of the rotor circuit, depending on the speed of the turbine.
The rotor-side converter of a doubly fed induction generator acts as a voltage source
in the rotor circuit. Most publications use the rotor currents as intermediate
variables, which are converted to voltages for injection into the rotor circuit. In some
models, the desired rotor currents may be used directly in the electrical model of the
machine. It has been found that reactive power is directly dependent on the direct
axis component of rotor current and that electromagnetic torque is directly dependent
on the quadrature axis component of rotor current.
The use of controllers in rotor current control varies across publications. In

particular, the choice of algebraic functions or PI controllers at different points in the
control loop varies. This is discussed in more detail in the sections below on
electromagnetic torque / speed control and reactive power / voltage control.
Electromagnetic Torque and Speed Control

In the doubly fed induction generator, power electronics in the rotor circuit allow the
electromagnetic torque of the generator to be controlled. This can be used to control
the torque exerted on the wind turbine shaft and therefore the speed of rotation.
184
Fluctuations in the torque from the wind turbine can exert significant mechanical
stresses on the turbine shaft. Adjusting the electromagnetic torque of the generator
to follow the changes in torque from the turbine can reduce these. The twisting
torque on the shaft is then reduced. However, varying the electromagnetic torque
results in variations in the output power from the generator. Thus, the converters can
reduce the mechanical stresses on the wind turbine or reduce the fluctuations in the
power delivered but there is some conflict between these objectives.
As described in the section on pitch angle control, for a given wind speed, there is an
optimum rotating speed for the wind turbine. The generator’s electromagnetic torque
can be adjusted to bring the turbine to that optimum speed. Thus, turbine speed can
be controlled through two separate means – pitch angle and electromagnetic torque –
and these must be co-ordinated within the control system.
Most publications on the doubly fed induction generator note that electromagnetic
torque is directly dependent on the quadrature axis component of rotor current.
Slootweg et al [9.8] present the following equation, Equation 9.44, as the relationship
between rotor quadrature current and electromagnetic torque.
Te is the electromagnetic torque from the generator
Lm et Lm is the mutual inductance
Te = − iqr
Ls + Lm
Ls is the stator leakage inductance
Equation 9.44 et is the terminal voltage
iqr is the quadrature component of the rotor current
The relationship between electromagnetic torque and the quadrature component of

rotor voltage, VQR, can be investigated using the standard equations for induction
machines – see section 9.3.4.2.
Figure 9.5 shows the resultant electromagnetic torque for changes in VQR and VDR,
representing control action by the rotor-side converter. VDR was held constant at
zero while VQR was varied, and vice-versa. Other parameters were held constant
185
throughout: VDS=0, VQS=0.99, RS=RR=0.01, LS=LR=0.1, LM=3.0, SLIP=-0.005.
A linear approximation is shown to TELECR as a function of VQR.
This analysis suggests that it should be possible to use VQR directly as the control
variable for electromagnetic torque, rather than using IQR as an intermediate
variable.
Figure 9.5 – Response of electromagnetic torque to changes in rotor voltages
TELECR for VQR TELECR for VDR Linear (TELECR for VQR)
15
y = 96.252x + 0.7771
10
TELECR (pu)
0
-0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1
-5
-10
VQR or VDR (pu)
The use of controllers in determining IQR and VQR varies across publications. In
some [9.8, 9.15], IQR is determined using an algebraic or numerical function of rotor
speed, reflecting the non-linear turbine characteristic. A dynamic component, in the
form of a PI controller, may be introduced in the conversion of rotor current to rotor
voltage. Alternatively [9.9, 9.16], the rotor currents can be the outputs of PI
controllers with rotor speed as inputs. These may then be transformed algebraically
to rotor voltages or passed through another PI dynamic block to produce the rotor
voltages for injection to the machine. Another alternative [9.10] is to use a PI
controller with rotor speed error as the input and electromagnetic torque as the output
then determine rotor quadrature current as a proportional function of electromagnetic
186
torque. Yet another alternative [9.17] is to use Proportional-Integral-Derivative
controllers with power and reactive power as inputs and the rotor currents as outputs.
Depending on its design, the control system may require inputs of actual and desired
rotor speed, actual and desired electromagnetic torque, and actual and desired rotor
currents and voltages. Speed, voltages and currents can be measured but in real
machines, electromechanical torque must be calculated from other parameters, as it
cannot be measured directly.
Rostoen et al [9.16] present a simple cascaded PI control system for control of

electromagnetic torque and rotor speed; see Figure 9.6. It can be seen that two PI
controllers are used in determining the rotor quadrature current, IQR, which is then
transformed, either algebraically or with another PI controller, to VQR for
application to the rotor side converter.
Figure 9.6 – Cascaded PI controller for rotor quadrature current
desired rotor
PI PI
rotor + + quadrature
- Controller - Controller
speed current
actual
rotor electromagnetic
speed torque
Rostoen et al [9.16] note that fast control of electric torque (with a small time
constant in the torque control loop) results in oscillations in the torque transmitted
through the shaft when there is a step change in the input torque from the turbine. If
there are mechanical reasons to avoid oscillations in shaft torque then a larger time
constant will help reduce those oscillations. To avoid unnecessary oscillations, the
gain at the resonance frequency of the shaft system should be small.
Torque pulsations from the turbine, perhaps from tower vortex interaction, can be
damped with a large proportional gain constant in the speed control loop. This
makes the electromagnetic torque vary more, thereby tracking the pulsations from
the turbine and reducing the pulsations transmitted in the shaft. However, the
pulsations in electromagnetic torque will also appear in the output power. Thus, in
187
setting the gain constant in the speed control loop, there is a trade off between
pulsations in the shaft torque and pulsations in the output power.
Reactive Power and Voltage Control

In the doubly fed induction generator, power electronics in the rotor circuit allow the
power factor of the generator and reactive power exchanged with the grid to be
controlled. The rotor side can control the reactive power of the generator, and the
grid side converter can act like a STATCOM, providing an alternative control option
for reactive power exchange with the grid.
Most publications on the doubly fed induction generator note that reactive power is
directly dependent on the direct axis component of rotor current. Slootweg et al [9.8]
present the following equation, Equation 9.45, as the relationship between rotor
direct current and reactive power.
Qtotal is the total reactive power produced by the machine

L i
Qtotal = − m dr
Ls + Lm Lm is the mutual inductance
Ls is the stator leakage inductance
Equation 9.45
idr is the direct component of the rotor current
Thus, in Slootweg’s model, the rotor direct current is used to control reactive power
in the machine. The direct component of the rotor current is split into two parts:
et2 Equation
idr , mag = − ,
ω s Lm 9.46
which is required to magnetise the generator itself; and

idr , gen ,
which determines reactive power exchanged with the grid.
The relationship between reactive power and the direct component of rotor voltage,
VDR, can be investigated using the standard equations for induction machines – see
section 9.3.4.2.
188
Figure 9.7 shows the resultant real and reactive power for changes in VDR,
representing control action by the rotor-side converter. Other parameters were held
constant: VDS=0, VQS=0.99, VQR=0, RS=RR=0.01, LS=LR=0.1, LM=3.0, SLIP=-
0.005. It can be seen that reactive power varies linearly with VDR.
This analysis suggests that it should be possible to use VDR directly as the control
variable for reactive power, rather than using IDR as an intermediate variable.
Figure 9.7 – Response of power and reactive power to changes in direct axis rotor voltage
P Q Linear (Q)
10
8
y = 95.771x - 0.3308
6
2
pu P and Q
0
-0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1
-2
-4
-6
-8
-10
VDR
As with the IQR/VQR controllers for electromagnetic torque and speed, there are a
variety of controllers for IDR/VDR presented in publications. This covers different
combinations of algebraic and dynamic functions for determining IDR and VDR.
Rostoen et al [9.16] present a very simple PI control system for control of reactive
power; see Figure 9.8. A single PI controller is used to determine IDR, which could
then be transformed, either algebraically or with another PI controller, to VDR for
application to the rotor side converter.
189
Figure 9.8 – Simple PI control system for reactive power
desired rotor
PI
reactive + quadrature
- Controller
power current
actual reactive
power
An alternative approach to reactive power and voltage control is to use the grid side
converter like a STATCOM device [9.18]. The rotor side converter can be used to
provide reactive power for magnetisation of the generator. The grid side converter
can then have sole control over reactive power exchange with the grid. In
publications on doubly fed induction generators, this approach is less common than
using only the rotor side converter.
The power electronics have limitations and their reactive power capabilities vary
depending on conditions. Meeting the reactive power requirements may affect other
aspects of the machine’s operation. For example, it has been claimed that operating
at low power factors results in an increase in flicker. This may be because the
voltages and currents required in the converters to produce the desired reactive
power reduce the scope to mitigate fluctuations in torque from the turbine.
9.3.5.3. Modelling Control Systems in PSS/E
Despite efforts by network operators and others over a number of years to have
information made available, wind turbine manufacturers are reluctant to release
details of their controllers, seeing them as valuable intellectual property. This
missing information presented one of the main challenges in modelling wind farms.
The only controller models available were those proposed by university researchers
in publications. For a DNO wishing to develop models and perform studies, the
work done and presented here should prove useful.
As described in the sections above, most publications suggest PI controllers for

control of reactive power and speed. Cascaded control loops are frequently used;
190
particularly to control speed where there can be cascaded control loops for blade
pitch angle, rotor speed and electromagnetic torque.
For example, Tapia et al [9.19] describe a control structure with two cascaded control
loops. Numerous transformations are required between the three-phase
measurements from the machine, the two-phase representations used in the
controllers, and the three-phase voltages injected into the rotor. The details of how
the desired rotor currents are specified for desired values of P and Q are not given.
PI controllers are used to convert the rotor current errors to desired rotor voltages.
Heier [9.2] identifies a number of advanced control methods that may be applied to
wind turbines including fuzzy controllers, self-tuning control systems, system-
oriented controller design, and neural networks. The non-linear nature of a wind
turbine means that parameters can cover a very broad range of values depending on
the status of the turbine [9.20]. It is argued that non-linear control systems are
required to improve control behaviour.
The models developed allow the direct and quadrature components of the rotor
voltage to be varied independently. The models have been set up such that external
controller models can control the rotor voltages. This is possible with the PSS/E
modelling framework, which allows the separate specification of generator,
governor/prime mover, and exciter/automatic voltage regulator (AVR) models. It is
also possible to specify power system stabilisers and other components. These
separate models facilitate the testing of different control strategies with the same,
generic generator model.
9.3.6. Ancillary Systems
The ancillary systems associated with wind farms will affect the dynamic response to
varying degrees. Ancillary systems may influence when machines are tripped off
during disturbances. For example, uninterruptible power supplies may be required to
keep control systems operating. Features such as compensating capacitors will also
affect the dynamic response. Many wind turbines use thyristor soft-start systems to
191
reduce the impact of induction machine start-up. In general, the modelling
requirements for ancillary systems are likely to vary on a case-by-case basis and are
not dealt with further here.
9.3.7. Protection
At the interface between a wind farm and the wider power network, protection
systems should be designed to comply with the relevant standards. However, the
introduction of very large wind farms may necessitate changes in the approach to
protection. Dynamic simulations with accurate wind turbine models will be required
to determine the most appropriate protection settings. There is scope to incorporate
protection models in dynamic simulations but the action of protection devices may
also be simulated through manually enacting switching events. Network operators
and wind farm operators will work together so that settings grade with network
protection.
New requirements for the connection of large wind farms to electricity networks
have been developed. These build on existing, long-standing recommendations for
connecting DG in the UK; Engineering Recommendation G59/1 requires that the
protection systems must be able to detect the following conditions:
• Over and under voltage
• Over and under frequency
• Loss of mains
• Neutral voltage displacement
• Over current
• Earth current
• Reverse power
9.3.7.1. Protection Modelling in PSS/E
PSS/E includes a library of models covering generators, excitation systems, turbine-

governors and some relays. User-developed models, written in a Fortran derivative
language called Flex, can also be added. The PSS/E library models were examined
192
to determine whether any were appropriate to wind farm protection modelling.
However, it was found necessary to write additional code to provide the functionality
required. A user-developed model of a reverse power relay was produced. It uses
the PSS/E function LINTRP to trip a line when the appropriate reverse power
conditions are met.
In conclusion, modelling protection was not a primary consideration in the design of

