Abstract image of a woman sitting on books and studying on a laptop

Thesis458

M
Matt Wardhaugh

This thesis investigates the impacts of electric vehicle charging on residential distribution systems. It finds that uncoordinated charging could overload low voltage networks, particularly overhead networks where voltage unbalance is a greater issue. Simple staggered off-peak charging is shown to mitigate these impacts and allow up to 100% electric vehicle penetration. Uncoordinated charging is also found to significantly impact zone substations, possibly requiring upgrades within the next decade without coordinated strategies. A graphical user interface was created to model different electric vehicle loading scenarios and assess their effects on the distribution network.

1 of 76
Downloaded 23 times
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
              The  Impacts  of  Electric  Vehicle  Charging   on  Residential  Distribution  Systems         Matthew  Wardhaugh   Bachelor  of  Engineering  (Electrical)             October,  2012               Supervisor:  Dr  Phil  Ciufo                
i     i         Abstract     A  significant  increase  in  the  number  of  electric  vehicles  is  expected  over  the  coming   years,  and  this  is  expected  to  create  issues  for  distribution  networks  when  charging   coincides  with  peak  demand  periods.  This  thesis  investigates  the  effects  of   uncoordinated  charging  on  the  residential  distribution  network,  and  looks  at  the   viability  of  coordinated  charging  to  mitigate  these  effects.  A  graphical  user  interface  was   created  to  aid  this  study  and  provide  a  tool  for  network  planners  to  easily  run  electric   vehicle  loading  scenarios.  This  thesis  finds  that  uncoordinated  charging  would  have  an   impact  on  low  voltage  networks,  particularly  for  overhead  networks  where  voltage   unbalance  is  a  greater  issue.  Simple  staggered  off-­‐peak  charging  was  investigated  and   found  to  mitigate  loading  effects  completely,  allowing  up  to  100%  electric  vehicle   penetration  for  the  highest  charger  rating  scenario.  The  impact  of  charging  was  found  to   be  significant  at  the  zone  substation  level  during  uncoordinated  charging  scenarios,   possibly  requiring  upgrades  within  the  next  decade  if  coordinated  charging  strategies   are  not  adopted.                                    
ii     ii       Acknowledgements         I  would  like  to  thank  my  supervisors  Dr.  Phil  Ciufo  and  Prof.  Danny  Soetanto  for  their   guidance,  and  Endeavour  Energy  for  providing  network  models  and  data.    
iii     iii         Statement  of  Originality         I,  Matthew  Wardhaugh,  declare  that  this  thesis,  submitted  as  part  of  the  requirements   for  the  award  of  Bachelor  of  Engineering,  in  the  School  of  Electrical,  Computer  and   Telecommunications  Engineering,  University  of  Wollongong,  is  wholly  my  own  work   unless  otherwise  referenced  or  acknowledged.  The  document  has  not  been  submitted   for  qualifications  or  assessment  at  any  other  academic  institution.         Signature:                     Print  Name:                     Student  ID  Number:   3667315     Date:                        
iv     iv     Contents             Abstract  ....................................................................................................................................................................  i   Acknowledgements  ...........................................................................................................................................  ii   Statement  of  Originality  .................................................................................................................................  iii   Contents  ................................................................................................................................................................  iv   List  of  Figures  ......................................................................................................................................................  vi   List  of  Tables  ......................................................................................................................................................  vii   List  of  Equations  .............................................................................................................................................  viii   Abbreviations  and  Symbols  ..........................................................................................................................  ix   List  of  Changes  ......................................................................................................................................................  x   1   Introduction  ................................................................................................................................................  1   2   Literature  Review  .....................................................................................................................................  3   2.1   Power  system  and  network  configuration  ............................................................................  3   2.1.1   Layout  of  grid  ............................................................................................................................  3   2.1.2   Feeder  Voltages  ........................................................................................................................  3   2.1.3   Voltage  Correction  ..................................................................................................................  4   2.2   Electric  Vehicles  ................................................................................................................................  5   2.2.1   EV,  PHEV,  Extended  Range  EV  ...........................................................................................  5   2.2.2    Configuration  ...........................................................................................................................  