PSS/E version 26. In particular, the program is not capable of dynamic simulation of
unbalanced conditions so some aspects of wind farm protection could not be
modelled. Nevertheless, as the program is used for other purposes, it is useful to
develop protection modelling capabilities. User-developed models offer flexibility in
modelling different types of protection. It is also possible to incorporate protection
effects within models of a wind farm or wind turbine.
9.3.8. Wind Farm Electrical Network
Large wind farms cover large areas of land resulting in a significant electrical
network within the wind farm site itself. This may necessitate modelling of the wind
farm electrical network in power systems analysis performed by the operator of the
public system. If the wind farm electrical network is a conventional design then this
should not pose any great challenges, except obtaining data from wind farm
operators. However, large wind farms of the future may utilise unconventional
electrical networks, such as direct current networks. Such designs would have to be
carefully assessed.
9.3.9. Electricity Network
To accurately determine the effect of wind farms on the network, and vice-versa, the
electricity network must be represented accurately. Network operators would use
their own data to model portions of their network as necessary.
The network representation used in PSS/E is greatly simplified because the program
is designed to simulate very large networks. This restricts the interaction between
the network and machines and influences the approach to machine modelling.
193
9.4. Model Validation or Verification
Operational data is required to validate dynamic models. Test results from

manufacturers and wind farm developers would support the development and
validation of models. However, efforts to obtain useful data from manufacturers
failed to produce the specific data required for dynamic modelling. This continues to
be a problem for those wishing to develop and validate dynamic models of wind
farms. Commercial interests and the competition between manufacturers and
between software providers mean that collaboration is restricted and information is
kept secret.
An alternative to full validation is verification by comparison of results with those

produced using another model. The results from the models described here were
sufficiently similar to those reported in literature [9.3, 9.8, 9.9, 9.13, 9.16, 9.17] to be
deemed acceptable for the purposes they were put to in this research. Likewise, if
other models match the results produced in this work then this will provide some
degree of verification.
9.5. Case Studies
The dynamic models developed in this work were tested and used in a number of
different studies. The sections below present highlights from two case studies.
Fuller results from the studies are presented in Appendix B and Appendix C. The
first examined the equivalence of multiple wind turbine models and a single model to
represent a complete wind farm. The second examined the impact of a high
penetration of wind farms on a distribution network.
9.5.1. Aggregate Models of Wind Farms
The modular approach to modelling a single wind turbine provides flexibility but
from the system operator’s perspective, the response of each individual turbine
matters less than the response of complete wind farms. Thus, aggregate models of
194
complete wind farms would be preferable. However, there are a number of issues to
be considered, including:
• The presence of multiple, independently-controlled machines
• A non-uniform wind speed pattern across a wind farm
• A substantial electrical network within a large wind farm
The scope for aggregate wind farm models was explored with tests conducted to
study whether a single machine could be used in power system simulation to
represent the multiple machines in a wind farm. The results suggest that from a
wider network perspective, a wind farm with multiple machines may be equally well
represented by a single machine as with multiple machines as long as the wind farm
electrical network is not too large and so places a much higher impedance between
the network and the generators.
This is illustrated in Figure 9.9, which shows an example of the very similar
responses obtained for three different studies using three different representations of
a wind farm:
• Category A studies used a single generator of rating 20MVA
• Category B studies used ten 2MVA generators connected at a single bus
• Category C studies used ten 2MVA generators connected to separate buses in an
electrical network with low impedance lines
195
Figure 9.9 – Real power in branch 102-103 for the three categories of study set one
a1 b1 c1
30
20
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-10
-20
-30
-40
Differences in response emerged when the impedance of the lines in the wind farm
network itself were increased, i.e. if the wind farm network is bigger and spread over
a larger area. Figure 9.10 shows that bus angle responses A and B behave quite
differently to response C. In A and B, bus angle drops and settles at a new value. In
C, bus angle starts rising but is restored to its initial value when the fault is cleared.
This is due to greater line impedances in the wind farm network. This higher
impedance alters the dynamic relationship between the network and the wind turbine
models. The result is that the turbines are more stable and the bus angle at bus 103
rises then returns to its original value in response C but changes rapidly and shifts to
a new steady state value in responses A and B.
196
Figure 9.10 – Bus angle at bus 103 for the three categories of study set three
a3 b3 c3
200
100
0
0 0.5 1 1.5 2 2.5
-100
-200
-300
-400
Full details of the study and all the results can be found in Appendix B.
9.5.2. Study of the Effect of a High Penetration of Wind Farms
In the second case study, tests were conducted to examine the effect of a high
penetration of wind farms on a distribution network. A modified version of a
standard IEEE test network was used and a single model of a doubly fed induction
generator was used to represent each wind farm. Studies were conducted under three
different load conditions and six different wind farm conditions. Two main types of
study were performed: the dynamic response to a line fault; and the dynamic
response to changes in wind power.
The results in Appendix C illustrate some of the possible effects of a high penetration
of wind farms on a distribution network. These include variations in system stability
depending on wind farm output, and voltage fluctuations as a result of changes in
wind power. Some areas for possible further investigation were identified.
197
The results for the line fault disturbance tests suggest that a high penetration of wind
farms may cause network instability. The example in Figure 9.11 shows how the
smoothness of post-fault recovery worsens as the wind farm outputs are increased.
Ultimately, wind farm outputs of 20MW result in post-fault instability.
Figure 9.11 – Bus 1 voltage for medium load condition and different wind farm conditions
The system only became unstable following a line fault when the wind farms were
operating at full output and all tripped off due to the fault disturbance. The results
illustrate why fault ride-through capabilities were of great interest to network
operators at the beginning of the decade as large numbers of wind farms were
installed. In the simulations performed in this study, the wind farms do little to
support the network and have inadequate fault ride-through performance. In the last
few years the fault ride-through performance of modern turbines has been improved
with changes to the control systems used. The models developed here could be
revised to implement these new control systems as information on their behaviour
becomes available.
198
The results for the varying input power tests suggest that changes in input power will
result in changes and oscillations in the power output from wind farms, which will
cause shifts and oscillations in network voltages. The grid in-feed to a network with
a high penetration of wind farms must be able to compensate for the variations in
power that result from changes in wind speed.
For example, Figure 9.12 shows how the voltage at bus 30 – far from the swing bus
and with a wind farm connected – varies for the different wind farm scenarios in the
low load condition. The variations in wind farm power are reflected in the bus
voltage. The base level of the voltage at this bus is different for the different wind
farm scenarios; higher wind farm outputs result in higher bus voltages.
Figure 9.12 – Bus 30 voltage for low load condition and different wind farm scenarios (different
MW base levels)
Furthermore, it was shown that the oscillations caused by changes in input power
interfere with one another. The timing of such changes lead to constructive or
199
destructive interference. A brief supplementary study revealed that unfortunate
timing of changes in input power could lead to significant constructive interference
of waveforms and the exceeding of protection limits. For example, Figure 9.13
shows that with the final ramp in input power delayed by three seconds, there is
constructive rather than destructive interference of the power output waveforms
resulting in oscillations of much greater amplitude. In fact, the additive effects are
enough in scenario d to drive the machine power and rotor current magnitude high
enough that the protection is operated and the machine trips. This issue could be
investigated further. In particular, the damping of the oscillations must be improved,
perhaps through improvements in the control systems.
Figure 9.13 – Bus 30 wind farm output for low load condition and different wind farm scenarios
(different MW base levels) and the adjusted pattern of input power
The results of these simulations, which are presented in full in Appendix C, illustrate
effects that may be experienced on distribution networks with a high penetration of
wind farms. However, as noted above, the models used have not been validated
against real-life systems.
200
This chapter explained the need for dynamic models of modern wind farms and
described the development of a dynamic model for power system simulation in
PSS/E. The various options available and the associated assumptions and
simplifications were explored to provide a comprehensive statement of the problem.
Models were developed and shown to work effectively in PSS/E. It was
acknowledged that the models require validation to be applied more fully and this
was identified as one of the areas for further work. The principal barrier to model
development was, and remains, the lack of publicly available detailed information on
the behaviour of modern wind turbines, particularly their control systems.
The development of dynamic models of wind farms provides an example of the

extensions required in power system simulation to manage the changes in the
electricity system, as discussed in previous Chapters. The expansion of renewables
and DG requires new approaches and new tools to support planning and design of the
network.
From the experience described in this Chapter, it can be concluded that the
development of models of new technologies is an onerous task. It requires an
understanding of the simulation software being used as well as the physical
characteristics of the device being modelled. Information is likely to be lacking
because new technologies are subject to intellectual property protection. For
example, the control systems on the model were designed rather than representing an
actual piece of plant. There are very many design decisions that can influence the
model in different ways and each decision normally reduces the scope or
applicability of the model. A formal design process was not used in the work
described here. The application of more formal and rigorous methods may lead to
more robust models but is also likely to incur additional costs in time and effort.
Thus, it can be concluded that while new models must be developed to support
analysis of new technologies in future distribution networks, this presents a very
201
considerable challenge to DNOs. In providing a comprehensive review of the task
and offering a non-proprietary model for others to use in their work, this research
contributes to the general understanding of the dynamic modelling of wind farms.
While the models described here are limited in their scope, being designed for
assessment of power system stability at electromechanical frequencies, and have not
been validated with experimental data, they can still be of value to system planners
as they endeavour to meet the challenges facing them.
The study of aggregate wind farm models suggested that for normal power system
simulation, a single machine could be used to represent a wind farm with multiple
turbines as long as the wind farm electrical network did not place high impedances
between turbines and the grid. This would greatly simplify the modelling
requirements and is in line with the perspective of system operators, who will wish to
view wind farms as single entities.
The study of the effect of a high penetration of wind farms on distribution networks
revealed that the response depended on the wind farm operating conditions. The
wind farms rode through faults when at medium and low power output levels but
became unstable when at full power output. Thus, the results indicated that it was
possible, although unlikely, for large amounts of wind power to trip off under certain
fault conditions and so highlighted the importance of fault ride-through capabilities.
The studies of varying input power highlighted the resultant variations in network
power flows and voltages and suggested that unfortunate timing of input changes
might cause constructive interference of oscillations and lead to instability.
It is understood that modern wind turbines have more advanced control systems and
have addressed some of the potential problems highlighted in this report. However,
while the control systems and detailed information on performance remain a closely
guarded secret, it is difficult for researchers and network operators without access to
this information to perform realistic studies. When information on the actual control
systems is made available, they can be modelled within the PSS/E model framework
established in this research.
202
9.1. Power Technologies Inc.; PSS/E-26 Documentation; December 1998

9.2. Heier,S.; “Grid Integration of Wind Energy Conversion Systems”; 1998; John
Wiley & Sons; ISBN 0 471 97143 X
9.3. Akhmatov,V., Knudsen,H., Nielsen,A.H.; “Advanced simulation of windmills
in the electric power supply”; Electrical Power and Energy Systems, 22 (2000),
p.421-434; Elsevier; ISSN 01420615
9.4. Akhmatov,V., Nielsen,A.H., Knudsen,H.; “Electromechanical interaction and
stability of power grids with windmills”; Proceedings of the IASTED
International Conference, Power and Energy Systems, September 19-22, 2000,
Marbella, Spain
9.5. Salman,S.K., Teo,A.L.J., Rida,I.M.; “The Effect of Shaft Modelling on the
Assessment of fault CCT and Power Quality of a Wind Farm”; Proceedings of
2000 International Conference on Harmonics and Quality of Power, vol.3, 1-4
October 2000, Orlando, FL, USA
9.6. Knudsen,H., Akhmatov,V.; “Induction Generator Models in Dynamic
Simulation Tools”; IPST’99, International Conference on Power Systems
Transients, June 20-24, 1999, Budapest, Hungary
9.7. Kundur,P.; “Power System Stability and Control”; 1993; McGraw-Hill Inc.;
ISBN 0 07 035958 X
9.8. Slootweg,J.G., Polinder,H., Kling,W.L.; “Dynamic Modelling of a Wind
Turbine with Doubly Fed Induction Generator”; IEEE Power Engineering
Society Summer Meeting, 15-19 July 2001, Vancouver, Canada
9.9. Usaola,J., Ledesma,P.; “Dynamic incidence of wind turbines in networks with
high wind penetration”; IEEE Power Engineering Society Summer Meeting,
15-19 July 2001, Vancouver, Canada
9.10. Ledesma,P., Usaola,J.; “Minimum Voltage Protections in Variable Speed Wind
Farms”; 2001 IEEE Porto Power Tech (PPT 2001) Conference, 10-13
September 2001, Porto, Portugal; 0-7803-7139-9
9.11. CIGRE; “CIGRE Technical Brochure on Modeling New Forms of Generation
and Storage”; TF 38.01.10; November 2000; contributions by N.Hatziargyriou,
203
M.Donnelly, S.Papathanassiou, J.A.Pecas Lopes, M.Takasaki, H.Chao,
J.Usaola, R.Lasseter, A.Efthymiadis, K.Karoui, S.Arabi
9.12. Efthymiadis,A.E.; “Report on the Status of Steady State and Transient
Modelling of Double Fed Induction Generators (DFIG)”; British Isles Wind
Technical Panel; November 2001
9.13. Saenz,J.R., Tapia,G., Ostolaza,X., Tapia,A., Criado,R., Berasategui,J.L.;
“Simulation of a wind farm performance under wind speed changes”; 16th
International Conference and Exhibition on Electricity Distribution, CIRED
2001, Amsterdam, The Netherlands; 18 - 21 June 2001
9.14. SP Power Systems Limited, (Dallachy,J.L.); “Transmission Connection
Requirements for Wind Farms – DRAFT”; Issue No.1.8, 04/02/02
9.15. Holdsworth,L., Ekanayake,J., Jenkins,N.; “Steady State and Transient
Behaviour of Induction Generators Connected to Distribution Networks”; IEE-
UMIST Tutorial on Principles and Modelling of Distributed Generators, 4 July
2002, Manchester, UK
9.16. Rostoen,H.O., Undeland,T.M., Gjengedal,T.; “Doubly Fed Induction
Generator in a Wind Turbine”; IEEE Norway Section Workshop on Wind
Power, Oslo, Norway, 17-18 June 2002
9.17. Atkinson,D.J., Lakin,R.A., Jones,R.; “A vector-controlled doubly-fed
induction generator for a variable-speed wind turbine application”;
Transactions of the Institute of Measurement and Control, vol.19, no.1, 1997,
p.2-12
9.18. Naess,B.I., Undeland,T.M., Gjengedal,T.; “Methods for Reduction of Voltage
Unbalance in weak Grids Connected to Wind Plants”; IEEE Norway Section
Workshop on Wind Power, Oslo, Norway, 17-18 June 2002
9.19. Tapia,A., Tapia,G., Ostolaza,J.X., Saenz,J.R., Criado,R., Berasategui,J.L.;
“Reactive Power Control of a Wind Farm made up with Doubly Fed Induction
Generators (I and II)”; 2001 IEEE Porto Power Tech Conference, 10-13
September 2001, Porto, Portugal
9.20. Chedid,R., Mrad,F., Basma,M.; “Intelligent Control of a Class of Wind Energy
Conversion Systems”; IEEE Transactions on Energy Conversion, vol.14, no.4,
December 1999, p.1597-1604
204
10. Conclusions
In the research reported here, it has been shown that there are new challenges in
electricity distribution network planning that mean the conventional approach must
be reviewed and updated to address a number of shortcomings. The expansion of
distributed generation (DG) presents the greatest challenge and must be addressed by
enhancing planners’ capabilities in dealing with it. It has been demonstrated that the
application of methods from other domains can help identify the shortcomings in the
conventional approach and help improve it. This research produced examples using
methods drawn from the domains of engineering design theory, multiple criteria
decision making (MCDM), information management and scenario analysis. In
conjunction with these new methods, the development of dynamic models of wind
farms demonstrated how modelling resources must be expanded to incorporate new
technologies. This is necessary if network planners are to properly understand the
impact on their networks and so design them to accommodate new technologies.
Thus, the thesis of the work has been proven.
A new model of the conventional approach to electricity distribution network

planning provides a means of analysing the conventional approach and identifying
shortcomings in it. The model was developed using a modified knowledge
modelling methodology, a stripped-down version of the KADS methodology, which
itself might be deployed in the analysis of other engineering activities. The
knowledge model provides a new perspective on the distribution planning activity by
identifying the separate tasks that must be fulfilled and the methods used to perform
these tasks.
An up-to-date review of the drivers and new directions in distribution network

planning revealed that DG is the primary issue presenting new challenges. DNOs
will move in new directions both in network architecture and operation, and in
institutional and organisational structures.
205
In constructing the new model of the conventional approach, engineering design
theory was applied and this helped to identify particular shortcomings in the growing
burden of power system simulation and analysis, limited effectiveness in information
and knowledge management, and a failure to incorporate new technologies as
options in network planning and design.
The conventional approach must change to improve the recording of the rationale
behind decisions and provide more effective management of knowledge. This is
necessary to enhance the productivity of planners and may be accomplished by the
introduction of structured decision making techniques, pre-defined justifications that
planners can re-use, and ex-post use of knowledge modelling methods.
The conventional approach must change to improve the management of information

because planners face new challenges in handling new technologies and new
commercial and operational systems. An assessment of the problem identified three
methods that could improve the collection of network planning information as a
structured approach, standard formats and generic devices or data categories. For the
purposes of distribution network planning, DG can be effectively characterised by
collecting information using the new structured approach developed here and
including eight categories: general description; power and energy capabilities;
variable costs; fixed costs; electrical connection; protection; control; and
communications. However, the problem with structured approaches for information
collection, standard formats and generic data is that they restrict the expression of
diversity and nuance. In certain circumstances this is certainly a bad thing but when
presented with having to collect and process vast amounts of information in a short
time, the value of such methods becomes clear. From study in this area it is
concluded that it is inevitable that software vendors – if not the utilities and
manufacturers – will gradually converge on standard formats. An important aspect
of most modern data formats is their extensibility and interoperability so the
traditional problems associated with lock-in to standard formats are mostly overcome
and the risk of adopting any given standard is thereby reduced.
206
The adoption of structured decision making through the use of a framework such as
that outlined in Chapter 6 would offer benefits to electricity distribution network
planning in the incorporation of new technologies and the recording of rationale.
The formalised collection of information facilitates solution-neutral identification of
issues and options, as is recommended in engineering design theory for the
incorporation of new technologies and development of novel ideas. This would mark
a significant change from the conventional approach, which assumes the use of
conventional solutions. The explicit identification of issues, options and the analysis
used to make decisions supports the recording of rationale, which is important in
enhancing planner productivity and in meeting the need for tractable and transparent
decisions.
The calculation of alternative cost-benefit ratios, where the benefit or “cost” need not
be expressed in purely financial terms but as a score or some measure that still allows
objective comparison of different options, offers a new perspective on planning
decisions, including a means of evaluating the relative environmental cost of
different options. This is of value to decision makers in a regulatory and social
environment that places ever more emphasis on protection of the environment.
Regulators may wish to enforce the calculation and presentation of such ratios to
ensure that issues that are sometimes peripheral to planning decisions are properly
taken account of.
The analysis of alternative options for expanding the supply to a remote industrial
facility found that grid reinforcement had the highest overall desirability score. It
also presented the best value in terms of the environmental cost / benefit ratio
(defined using non-financial measures). This confirms that despite the excitement
over DG and its growth, conventional grid reinforcement is still an attractive option.
However, the option with the lowest net present cost was found to be the single gas
turbine and the option with the highest benefit / cost ratio was the multiple gas
turbine option. This shows that DG options are valid alternatives that may be the
best option depending on the specific decision criteria being applied in a given
circumstance.
207
The fair evaluation of DG is subject to the regulatory framework, which may forbid
network operators from owning DG and require complex commercial agreements to
be reached between parties, thereby obscuring the opportunities to find and
implement the least cost or best value solution. This complexity in commercial
agreements presents another new challenge for network operators but it is one that
they should face up to because it will facilitate the exploitation of new resources like
DG.
Scenario analysis was identified as the premier method for managing uncertainty and
risk in network planning and handling DG and its growing impact. While driven by
parties separate from the network operator, DG presents particular challenges
because its expansion, in terms of size, location and timing, is very different from
load growth, which is what the conventional approach is focused upon. Of the
methods described for managing uncertainty and risk, scenario analysis was
highlighted as fitting within an MCDM framework and facilitating the flexible
assessment of a range of issues and options. It also supports risk-based analysis
rather than relying on the law of large numbers necessary to validate probabilistic
results.
An original methodology was devised for the creation of scenarios with DG

connected to LV grids. This methodology is original in its combination of a number
of features: firstly, interpreting high-level national forecasts for DG or renewable
power in terms of the likely impact on LV grids; secondly, by producing quantitative
technology rankings based on a survey of expert opinion; and thirdly, by offering
three strategies for combining the two perspectives to produce scenarios of interest.
This combined approach provides results that are robust from the perspective of
high-level forecasts of DG penetration and from the perspective of how likely each
technology is to be installed in a particular LV grid.
The top-down forecasts used in the study suggest overall levels of DG that are quite
low but in reality some LV grids will contain much more DG than others. The
208
penetration level will depend on the suitability of local conditions to DG deployment,
particularly in terms of the weather, the capability of the existing grid, and the nature
of the load customers. It will also depend on the regulatory framework and the
sophistication of commercial agreements between network operators and owners of
DG. The more complexity required in commercial agreements, the greater the
barrier to DG expansion, so vertically integrated utilities, who may have more
flexibility in defining the commercial framework, are in a better position to exploit
the benefits of DG in their networks more quickly.
The survey of expert opinion identified photovoltaics as the dominant DG