6   2.2.3    Battery  system  .........................................................................................................................  6   2.2.4   Charging  ......................................................................................................................................  7   2.2.5   Growth  ..........................................................................................................................................  8   2.3   Impacts  of  Charging  .........................................................................................................................  9   2.3.1     Uncoordinated  Charging  ......................................................................................................  9   2.3.2    Coordinated  Charging  .........................................................................................................  10   2.4   Summary  ............................................................................................................................................  12   3   Methodology  .............................................................................................................................................  13   3.1   Load  Flow  ..........................................................................................................................................  13   3.1.1   Load-­‐Flow  Solutions  .............................................................................................................  13   3.1.2   Load  Types  ...............................................................................................................................  13   3.2   Modelling  ...........................................................................................................................................  15   3.2.1   DIgSILENT  PowerFactory  Models  ..................................................................................  15   3.2.2   DIgSILENT  Programming  Language  (DPL)  Script  ...................................................  18  
v       v       3.2.3   Load  Profiles  ............................................................................................................................  18   3.2.4   Loading  Assumptions  ..........................................................................................................  20   3.2.5   Load  Scaling  .............................................................................................................................  21   3.3   Simulation  ..........................................................................................................................................  26   3.3.1   Graphical  User  Interface  .....................................................................................................  26   3.3.2   GUI  Structure  ...........................................................................................................................  27   3.4   Scenarios  ............................................................................................................................................  29   3.4.1   Uncoordinated  Charging  ....................................................................................................  29   3.4.2   Coordinated  Charging  ..........................................................................................................  30   3.4.3   11  kV  ...........................................................................................................................................  31   4   Results  .........................................................................................................................................................  33   4.1   Base  Load  Profile  ............................................................................................................................  33   4.1.1   Effects  of  Temperature  on  Substation  Loading  ........................................................  33   4.1.2   Load  Scaling  .............................................................................................................................  33   4.1.3   Network  Type  .........................................................................................................................  34   4.2   Uncoordinated  Charging  .............................................................................................................  35   4.2.1   11  kV  Voltage  Regulation  ...................................................................................................  35   4.2.2   400  V    Transformer  and  Feeder  Loading  ....................................................................  36   4.2.3   11  kV  Transformer  Loading  ..............................................................................................  42   4.3   Coordinated  Charging  ...................................................................................................................  43   4.3.1   3-­‐Group  Charging  ..................................................................................................................  43   4.3.2   Six-­‐Group  Charging  ...............................................................................................................  45   4.3.3   11  kV  ...........................................................................................................................................  45   5   Conclusion  .................................................................................................................................................  46   References  ...........................................................................................................................................................  48   Appendix  A  ..........................................................................................................................................................  51   Appendix  B  ..........................................................................................................................................................  54   Appendix  C  ..........................................................................................................................................................  55                        