technology in LV grids in both 2010 and 2020. Reciprocating engines and wind
turbines also appeared frequently in the top three ranked technologies in the fifteen
scenarios, with wind turbines coming first in a number of scenarios based on rural
areas in northern Europe. In 2020, the pattern remains much the same but with new
technologies like micro gas turbines, biogas engines and fuel cells featuring more
prominently.
The expansion of renewables and DG requires new models to support planning and
design of the network. The development of dynamic models of wind farms is one
example of the extensions required in power system simulation to manage the
changes in the electricity system. Models such as these are essential to system
planners as they endeavour to meet the challenges facing them.
A comprehensive analysis was performed of the modelling challenges posed by

modern wind farms and the need for dynamic models. Models were developed and
shown to work effectively in PSS/E, a widely used power system simulation
package. It was acknowledged that the models require validation to be applied more
fully and this was identified as one of the areas for further work. However, the lack
of publicly available information on modern wind turbines, particularly their control
systems, was a significant barrier to this research and remains so now, hindering the
development of models by others including network operators.
209
The study of aggregate wind farm models suggested that for normal power system
simulation, a single machine could be used to represent a wind farm with multiple
turbines. This would greatly simplify the modelling requirements and is in line with
the perspective of system operators, who will wish to view wind farms as single
entities. The results start to diverge as the wind farm electrical network becomes
large and its high impedance reduces the coupling between the network and
individual machines. This indicates that wind farms spread over very large areas
may not be accurately represented with an aggregate model, although the studies do
not support firm conclusions on the exact size of networks affected. This could be an
area for further work.
The study of the effect of a high penetration of wind farms on distribution networks
revealed that the wind farms rode through faults when at medium and low power
output levels but became unstable when at full power output. Thus, the results
indicated that it was possible for large amounts of wind power to trip off under fault
conditions. This illustrates why fault ride-through capabilities on wind turbines are
so important to system operators wishing to maintain system stability. However, the
probability of instability problems caused by wind power dropping off during a
disturbance is low due to the low probability that the wind farms will actually be
producing close to their maximum output. The studies of varying input power
highlighted the resultant variations in network power flows and voltages and
suggested that unfortunate timing of input changes might cause constructive
interference of oscillations and lead to instability.
It is understood that modern wind turbines have more advanced control systems and
have addressed some of the potential problems highlighted in this report. However,
while the control systems and detailed information on performance remain a closely
guarded secret, it is difficult for researchers and network operators without access to
this information to perform realistic studies. When information on the actual control
systems is made available, they can be modelled within the PSS/E model framework
established in this research.
210
It was concluded that while new models must be developed to support analysis of
new technologies in future distribution networks, this presents a very considerable
challenge to DNOs. It requires an understanding of the physical characteristics of
the device being modelled, which is obviously difficult for new technologies.
Information is likely to be lacking and there are very many design decisions that can
influence the model in different ways. It also requires a deep understanding of the
simulation software being used. The expertise in modelling and simulation will vary
from DNO to DNO but some of them are likely to require external assistance. This
provides another example of the organisational shortcomings highlighted in Chapter
4.
In conclusion, the changes taking place in electricity distribution network planning

are dramatic and the challenges are great. The conventional approach must be
updated and this can be supported by the application of methods from other domains
and the development of new tools such as those described here. Engineering design
theory, decision support and scoring methods for quantitative assessment,
information management, particularly for DG, a methodology for robust scenario
preparation, and new dynamic models of wind farms have all been investigated and
offer a valuable contribution to the development of electricity distribution network
planning and the changes necessary to meet the new challenges of the 21st century.
211
Appendix A. DG Penetration Survey Results
This appendix details the results of the survey used in the bottom-up assessment of
DG penetration described in section 8.2.3. For each of the LV grid scenarios,
respondents to the questionnaire ranked the DG technologies in terms of which are
most likely to be installed in the scenario conditions by 2010 and by 2020. A
ranking of “1” indicates the technology is most likely to be installed in the associated
grid conditions. Rankings of “2”, “3”, “4”, etc. indicate decreasing likelihood of
technologies being installed. The rankings from different respondents were
combined to calculate average rankings or scores for 2010 and 2020 for each
scenario. These scores, which give an indication of which technologies are
considered most suited to each grid scenario and are most likely to be installed, are
shown in the tables below. Respondents also provided comments to support their
rankings. The comments are summarised in the results shown below.
212
A.1. Scenario 1. Residential Ring, Germany
Table A.1 – Questionnaire results for scenario 1
DG Technology Score Score Comments

2010 2020
Reciprocating engines 2.3 3.7 Governments support CHP engines because of
potential CO2 reductions.
Diesel engines may be used to provide backup power
but other technologies are expected to take over by
2020.
Micro gas turbines 6.7 5.3 Micro gas turbines are more likely in industrial
applications but may be used in domestic CHP by
2020
Fuel cells 5.3 3.7 Governments support fuel cells because of the
Fuel cells are still in the early stages of development
but may find applications in domestic combined heat
and power, especially by 2020.
Photovoltaics 1.3 1.0 PV is the most advanced DG technology at the
moment and in Germany benefits from the highest
feed-in tariff.
Wind turbines 7.3 7.0 Wind turbines are more likely to be connected at MV
or HV and are unlikely in an urban area.
Micro hydro 7.7 7.7 Micro hydro is unlikely in an urban area.
Biogas engines 7.3 7.3 Biogas is unlikely in an urban area with residential
loads.
Other DG 5.0 4.7 Domestic CHP systems with Stirling Engines are now
available and may be suited to this scenario.
If available, geothermal might be used.
No distributed generation 6.3 6.7
Photovoltaics are clearly considered the most likely technology in this scenario.
Reciprocating engines, in CHP or backup power applications, are in second place in
2010 but fuel cells are predicted to catch up by 2020. Domestic CHP systems with
Stirling Engines are considered the next most likely technology to be installed in this
scenario.
213
A.2. Scenario 2. Commercial Mesh, Germany

2010 2020
Reciprocating engines 2.0 3.7 Governments support CHP engines because of
In these circumstances, especially with commercial
loads, backup power may be provided with diesel
engines most likely in 2010 but other technologies
expected to take over by 2020.
Micro gas turbines 6.0 4.7 Micro gas turbines are more likely in industrial
applications although they may also be used in retail or
office applications, especially by 2020.
Fuel cells 5.3 4.3 Governments support fuel cells because of the potential
CO2 reductions.
Fuel cells are still in the early stages of development
but may find applications in small-scale combined heat
and power, especially by 2020.
Photovoltaics 1.3 1.0 PV is the most advanced DG technology at the moment
and in Germany benefits from the highest feed-in tariff.
Wind turbines 7.0 6.7 Wind turbines are more likely to be connected at MV
or HV and are unlikely in an urban area.
Biogas engines 7.3 7.3 Biogas is unlikely in an urban area.
Other DG 5.3 4.3 CHP systems with Stirling Engines are now available
and may be suited to this scenario.
Reciprocating engines, in CHP or backup power applications, are in second place.
Small-scale CHP systems with Stirling Engines or fuel cells are considered the next
most likely technology to be installed in this scenario.
214
A.3. Scenario 3. Mixed Radial, Germany

2010 2020
Reciprocating engines 3.0 5.0 In these circumstances, especially with commercial
loads, backup power may be provided, with diesel
Micro gas turbines 5.5 3.0 Micro gas turbines may find applications in small-scale
combined heat and power and in providing backup
supplies to commercial loads, especially by 2020.
Fuel cells 5.5 4.0 Fuel cells may find applications in small-scale
Photovoltaics 1.5 1.0 This scenario does not have the best conditions for PV
but it is still likely to be the most common form of DG.
Wind turbines 6.5 6.0 Wind turbines are unlikely in an urban area.
Other DG 6.5 5.0 Micro CHP systems with Stirling Engines are now
Reciprocating engines, in CHP or backup power applications, are in second place in
2010 but fall behind micro gas turbines and fuel cells in 2020.
215
A.4. Scenario 4. Urban Meshed, UK

2010 2020
Reciprocating engines 3.0 4.5 In these circumstances, with commercial loads, backup
power may be provided, with diesel engines most likely
in 2010 but other technologies expected to take over by
2020.
Photovoltaics 3.5 2.5 The lower solar radiation and lower temperatures in the
UK makes small-scale CHP more attractive than PV
but PV is still likely to be installed in significant
numbers.
Wind turbines 5.5 5.0 Wind turbines are unlikely in an urban area unless there
are significant developments in small-scale,
unobtrusive designs.
available and may be suited to this scenario where
central heating is likely to be widely installed.
No technologies emerge with very high scores in this scenario. Reciprocating

engines, probably to provide backup power, and photovoltaics, despite the less than
perfect conditions, come out highest in 2010. By 2020 it is predicted that micro gas
turbines will have overtaken reciprocating engines.
216
A.5. Scenario 5. Rural, Poland

2010 2020
Reciprocating engines 5.7 5.7 In a rural area with mostly residential loads, it is
unlikely that there will be many applications for
reciprocating engines although they may be used as
backup supplies.
Micro gas turbines 6.0 5.0 In a rural area with mostly residential loads, it is
unlikely that there will be many applications for micro
gas turbines.
Fuel cells 7.0 6.3 In a rural area with mostly residential loads, it is
unlikely that there will be many applications for fuel
cells.
Photovoltaics 3.0 2.7 With relatively high solar irradiation, PV could be
attractive in this scenario if the necessary financial
incentives are there.
Solar energy may also be used for heating.
Wind turbines 3.0 2.0 Wind speed is sufficient to make wind turbines viable
and they may be accepted in the rural area.
Micro hydro 3.7 4.0 If the resources are available, micro hydro may feature
although the potential is limited.
Biogas engines 4.3 4.0 Biogas fuel may be available from farms or sewage
refineries but the residential loads in this scenario make
biogas fuelled DG less likely.
Other DG 5.0 4.7 Small CHP fed by agricultural waste, wood crops and
forest residues.
Geothermal, if available
The renewable power sources – photovoltaics, wind turbines and micro hydro – score
most highly in this scenario although none of them are favoured particularly strongly.
Biogas engines are the next most likely technology to be installed.
217
A.6. Scenario 6. Urban Link, Poland

2010 2020
Reciprocating engines 2.3 4.0 In these circumstances, especially with commercial
loads, backup power may be provided with diesel
Photovoltaics 2.7 2.0 Solar irradiation is not as high as other areas but PV is
still appropriate, although the expense means some
financial support must be provided.
Wind turbines 6.0 5.7 Wind turbines are unlikely in an urban area.
Biogas engines 5.3 5.3 Biogas might be available from refuse dumps and
sewage refineries but is unlikely in an urban area.
Other DG 5.0 3.7 Small CHP may be fed by municipal waste
Micro CHP systems with Stirling Engines are now
Geothermal may be used, if available
Reciprocating engines and photovoltaics emerge as the highest scorers in this

scenario for 2010 and PV moves further ahead in 2020. Micro gas turbines move
into second place by 2020, followed in third place by other DG – probably small
CHP fed by municipal waste or with a Stirling engine.
218
A.7. Scenario 7. Urban Radial, France

2010 2020
Reciprocating engines 3.0 5.0 In these circumstances, backup power may be provided
with diesel engines most likely in 2010 but other
technologies expected to take over by 2020.
Micro gas turbines 5.5 3.5 Micro gas turbines may find applications in backup
power, especially by 2020.
Fuel cells 5.5 4.0 Fuel cells may find applications in backup power,
especially by 2020.
Photovoltaics 1.5 1.0 This scenario is most suited to PV.
Wind turbines 7.0 7.0 Wind turbines are unlikely in an urban area with a low
wind speed.
Geothermal may be used, if available.
Photovoltaics are the clear winner in this scenario because of the high solar
irradiation, far ahead of other technologies in both 2010 and 2020. Reciprocating
engines are in second place in 2010, most likely in backup power applications, but
are expected to fall behind micro gas turbines and fuel cells by 2020.
219
A.8. Scenario 8. Rural, Italy

2010 2020
Reciprocating engines 3.7 4.0 In these circumstances, backup power may be provided
to farms. Diesel engines are most likely because they
are good at partial load and are reliable.
Micro gas turbines 6.7 6.0 Micro gas turbines may find applications in backup
power if there is a gas network available but they are
better used to supply a continuous CHP load.
Also, their reliability has still to be demonstrated and
costs are too high so new technology development is
required.
Fuel cells 7.7 6.3 Currently, only phosphoric acid fuel cells are
commercial. Other types are too expensive.
Fuel cells may find applications in backup power if
there is a fuel delivery network.
Photovoltaics 1.7 1.3 PV is the technology most suited to this scenario but its
widespread adoption will depend on government
finance programmes.
Wind turbines 5.0 4.3 Although farms probably provide the best location for
small wind turbines, the wind speed is too low and
small turbines are too expensive.
Micro hydro 5.0 5.0 If the resources are available, micro hydro may be
installed but there are limited sites where the
geography makes it economic.
Biogas engines 4.0 4.0 If biogas is available it may be possible but a sparsely
populated area may not have the electrical network
infrastructure to support it.
Other DG 6.7 6.0 Biomass-fired CHP with external combustion engines
(Stirling cycle or organic Rankine cycle) but these need
new technology development
If resources are available, geothermal may be possible
but it would depend on the strength of the electrical
network.
Photovoltaics again emerge as the clear winner in this scenario. The second highest
ranking in 2010 goes to No DG with reciprocating engines for backup power in third
place. By 2020, it is thought that reciprocating engines and biogas engines will be as
likely as No DG.
220
A.9. Scenario 9. Urban Radial, Italy

2010 2020
Reciprocating engines 2.3 4.3 Some of the commercial or industrial loads may use
diesel or gas engines as backup power supplies because
they are good at partial load and are reliable.
Micro gas turbines 4.7 2.7 Micro gas turbines require a gas supply network, their
reliability has still to be demonstrated and their cost is
currently too great. New technology development is
required.
They are better used to supply a continuous CHP load
but may surpass diesels as the primary choice for
backup supplies.
Fuel cells 6.0 4.0 Currently, only phosphoric acid fuel cells are
commercial. Other types are too expensive.
Fuel cells may find applications in backup power if
there is a fuel delivery network.
Photovoltaics 1.7 1.3 With high solar irradiation and suitable loads, PV
should be the most common form of DG but their high
ranking here assumes government finance programmes.
Wind turbines 8.0 8.0 Low average wind speed in an urban area suggests no
wind turbines.
Small turbines are too expensive.
Micro hydro 6.3 6.3 There are limited sites where the geography makes
micro hydro economic but it is unlikely in an urban
area.
Biogas engines 7.0 7.0 Biogas is limited availability of the fuel but is unlikely
in an urban area where other options are available.
available and may be suited to this scenario. Biomass-
fired CHP with external combustion engines (Stirling
cycle or organic Rankine cycle) might be used but
require new technology development.
Geothermal might be used, if available
Photovoltaics are ranked most highly in this scenario for both 2010 and 2020.
Reciprocating engines come second in 2010 but are replaced by micro gas turbines in
2020. In both 2010 and 2020, the third ranked option is No DG.
221
A.10. Scenario 10. Urban Link, Greece

2010 2020
Reciprocating engines 3.5 5.0 Some of the commercial or industrial loads may use
diesel engines as backup power supplies.
Micro gas turbines 5.5 3.5 Micro gas turbines may surpass diesels as the primary
choice for backup power and may find additional
applications in CHP.
Fuel cells 5.5 4.0 Fuel cells may be used as backup and for CHP but are
likely to remain more expensive than the alternatives
Photovoltaics 1.5 1.0 With high solar irradiation and suitable loads, PV
should be the most common form of DG.
Wind turbines 7.5 7.5 Low average wind speed in an urban area suggests no
wind turbines.
Biogas engines 6.0 6.0 Biogas is unlikely in an urban area where other options
are available.
Other DG 8.0 7.5 If resources are available, geothermal may be possible.
Photovoltaics are ranked highest in 2010 and 2020. In 2010, reciprocating engines
are thought next most likely to be installed. Other technologies all come behind No
DG, which is ranked third. However, by 2020 micro gas turbines are ranked second
and fuel cells third.
222
A.11. Scenario 11. Rural South Coast, Spain

2010 2020
Reciprocating engines 5.0 5.7 Some farms may use diesel engines as backup supplies.
Micro gas turbines 6.0 5.7 It is unlikely that micro gas turbines will find an
application in this scenario.
Fuel cells 7.7 6.0 It is unlikely that fuel cells will find an application in
this scenario.
Photovoltaics 1.0 1.0 High solar irradiation makes PV the most promising
DG technology in this scenario.
Wind turbines 2.0 2.0 A relatively high wind speed in a rural area with mostly
farm loads makes wind turbines an attractive DG
option.
Micro hydro 5.3 5.7 If the resources are available, micro hydro might be
exploited but PV and wind seem like the more likely
options.
Biogas engines 4.0 4.0 If the biogas fuel is available from the farms then this
might be an option but PV and wind seem more
promising.
Other DG 8.0 7.7 If geothermal resources are available they might be
exploited but PV and wind are likely to be used instead.
Storage technologies like flywheels may be used.
Photovoltaics are ranked highest in this scenario but are closely followed by wind
turbines. The next highest ranked technology is biogas engines. These rankings
apply in both 2010 and 2020.
223
A.12. Scenario 12. Rural Link, Netherlands

2010 2020
Reciprocating engines 4.5 4.5 Some farms may use reciprocating engines for backup
power.
Micro gas turbines 7.0 6.5 Micro gas turbines are unlikely unless there is a
specific application for them.
Fuel cells 7.0 6.0 Fuel cells are unlikely unless there is a specific
application but even then other options will probably
be favoured.
Photovoltaics 3.5 2.5 Solar irradiation is not particularly high but PV may
still find a place in the DG mix.
Wind turbines 1.5 1.5 The high average wind speed makes wind turbines
attractive and the farms should provide a place to host
them.
Micro hydro 4.5 4.5 If the resource is available, micro hydro could be
exploited.
Biogas engines 4.0 4.0 Biogas engines may be utilised if the fuel was readily
available.
Other DG 8.0 7.5 If the resource is available, geothermal could be
exploited.
Wind turbines are the highest ranked technology in this scenario, in both 2010 and
2020. Photovoltaics come second and are closer to wind in 2020 than in 2010.
Biogas engines are ranked third most likely to be installed in both time horizons.
224
A.13. Scenario 13. Rural Ring, Belgium

2010 2020
power.
be favoured.
Photovoltaics 3.5 2.5 Solar irradiation is not particularly high but PV may
still find a place in the DG mix.
Wind turbines 1.5 1.5 The relatively high average wind speed makes wind
turbines attractive and the farms should provide a place
to host them.
exploited.
available.
exploited.
No distributed generation 5.0 6.5 The good wind resource in this scenario means there
should be some DG by 2010 and more by 2020.
This scenario produces exactly the same predictions as scenario 12. Wind turbines
are the highest ranked technology, in both 2010 and 2020. Photovoltaics come
second and are closer to wind in 2020 than in 2010. Biogas engines are ranked third
most likely to be installed in both time horizons.
225
A.14. Scenario 14. Urban Ring, Denmark

2010 2020
Reciprocating engines 3.0 4.5 Reciprocating engines may be installed to provide
backup power but are likely to be superseded.
Micro gas turbines 4.0 2.0 Micro gas turbines could be used as providers of
backup power and CHP to the commercial and
industrial loads.
Fuel cells 5.5 4.5 Fuel cells could be used as providers of backup power
and CHP to the commercial and industrial loads but are
less likely than micro gas turbines.
Photovoltaics 4.0 2.5 The relatively low solar irradiation and availability of
other options reduce the attractiveness of PV but it is
still likely to be installed in small numbers.
Wind turbines 5.5 5.0 The wind resource is there but it is unlikely that the
commercial and industrial customers will favour wind
power.
Micro hydro 7.0 7.0 A micro hydro resource is unlikely to be available in
this urban area.
Biogas engines 7.0 7.0 Biogas engines running on waste or other fuels
available in the urban area may be installed in this area
as long as the commercial and industrial customers are
willing.
Other DG 6.0 5.0 Micro CHP with Stirling Engine is available now in
small sizes and may provide a CHP solution for some
of these commercial and small industrial loads.
Geothermal might be used, if available
These rankings are based on the general scenario description but it is accepted that
Denmark already has large amounts of what may be considered DG, with many wind
farms and district CHP schemes. In 2010, reciprocating engines are ranked highest
followed by micro gas turbines and photovoltaics. In 2020, micro gas turbines move
into first place, photovoltaics into second, and reciprocating engines drop to be level
with fuel cells.
226
A.15. Scenario 15. Rural Ring, Austria

2010 2020
power.
be favoured.
Photovoltaics 2.0 1.5 Solar irradiation is relatively high and PV is likely to
find a place in the DG mix.
Wind turbines 5.5 5.0 The low average wind speed makes wind turbines
unattractive.
exploited.
available.
exploited.
Photovoltaics are ranked most highly in this scenario, micro hydro comes second and
biogas engines third. These rankings apply to both 2010 and 2020.
227
Appendix B. Aggregate Models of a Wind Farm
B.1. Methodology for Study of Aggregate Models of a Wind Farm
B.1.1. Test Networks
The models were tested on a simple four-bus network, which was extended to
represent a wind farm network in some studies. The studies were split into three
categories:
• Category A studies used a single 20 MVA generator connected at bus 103 as
shown in Figure B.1.
• Category B studies used ten 2 MVA generators connected at bus 103 as shown in
Figure B.2.
• Category C studies included ten additional buses and lines to represent the
electrical network of a wind farm with one 2 MVA generator at each of the extra
buses, as shown in Figure B.3.
As indicated in the figures, the base voltage at all buses was 33kV and all of the main
branches had resistance 0.01pu and reactance 0.05pu. In most category C studies,
the extra lines representing the wind farm network had resistance 0.001pu and
reactance 0.005pu but this was varied in one study, as explained below. System base
was 100 MVA. The load at bus 104 was 50 MW, 20 MVAr in all studies but the
generator outputs were varied in the different studies as explained below.
228
Figure B.1 – Test network for category A studies
Figure B.2 – Test network for category B studies
229
Figure B.3 – Test network for category C studies
B.1.2. Test Models
A second-order electrical machine model with a first-order turbine shaft model