vi     vi       List  of  Figures         Figure  2.1:  Radial  Feeder  Distribution  ......................................................................................................  3   Figure  2.2:  Feeder  Voltage  Profiles  ............................................................................................................  4   Figure  2.3:  Electric  Vehicle  Configuration  ...............................................................................................  6   Figure  2.4:  Lithium-­‐Ion  Charge  Curve  [26]  .............................................................................................  8   Figure  3.1:  Load  Flow  Analysis  [4]  ...........................................................................................................  13   Figure  3.2:  400  V  Overhead/Underground  DIgSILENT  Model  .....................................................  16   Figure  3.3:  11  kV  Overhead  DIgSILENT  Model  ....................................................................................  17   Figure  3.4:  Average  number  of  travellers  in  NSW  on  weekdays  in  2010/11  ........................  19   Figure  3.5:  Scaled  driver  arrival  times  ....................................................................................................  20   Figure  3.6:  Feeder  voltage  profile,  moving  from  last  premise  to  transformer  from  right  to   left  ...........................................................................................................................................................................  23   Figure  3.7:  MATLAB  GUI  ...............................................................................................................................  26   Figure  3.8:  Flowchart  displaying  the  interaction  of  programs  required  for  GUI   simulations  ..........................................................................................................................................................  27   Figure  4.1:  Woodlands  Drive  substation  loading  for  38.7  and  19.9  degrees  celsius  days33   Figure  4.2:  Woodlands  Drive  substation  total  load  compared  to  scaled  sample  loads  .....  34   Figure  4.3:  Woodlands  Drive  substation  load  for  overhead  and  underground  networks34   Figure  4.4:  Impact  of  increasing  charger  rating  on  undergroudn  network  at  100%  EV   penetration  .........................................................................................................................................................  40   Figure  4.5:  4  kW  three-­‐group  coordinated  charging  for  different  transformer  base  levels  ..................................................................................................................................................................................  43   Figure  4.6:  Six-­‐group  coordinated  charging  for  a  95%  loaded  transformer  ..........................  45      
vii     vii       List  of  Tables       Table  2.1:  Current  EV  Battery  Capacities  [11][13-­‐16]  .......................................................................  7   Table  2.2:  International  EV  Charging  Standards  ..................................................................................  7   Table  3.1:  Network  Equipment  Parameters  .........................................................................................  17   Table  3.2:  Variable  Options  Structure  .....................................................................................................  27   Table  4.1:  Woodlands  Drive  substation  transformer  loading  and  voltage  regulation  for   varying  EV  penetrations  ................................................................................................................................  36   Table  4.2:  Maximum  EV  penetration  for  4  kW  LV  uncoordinated  charging  ...........................  38   Table  4.3:  Maximum  EV  penetration  for  7  kW  LV  uncoordinated  charging  ...........................  39   Table  4.4:  Maximum  EV  penetration  for  10  kW  LV  uncoordinated  charging  ........................  40   Table  4.5:  Maximum  EV  penetration  at  zone  substation  assuming  worst  loading  day  in   2010/11  ...............................................................................................................................................................  42   Table  4.6:  Maximum  EV  penetration  for  7kW  LV  coordinated  charging  .................................  44   Table  4.7:  Maximum  EV  penetration  for  10  kW  LV  coordinated  charging  .............................  44                        
viii     viii       List  of  Equations       Equation  3.1  ........................................................................................................................................................  14   Equation  3.2  ........................................................................................................................................................  14   Equation  3.3  ........................................................................................................................................................  21   Equation  3.4  ........................................................................................................................................................  22   Equation  3.5  ........................................................................................................................................................  23   Equation  3.6  ........................................................................................................................................................  23   Equation  3.7  ........................................................................................................................................................  24   Equation  3.8  ........................................................................................................................................................  24   Equation  3.9  ........................................................................................................................................................  24   Equation  3.10  .....................................................................................................................................................  24   Equation  3.11  .....................................................................................................................................................  25   Equation  3.12  .....................................................................................................................................................  25   Equation  3.13  .....................................................................................................................................................  25                          