represented the doubly fed induction generators. The machine rotor voltages were
controlled by simple PI controllers with target values of voltage and rotor speed set
equal to their initial values. Power input to the generators was held constant
throughout the simulations. The doubly fed induction generators and rotor voltage
controllers were modelled with the parameters as shown in Table B.1 and Table B.2
respectively.
Table B.1 – Doubly fed induction generator parameters
Parameter Value
LM 4.0 pu
LS 0.1 pu
LR 0.1 pu
RS 0.005 pu
RR 0.005 pu
HROTOR 3.0 sec
230
HGEN 0.5 sec
DAMP 0.01
Table B.2 – Rotor voltage controller parameters
Parameter Value
KDP 1
KDI 0.1
KQP 1
KQI 1
VMRLIM 0.01 pu
MSPEED 0
MVOLTS 0
IMRLIM 100 pu
In the studies, the machines were given different base MVA values depending on
whether there were multiple machines or a single machine. In studies with ten
machines, each machine was assigned a base of 2MVA. In studies with a single
machine, the machine was assigned a base of 20MVA. However, the same model
parameter values were used in all studies.
In all studies, the generator at bus 101 was modelled using the GENCLS model from
the PSS/E library with zero inertia and zero damping to emulate the response of an
infinite bus.
B.1.3. Test Procedure
The simulations were run with a time step of 10ms. All studies involved the same
basic procedure:
• The simulation was run up to 1 second.
• A fault was applied on circuit 2 between buses 102 and 104.
• The simulation was run up to 1.2 seconds.
• Circuit 2 between buses 102 and 104 was tripped.
• The simulation was run up to 10 seconds.
A wide variety of simulation variables were recorded during simulation but the post-
simulation analysis focused on the real and reactive power flowing from bus 103 to
231
102 and the bus voltage and bus angle at bus 103. These variables were thought to
give the most effective means of comparing the different wind farm representations
from the perspective of the network.
Four sets of three studies were performed; each of the three categories (A, B and C)
being studied under four different conditions.
B.2. Results for Study of Aggregate Models of a Wind Farm
B.2.1. Study Set One
In study set one, the total output from the wind farm representation at bus 103 was
set to 15 MW and 0 MVAr. In categories B and C the real and reactive power was
divided evenly between the ten machines.
Figure B.4 shows that the three real power responses are very similar. A and B are
exactly the same as one another. Response C shows slight differences in the time
steps immediately following fault clearance.
Figure B.4 – Real power in branch 102-103 for the three categories of study set one
a1 b1 c1
30
20
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-10
-20
-30
-40
232
Figure B.5 shows that the three reactive power responses are very similar. There are
slight differences in the short-term transient response when the fault is cleared. All
three categories show an oscillation as the controllers try to restore reactive power to
its initial value.
Figure B.5 – Reactive power in branch 102-103 for the three categories of study set one
a1 b1 c1
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-2
-4
Figure B.6 shows that the three bus voltage responses are very similar. Response C
is slightly different to responses A and B at the point where the fault is cleared and
voltage is restored.
233
Figure B.6 – Bus voltage at bus 103 for the three categories of study set one
a1 b1 c1
1.2
0.8
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5
Figure B.7 shows that the three bus angle responses are very similar. However, all
three categories display an undesirable shift in bus voltage angle.
Figure B.7 – Bus angle at bus 103 for the three categories of study set one
a1 b1 c1
400
350
300
250
200
150
100
50
0
0 0.5 1 1.5 2 2.5
-50
234
B.2.2. Study Set Two
In study set two, the total output from the wind farm representation at bus 103 was
set to 10 MW and -5 MVAr. In categories B and C the real and reactive power was
divided evenly between the ten machines.
Figure B.8 shows that the three real power responses are very similar.
Figure B.8 – Real power in branch 102-103 for the three categories of study set two
a2 b2 c2
20
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-10
-20
-30
-40
-50
Figure B.9 shows that the three reactive power responses are very similar. There are
slight differences in the short-term transient response when the fault is cleared.
235
Figure B.9 – Reactive power in branch 102-103 for the three categories of study set two
a2 b2 c2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-2
-4
-6
-8
Figure B.10 – Bus voltage at bus 103 for the three categories of study set two
a2 b2 c2
1.2
0.8
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5
236
three categories display an undesirable shift in bus voltage angle.
Figure B.11 – Bus angle at bus 103 for the three categories of study set two
a2 b2 c2
400
350
300
250
200
150
100
50
0
0 0.5 1 1.5 2 2.5
-50
B.2.3. Study Set Three
In study set three, the total output from the wind farm representation at bus 103 was
set to 15 MW and 5 MVAr. In categories B and C the real and reactive power was
divided evenly between the ten machines. In this study, the line impedances within
the wind farm network in category C were made the same as in the main network, i.e.
R=0.01pu and X=0.05pu.
Figure B.12 shows that the three real power responses are very similar. A and B are
exactly the same as one another. Response C is slightly different in the period
immediately following fault clearance.
237
Figure B.12 – Real power in branch 102-103 for the three categories of study set three
a3 b3 c3
70
60
50
40
30
20
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-10
Figure B.13 shows that the three reactive power responses are very similar. There
are slight differences in the short-term transient response when the fault is cleared.
Figure B.13 – Reactive power in branch 102-103 for the three categories of study set three
a3 b3 c3
20
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-10
-20
-30
-40
-50
-60
238
Figure B.14 – Bus voltage at bus 103 for the three categories of study set three
a3 b3 c3
1.2
0.8
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5
Figure B.15 shows that bus angle responses A and B behave quite differently to
response C. In A and B, bus angle drops and settles at a new value. In C, bus angle
starts rising but is restored to its initial value when the fault is cleared. This is due to
the greater line impedances in the wind farm network: R=0.01pu and X=0.05pu
rather than R=0.001pu and X=0.005pu as in the other studies. This higher
impedance alters the dynamic relationship between the network and the wind turbine
models. The result is that the turbines are more stable and the bus angle at bus 103
rises then returns to its original value in response C but changes rapidly and shifts to
a new steady state value in responses A and B.
239
Figure B.15 – Bus angle at bus 103 for the three categories of study set three
a3 b3 c3
200
100
0
0 0.5 1 1.5 2 2.5
-100
-200
-300
-400
B.2.4. Study Set Four
In study set four, the total output from the wind farm representation at bus 103 was
set to 10 MW and -5 MVAr as in study set two. However, in categories B and C the
real and reactive power was divided between the ten machines according to the
pattern shown in Table B.3.
Table B.3 – Division of power between machines in test categories B and C
Category B Category C Real Power Reactive Power

Machine Machine (MW) (MVAr)
Bus 103 Machine 1 Bus 110 Machine 1 0.5 0.0
Bus 103 Machine 2 Bus 111 Machine 1 0.6 -0.1
Total 10.0 -5.0
240
Figure B.16 shows that the three real power responses are very similar.
Figure B.16 – Real power in branch 102-103 for the three categories of study set four
a4 b4 c4
20
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-10
-20
-30
-40
-50
Figure B.17 shows that the three reactive power responses are very similar. There
are slight differences in the short-term transient response when the fault is cleared.
Figure B.17 – Reactive power in branch 102-103 for the three categories of study set four
a4 b4 c4
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
-2
-4
-6
-8
241
Figure B.18 – Bus voltage at bus 103 for the three categories of study set four
a4 b4 c4
1.2
0.8
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5
242
three categories display an undesirable shift in bus voltage.
Figure B.19 – Bus angle at bus 103 for the three categories of study set four
a4 b4 c4
400
350
300
250
200
150
100
50
0
0 0.5 1 1.5 2 2.5
-50
243
Appendix C. The Effects of a High Penetration of Wind Farms
C.1. Methodology for Study of the Effects of a High Penetration of Wind Farms
C.1.1. Test Network
The tests were conducted using a modified version of the IEEE 30-bus test network.
The network has buses at 132kV and 33kV; there are 21 loads and four conventional
generators. Fifteen wind farms were added, spread across the network. Some
capacitors were also added to provide power factor correction. A diagram of the
network is shown in Figure C.1. Information on network loads is given below.
Figure C.1 – Diagram of 30-bus network used in wind farm studies
244
C.1.2. Test Models
C.1.2.1. Wind Farms
The wind farms were each represented by a model of a doubly fed induction
generator: a second-order electrical machine model with a first-order turbine shaft
model. The equivalence of single and multiple turbine models was demonstrated in
other studies. The machine rotor voltages were controlled by simple PI controllers
with target values of voltage and rotor speed set equal to their initial values.
The tests were conducted using the versions of the different model components
described in Table C.1. Four separate model components allow the different
components to be updated on their own. The four models all sit within the PSS/E
modelling environment and communicate with each other using PSS/E variables.
Table C.1 – Models used in the study of high penetration of wind farms
Model Description
DFIGBB-16 Second-order model of the doubly fed induction generator and
wind turbine shaft
DFIGVC-15 Rotor voltage controllers for the DFIG
DFIGIN-03 Power input controller for the DFIG
DFIGPR-01 Protection system for the DFIG
The doubly fed induction generators and rotor voltage controllers were modelled
with parameters shown in Table C.2 and Table C.3 respectively.
Table C.2 – Doubly fed induction generator parameters
Parameter Value
LM 4.0 pu
LS 0.1 pu
LR 0.1 pu
RS 0.005 pu
RR 0.005 pu
HROTOR 3.0 sec
HGEN 0.5 sec
DAMP 0.01
245
Table C.3 – Rotor voltage controller parameters
Parameter Value
KDP 1
KDI 0.1
KQP 1
KQI 1
VMRLIM 0.01 pu
MSPEED 0
MVOLTS 0
IMRLIM 100 pu
For the fault disturbance tests, no power input model was used so power input to the
wind farms was held constant through each simulation. For the power variation tests,
a version of the DFIGIN power input model was used with a pattern of power input
variations as described in Table C.4. This pattern of input power variations was
adjusted for a supplementary set of studies, as explained below.
Table C.4 – Pattern of power input variations
Time Period Input Power Variations

0 to 5 seconds Power held at its initial value
5 to 10 seconds Power ramped up to 100% of MVA base
10 to 20 seconds Power held at 100% of MVA base
20 to 30 seconds Power ramped down to zero
30 to 40 seconds Power held at zero
40 to 45 seconds Power ramped back up to its initial value
After 45 seconds Power held at its value
A protection system model was introduced to model the effect of wind farms tripping
under disturbance conditions. This first version of the protection model only
monitors one condition. If the rotor current magnitude rises too high then the
machine is tripped. After some preliminary studies, the rotor current protection was
set to trip if the magnitude reached 1.5pu. This level was chosen because it results in
some of the wind farms tripping and some of them riding through the line fault
disturbance for most of the wind farm scenarios.
246
The rotor current protection setting was adjusted to 50pu for a supplementary set of
studies to examine the effect of all wind farms staying connected, as explained
below.
C.1.2.2. Conventional Generators
The conventional generator at bus 2, the swing bus, was modelled using the classic
generator model from the PSS/E library (GENCLS). Inertia and damping values
were set to 10 to simulate large grid inertia with considerable damping. The
parameters for the other conventional generators were extracted from typical and
recommended values provided in the PSS/E user manuals. The conventional
generator at bus 1 was modelled using the salient pole generator model from the
PSS/E library (GENSAL), with parameters as shown in Table C.5.
Table C.5 – Salient pole generator parameters
GENSAL Parameter Value

T'do 8
T"do 0.05
T"qo 0.12
H 1.1
D 0
Xd 1.8
Xq 1.35
X'd 0.6
X"d 0.2
Xl 0.02
S(1.0) 0.03
S(1.2) 0.25
The conventional generators at buses 5 and 8 were modelled using the round rotor
generator model from the PSS/E library (GENROU), with parameters as shown in
Table C.6.
247
Table C.6 – Round rotor generator parameters
GENROU Parameter Value

T'do 2
T"do 0.05
T'qo 1
T"qo 0.05
H 5
D 0
Xd 1.4
Xq 1.35
X'd 0.3
X'q 0.6
X"d 0.2
Xl 0.1
S(1.0) 0.03
S(1.2) 0.4
The conventional generators at buses 1, 5 and 8 were modelled with the same
governor and excitation systems. The steam turbine governor, TGOV1, model and
simple excitation system, SEXS, model were used at all three machines with
parameters as shown in Table C.7 in Table C.8.
Table C.7 – Turbine governor parameters
TGOV1 Parameter Value

R 0.05
T1 0.05
VMAX 1.1
VMIN 0.1
T2 2
T3 6
Dt 0
Table C.8 – Simple excitation system parameters
SEXS Parameter Value

TA/TB 0.1
TB 10
K 100
TE 0.01
EMIN 0
EMAX 4
248
C.1.3. Test Scenarios
Tests were conducted under three different load conditions and six different wind
farm conditions. The load and wind farm conditions were combined to produce
eighteen test scenarios for each of the test procedures.
C.1.3.1. Load Conditions
Load conditions were varied on the network to produce three scenarios: low (“lo”),
medium (“md”) and high (“hi”). In the low load scenario, all loads were assigned
values 50% of the values on the medium load scenario, for both real and reactive
power. In the high load scenario, all loads were assigned values 150% of the
medium load scenario. The load values for the three load scenarios are shown in
Table C.9.
Table C.9 – Load values for the three scenarios
Bus Low Load Scenario Medium Load Scenario High Load Scenario
P (MW) Q (MVAr) P (MW) Q (MVAr) P (MW) Q (MVAr)
2 10.85 6.35 21.7 12.7 32.55 19.05
3 1.2 0.6 2.4 1.2 3.6 1.8
4 3.8 0.8 7.6 1.6 11.4 2.4
5 47.1 9.5 94.2 19 141.3 28.5
7 11.4 5.45 22.8 10.9 34.2 16.35
8 15 15 30 30 45 45
9 2.9 1 5.8 2 8.7 3
10 5.6 3.75 11.2 7.5 16.8 11.25
11 3.1 0.8 6.2 1.6 9.3 2.4
12 4.1 1.25 8.2 2.5 12.3 3.75
13 1.75 0.9 3.5 1.8 5.25 2.7
14 4.5 2.9 9 5.8 13.5 8.7
15 1.6 0.45 3.2 0.9 4.8 1.35
16 4.75 1.7 9.5 3.4 14.25 5.1
17 1.1 0.35 2.2 0.7 3.3 1.05
18 8.75 5.6 17.5 11.2 26.25 16.8
20 1.6 0.8 3.2 1.6 4.8 2.4
21 4.35 3.35 8.7 6.7 13.05 10.05
23 1.75 1.15 3.5 2.3 5.25 3.45
26 1.2 0.45 2.4 0.9 3.6 1.35
27 5.3 0.95 10.6 1.9 15.9 2.85
Totals 141.7 63.1 283.4 126.2 425.1 189.3
249
C.1.3.2. Wind Farm Conditions
The wind farm conditions were varied by changing the status and power output of
each wind farm. All fifteen wind farms were set to the same condition. In all
scenarios, the wind farms were specified to operate at unity power factor with
reactive power of zero. The wind farm scenarios were as described in Table C.10.
Table C.10 – Wind farm scenarios
Wind Farm Wind Farm Output of each Total Output

Scenario Status Wind Farm (MW) from 17 Wind
Farms (MW)
Scenario X Off 0 0
Scenario 0 On 0 0
Scenario A On 5 85
Scenario B On 10 170
Scenario C On 15 255
Scenario D On 20 340
C.1.3.3. Output of Conventional Generators
The output of the conventional generators at buses 1,5 and 8 were set to 100MW in
each simulation. The swing bus made up the balance depending on the loads, wind
farm outputs and network losses. Reactive power varied at each generator depending
on voltage control requirements in each scenario. Reactive power limits for the four
conventional generators were set to ±240MVAr.
C.1.4. Test Procedures
For each of the eighteen scenarios (combining three load conditions with six
generator conditions), two separate tests were performed: a line fault disturbance test
and a varying power input test.
C.1.4.1. Line Fault Disturbance Test
The line fault disturbance test was designed to study the effect on network voltages
of the response of a high penetration of wind farms to a significant fault disturbance.
The line fault disturbance test involved:
250
• running the simulation to one second;
• applying a line fault on line 1-2;
• running the simulation for 200ms with the fault applied;
• tripping the faulted line; and
• running the simulation on to 5 seconds
The simulation was run with a time step of 0.01seconds. Bus voltages were
monitored at every time step.
C.1.4.2. Varying Power Input Test
The varying power input test was designed to assess the effect on the network of a
varying power output from wind farms. Variation in wind farm power was achieved
using the DFIGIN-03 model, as described above. The test involved running a
simulation for 60 seconds.
The simulation was run with a time step of 0.01 seconds. Given the nature of the
test, bus voltages and machine powers were monitored at every ten time steps.
C.2. Results for Study of the Effects of a High Penetration of Wind Farms
For each of the two test procedures, results are presented using plots with some
comments attached. Each plot compares the results for a single load condition and
multiple wind farm outputs or a single wind farm output and multiple load
conditions.
251
C.2.1.1. Line Fault Disturbance Test
C.2.1.1.1. No Wind Farms and Different Load Conditions
Figure C.2 – Bus 1 voltage for no wind farms and different load conditions
This plot demonstrates how the network recovers after the line fault under all three
load conditions when there are no wind farms connected.
252
Figure C.3 – Bus 30 voltage for no wind farms and different load conditions
This plot shows how the voltage at bus 30, far from the fault on line 1-2, does not dip
as low as the voltage at buses closer to the fault and has a satisfactory post-fault
recovery.
253
C.2.1.1.2. 0MW Wind Farm Outputs and Different Load Conditions
load conditions when the wind farms are connected. Comparison with Figure C.2
shows that the presence of wind farms does not prevent the voltage at bus 1 dropping
to zero during the fault. Although not evident in the plot, the wind farm at bus 3 trips
as a result of the fault for all three load conditions.
254
This plot shows how the voltage at bus 30 is maintained close to its target level
throughout the line-fault simulation. Comparison with Figure C.3 shows that the
presence of the wind farms has helped to maintain the voltage at bus 30.
255
load conditions when the wind farms are connected and are producing 5MW each.
Comparison with Figure C.2 and Figure C.4 shows that the presence of wind farms
does not prevent the voltage at bus 1 dropping to zero during the fault, even if they
are producing power. Although not evident in the plot, the wind farm at bus 3 trips
as a result of the fault for all three load conditions.
256
throughout the line-fault simulation. Comparison with Figure C.5 shows greater
variation in the bus 30 voltage under the different load conditions when the wind
farms are producing power.
257
This plot demonstrates how the network recovers after the line fault with the wind
farms producing 10MW. Greater power output from the wind farms does not
prevent the voltage at bus 1 dropping to zero during the fault. Comparisons with
previous plots of the bus voltage show more obvious differences between the traces
for the three load conditions. Under the low and medium load conditions, wind
farms at buses 3, 7 and 28 trip as a result of the fault. Under the high load condition,
only the wind farm at bus 3 trips.
258
through the line-fault but then settles slightly below its original value.
259
This plot demonstrates how the network recovers after the line fault with the wind
farms producing 15MW. The post-fault recovery is less smooth than in previous
simulations. The voltage at bus 1 still drops to zero during the fault. Under all three
load conditions, wind farms at buses 3, 7 and 28 trip as a result on the fault.
260
through the line-fault but then settles slightly below its original value.
261
This plot shows how the line fault leads to instability when the wind farms are each
producing 20MW at the time of the fault. With the wind farm machine models
operating at a higher power output level, the line fault causes the rotor current
magnitudes to exceed the limit specified for protection. All 15 wind farms trip for all
three load conditions.
262
This plot shows how the voltage collapses at bus 30, with small oscillations
established due to the unstable swings of the conventional generators.
263
C.2.1.1.7. 20MW Wind Farm Outputs and Different Load Conditions and High
Protection Setting
A further study was performed for wind farm scenario d (20MW) with the protection
setting raised to a level (50pu) that avoids all wind farm trips. This produced a
different set of results, as shown below.
Figure C.14 – Bus 1 voltage for 20MW wind farm outputs and different load conditions with
high protection setting
This plot shows the voltage at bus 1 recovering after the line fault under the high
load condition. Through comparison with Figure C.12 it is clear that it is the tripping
of the wind farms and the loss of their power output that causes instability.
264
Figure C.15 – Bus 30 voltage for 20MW wind farm outputs and different load conditions with
high protection setting
This plot shows that if the wind farms remain connected through the fault
disturbance then the voltage at buses deep in the network, such as bus 30 shown here,
can be maintained through a fault disturbance.
This issue of fault ride-through depending on different protection settings is worthy

of further investigation but was not examined any further in these studies.
265
C.2.1.1.8. Low Load Condition and Different Wind Farm Conditions
Figure C.16 – Bus 1 voltage for low load condition and different wind farm conditions
This plot shows how the smoothness of post-fault recovery worsens as the wind farm
outputs are increased. Ultimately, wind farm outputs of 20MW result in post-fault
instability.
266
Figure C.17 – Bus 30 voltage for low load condition and different wind farm conditions
This plot shows firstly how the pre-fault voltage at bus 30 is different for each of the
wind farm conditions. The voltage drop during the fault depends on whether wind
farms are connected (no wind farms in scenario x) and whether they stay connected
(all wind farms trip in scenario d). The voltage recovers post-fault for all scenarios
except for scenario d.
267
C.2.1.1.9. Medium Load Condition and Different Wind Farm Conditions
Figure C.18 – Bus 1 voltage for medium load condition and different wind farm conditions
instability.
268
Figure C.19 – Bus 30 voltage for medium load condition and different wind farm conditions
269
C.2.1.1.10. High Load Condition and Different Wind Farm Conditions
Figure C.20 – Bus 1 voltage for high load condition and different wind farm conditions
instability. Comparison with Figure C.16 and Figure C.18 show how in this case, the
response for scenario b is grouped with those of scenarios 0 and a, whereas in the
low and medium load conditions, scenario b is distinct. For all three load conditions,
only the wind farm at bus 3 trips for scenarios 0 and a; scenario c results in trips at
buses 3, 7 and 28; and scenario d results in all the wind farms tripping. For scenario
b, under low and medium load conditions, the wind farms at buses 3, 7 and 28 trip;
but under high load conditions, only the wind farm at bus 3 trips.
270
Figure C.21 – Bus 30 voltage for high load condition and different wind farm conditions
271
C.2.1.2. Varying Power Input Test
C.2.1.2.1. Wind Farm Scenario 0 (0MW Base Level) and Different Load
Conditions
Figure C.22 – Bus 30 wind farm output for wind farm scenario 0 (0MW base level) and different
load conditions
This plot shows how the output from a wind farm varies over the 60-second
simulation. The abrupt changes in input power defined in the model code become
overshoots and oscillations in the output power from the wind farm.
272
Figure C.23 – Bus 2 generator output for wind farm scenario 0 (0MW base level) and different
load conditions
This plot shows the output from the generator at the swing bus. Its base level is
different for the different load conditions and the output changes in response to the
changing outputs from the wind farms.
273
Figure C.24 – Bus 30 voltage for wind farm scenario 0 (0MW base level) and different load
conditions
This plot shows how the voltage at bus 30 – far from the swing bus and with a wind
farm connected – varies as the power from the wind farms vary. The variations in
wind farm power are reflected in the bus voltage. The base level of the voltage at
this bus is the same for the low and medium load conditions and different for the
high load condition.
274
Figure C.25 – Bus 2 voltage for wind farm scenario 0 (0MW base level) and different load
conditions
This plot shows how the voltage at bus 2 – the swing bus – varies as the power from
the wind farms vary. The base level of the voltage at this bus is the same for the
different load conditions.
275
C.2.1.2.2. Wind Farm Scenario A (5MW Base Level) and Different Load
Conditions
Figure C.26 – Bus 30 wind farm output for wind farm scenario A (5MW base level) and
different load conditions
This plot shows how the output from the wind farm varies over the 60-second
overshoots and oscillations in the output power from the wind farm. It is apparent
that the final oscillations at the end of the simulation are slightly less for the lower
load conditions.
276
Figure C.27 – Bus 2 generator output for wind farm scenario A (5MW base level) and different
load conditions
277
Figure C.28 – Bus 30 voltage for wind farm scenario A (5MW base level) and different load
conditions
this bus is different for the different load conditions.
278
Figure C.29 – Bus 2 voltage for wind farm scenario A (5MW base level) and different load
conditions
the wind farms vary. The base level of the voltage at this bus is the same for the
different load conditions.
279
C.2.1.2.3. Wind Farm Scenario B (10MW Base Level) and Different Load
Conditions
Figure C.30 – Bus 30 wind farm output for wind farm scenario B (10MW base level) and
load conditions.
280
Figure C.31 – Bus 2 generator output for wind farm scenario B (10MW base level) and different
load conditions
281
Figure C.32 – Bus 30 voltage for wind farm scenario B (10MW base level) and different load
conditions
this bus is different for the different load conditions. The voltage varies most for the
high load condition and stays within the narrowest band for the low load condition.
282
Figure C.33 – Bus 2 voltage for wind farm scenario B (10MW base level) and different load
conditions
the wind farms vary. The variations in power from the wind farms and conventional
generators are reflected in the bus voltage. The base level of the voltage at this bus is
the same for the different load conditions.
283
C.2.1.2.4. Wind Farm Scenario C (15MW Base Level) and Different Load
Conditions
Figure C.34 – Bus 30 wind farm output for wind farm scenario C (15MW base level) and
load conditions.
284
Figure C.35 – Bus 2 generator output for wind farm scenario C (15MW base level) and different
load conditions
285
Figure C.36 – Bus 30 voltage for wind farm scenario C (15MW base level) and different load
conditions
286
Figure C.37 – Bus 2 voltage for wind farm scenario C (15MW base level) and different load
conditions
the same for the different load conditions.
287
C.2.1.2.5. Wind Farm Scenario D (20MW Base Level) and Different Load
Conditions
Figure C.38 – Bus 30 wind farm output for wind farm scenario D (20MW base level) and
load conditions.
Wind Farm output starts at 20MW but input is actually higher due to losses in the
machine. Thus, moving the power input to 20MW actually reduces the power
slightly, as can be seen at the start of this simulation.
288
Figure C.39 – Bus 2 generator output for wind farm scenario D (20MW base level) and different
load conditions
289
Figure C.40 – Bus 30 voltage for wind farm scenario D (20MW base level) and different load
conditions
290
Figure C.41 – Bus 2 voltage for wind farm scenario D (20MW base level) and different load
conditions
the same for the different load conditions. The voltage variation is greatest for the
low load condition.
291
C.2.1.2.6. Low Load Condition and Different Wind Farm Scenarios
Figure C.42 – Bus 30 wind farm output for low load condition and different wind farm scenarios
(different MW base levels)
This plot shows how the different outputs from the wind farm at bus 30 vary over the
60-second simulations for the low load condition. The abrupt changes in input
power defined in the model code become overshoots and oscillations in the output
power from the wind farm. At the end of the simulation, the oscillations are smallest
for the scenario B wind farm output (10MW).
These differences in the amplitudes of oscillation were thought to be because of

interference between the different oscillations instigated by the different changes in
input power. To test this hypothesis, a further set of studies was conducted for the
low load condition with a slightly different pattern of wind power input changes.
These are described in the next section.
292
Figure C.43 – Bus 2 generator output for low load condition and different wind farm scenarios
different for the different wind farm scenarios (different wind farm output base
levels). The output changes in response to the changing outputs from the wind
farms.
293
Figure C.44 – Bus 30 voltage for low load condition and different wind farm scenarios (different
MW base levels)
farm connected – varies for the different wind farm scenarios in the low load
condition. The variations in wind farm power are reflected in the bus voltage. The
base level of the voltage at this bus is different for the different wind farm scenarios;
higher wind farm outputs result in higher bus voltages.
294
MW base levels)
This plot shows how the voltage at bus 2 – the swing bus – varies for the different
wind farm scenarios in the low load condition. The variations in power from the
wind farms and conventional generators are reflected in the bus voltage. The base
level of the voltage at this bus is the same for the different wind farm scenarios.
C.2.1.2.7. Low Load Condition and Different Wind Farm Scenarios with
Different Power Input Pattern
In the previous section it was noted that the power from a wind farm displayed
oscillations of different amplitudes for different wind farm scenarios. These
differences were thought to be because of interference between the different
oscillations instigated by the different changes in input power. To test this
hypothesis, a further set of studies was conducted for the low load condition with a
slightly different pattern of wind power input changes.
295
The oscillations appear to have a period of approximately six seconds so the power
input pattern was altered by delaying the final ramp up by three seconds. Thus, the
new power input pattern was as shown in Table C.11.
Table C.11 – New pattern of input power variations
Time Period Input Power Variations

0 to 5 seconds Power held at its initial value
5 to 10 seconds Power ramped up to 100% of MVA base
10 to 20 seconds Power held at 100% of MVA base
20 to 30 seconds Power ramped down to zero
30 to 43 seconds Power held at zero
43 to 48 seconds Power ramped back up to its initial value
After 48 seconds Power held at its value
Figure C.46 – Bus 30 wind farm output for low load condition and different wind farm scenarios
This plot shows that with the final ramp in input power delayed by three seconds,
there is constructive rather than destructive interference of the power output
296
waveforms resulting in oscillations of much greater amplitude. In fact, the additive
effects are enough in scenario d to drive the machine power and rotor current
magnitude high enough that the protection is operated and the machine trips.
Figure C.47 – Bus 2 generator output for low load condition and different wind farm scenarios
This plot shows how the generator at the swing bus responds to the varying power
from the wind farms and adjusts when wind farms trip in scenario d.
297
MW base levels) and the adjusted pattern of input power
This plot shows how the voltage at bus 30 collapses in scenario d when the wind
farm trips due to excessive rotor current magnitude, which itself is due to
constructive interference between the oscillations instigated by changes in the input
power.
298
MW base levels) and the adjusted pattern of input power
This plot shows how the voltage at bus 2 is affected by the loss of wind farms caused
by the constructive interference of oscillations in power.
This issue of interference between the oscillations established by changes in input

power is worthy of further investigation but was not examined any further in these
studies.
299
C.2.1.2.8. Medium Load Condition and Different Wind Farm Scenarios
Figure C.50 – Bus 30 wind farm output for medium load condition and different wind farm
scenarios (different MW base levels)
60-second simulations for the medium load condition. The abrupt changes in input
power from the wind farm. At the end of the simulation, the oscillations are smallest
for the scenario B wind farm output (10MW). This is thought to be because the
oscillations instigated by upward and downward changes in input power cancel each
other out; and scenario B has the best balance of upward and downward changes.
300
Figure C.51 – Bus 2 generator output for medium load condition and different wind farm
farms.
301
Figure C.52 – Bus 30 voltage for medium load condition and different wind farm scenarios
farm connected – varies for the different wind farm scenarios in the medium load
302
Figure C.53 – Bus 2 voltage for medium load condition and different wind farm scenarios
wind farm scenarios in the medium load condition. The variations in power from the
303
C.2.1.2.9. High Load Condition and Different Wind Farm Scenarios
Figure C.54 – Bus 30 wind farm output for high load condition and different wind farm
60-second simulations for the high load condition. The abrupt changes in input
power from the wind farm.
304
Figure C.55 – Bus 2 generator output for high load condition and different wind farm scenarios
farms.
305
Figure C.56 – Bus 30 voltage for high load condition and different wind farm scenarios
farm connected – varies for the different wind farm scenarios in the high load
306
Figure C.57 – Bus 2 voltage for high load condition and different wind farm scenarios (different
MW base levels)
wind farm scenarios in the high load condition. The variations in power from the
307

Colin Foote PHD Thesis March 2007

Uploaded by

Copyright:

Available Formats

Colin Foote PHD Thesis March 2007

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Colin Foote PHD Thesis March 2007

Uploaded by

Copyright:

Available Formats

New Methods for New Challenges in Electricity

Distribution Network Planning

Submitted for the Degree

Institute for Energy and Environment

Figure 2.1 – Symbols used in knowledge models........................................................ 9

Table 6.1 – Initial Construction Cost Data................................................................. 87

1.2. Principal Contributions

Following this review of conventional practice, new challenges and shortcomings, a

In this Chapter, the conventional perspective of electricity distribution network

2.1. The Distribution Network Planning “Problem”

The basic purpose of electricity distribution networks is to deliver electrical power

In electricity distribution network planning, objectives and constraints are concerned

2.2. Model of the Conventional Approach to Distribution Network Planning

2.2.1. Knowledge Modelling Methodology

Figure 2.1 – Symbols used in knowledge models

2.2.2. Knowledge Models

Level One: Plan and design the distribution network

Plan and design the

Level Two: Identify requirements

In identifying the requirements of the distribution network (Figure 2.3), planners

Figure 2.3 – Task Model: Identify requirements

Consider goals and

Level Three: Consider goals and responsibilities

Distribution network companies are normally subject to strong government

Parent: Level Two: Identify requirements

Satisfy all relevant

Consider goals and

Level Three: Consider development drivers

In the conventional approach to distribution network planning and design, six

Figure 2.5 – Task Model: Consider development drivers

Consider general load

Consider new products

Level Four: Consider new connections

In considering new connections to the network (Figure 2.6), planners must be

Parent: Level Three: Consider development drivers

Figure 2.6 – Task Model: Consider new connections

Assess applications for

Level Four: Consider general load growth

In conventional distribution network planning, an important element is forecasting

Parent: Level Three: Consider development drivers

Consider general load

Identify insecurities and

Level Five: Forecast load

Parent: Level Four: Consider general load growth

Figure 2.8 – Task Model: Forecast load

Level Five: Identify insecurities and exceeded ratings

The identification of insecurities and exceeded ratings relies on feedback from

Parent: Level Four: Consider general load growth

Figure 2.9 – Task Model: Identify insecurities and exceeded ratings

Level Four: Consider asset replacement or refurbishment

Figure 2.10 – Task Model: Consider asset replacement or refurbishment

Level Five: Assess condition of plant

Parent: Level Four: Consider asset replacement or refurbishment

Level Five: Identify redundant network

Identifying redundant network requires the same combination of feedback from

Parent: Level Four: Consider asset replacement or refurbishment

Figure 2.12 – Task Model: Identify redundant network

Level Four: Consider network performance

Parent: Level Three: Consider development drivers