ix     ix         Abbreviations  and  Symbols       EV     Electric  Vehicle   BEV     Battery  Electric  Vehicle   PHEV     Plug-­‐In  Hybrid  Electric  Vehicle   IC     Internal  Combustion   V2G     Vehicle  to  Grid   OLTC     On-­‐load  tap  changer   SC     Switched  capacitor   SoC     State  of  Charge   Li-­‐ion     Lithium  ion   NiMH     Nickel-­‐metal  hydride   PV     Photovoltaic   DC     Direct  current   AC     Alternating  current   pu     per  unit   𝑗𝑋     Reactance,  Ohms   𝑅     Resistance,  Ohms   𝑍     Impedance,  Ohms   𝑃     Power,  Watts   𝑉     Voltage,  Volts        
x       x         List  of  Changes         Section   Statement  of  Changes   Page  Number   1   Removed  references  to  solar  and  V2G,  added  description   of  new  work   1,2   2.2   Removed  sentence  relating  to  V2G   5   2   Removed  Solar  section   -­‐   2.3.2   Removed  Solar  sub-­‐subsection   11   2   Removed  ‘V2G  Benefits’  section   -­‐   2.3.1   Added  analysis  of  loading  assumptions  in  literature   10   3   Replaced  Methodology  section   32   4   Replaced  Results  section   13  
1     1     1 Introduction       The  world  is  currently  experiencing  a  major  shift  in  the  way  energy  is  generated  and   consumed.  Pressing  issues  such  as  climate  change  and  declining  fossil  fuel  reserves  are   changing  the  way  people  think  about  the  environment.  Also,  technological  advances  are   allowing  renewable  generation  and  energy  storage  to  become  technically  and   economically  viable,  paving  the  way  for  an  emissions  free  future.   Electric  vehicles  (EV)  and  plug  in  hybrid  electric  vehicles  (PHEV)  (used   interchangeably  in  this  text)  are  becoming  increasingly  popular  due  to  the  impetus  of   these  factors.  Significant  advances  in  battery  storage  capabilities  are  allowing  EVs  to   become  a  viable  alternative  to  internal  combustion  (IC)  vehicles.  Their  storage  of   electricity  allows  energy  to  be  sourced  from  renewable  sources  such  as  wind  and  solar,   allowing  for  zero  emission  driving.  This  is  significant,  as  it  would  play  a  large  role  in   reducing  CO₂  emissions  and  localised  air  pollution  levels  [1].    Without  proper  planning,  however,  EVs  are  expected  to  produce  undesired   impacts  on  the  low  voltage  distribution  network  when  charged  in  an  uncoordinated   manner.  Charging  will  occur  whenever  convenient  for  the  driver,  such  as  on  arrival   home  from  work,  increasing  the  evening  peak  load  and  causing  stress  to  network   equipment,  particularly  at  distribution  levels.  Due  to  the  large  amount  of  energy  drawn   during  charging  periods,  it  is  expected  that  at  high  penetration  levels  this  will  present   serious  power  quality  issues  for  the  grid,  including  potential  transformer  overloading   and  voltage  sags,  resulting  in  outages,  equipment  damage  and  energy  loss  [2][3].   This  outcome  may  be  avoided  if  electric  vehicle  charging  can  be  coordinated  in   such  a  way  to  avoid  the  evening  load,  and  instead  be  automated  for  charging  during  low-­‐ demand  periods,  such  as  late  at  night.  Smart  infrastructure  currently  being   contemplated  will  allow  charging  times  to  be  staggered  between  different  households  to   allow  a  more  evenly  distributed  feeder  load.   The   proposed   focus   of   this   thesis   is   to   investigate   the   impact   of   introducing   a   significant   number   of   EVs   on   the   residential   distribution   system,   particularly   during   uncoordinated  charging  periods  that  coincide  with  peak  load.  The  load  flow  simulation   package  DIgSILENT  PowerFactory  will  be  used  to  carry  out  the  investigations.  Means  of   avoiding   the   undesirable   impacts   of   EV   charging   will   be   investigated,   using   several  
2     2     scenarios   to   determine   the   viability   of   load   levelling.   This   study   will   determine   the   effects   of   charging   on   residential   feeder   voltage   levels,   consequently   discerning   the   associated  impacts  on  transformer  loading  and  energy  loss.   In  order  to  study  the  impacts  of  charging  on  the  residential  distribution  network,   typical  400V  and  11  kV  radial  residential  feeders  have  been  modelled  in  PowerFactory,   using  smart  metering  data  from  premises  in  the  Endeavour  Energy  network  area  of   Glenmore  Park.  Associated  variables  have  been  accounted  for,  including  battery   capacities,  charging  power,  base  load  demand,  load  power  factor  and  phase  unbalance.   To  aid  network  planners  in  making  decisions  based  on  future  electric  vehicle  loading,  a   graphical  user  interface  has  been  developed  using  MATLAB  GUIDE.  This  allows   DIgSILENT  PowerFactory  to  be  controlled  remotely  to  run  various  EV  loading  scenarios,   displaying  transformer  loading  and  voltage  regulation  results  both  numerically  and   graphically  for  analysis.                                          
3     3     2 Literature  Review     2.1 Power  system  and  network  configuration     2.1.1   Layout  of  grid     The  electricity  grid  is  a  complex  network  that  acts  as  a  path  for  electricity  from  generators   to  consumers.  The  layout  of  the  grid  is  an  important  concept  that  must  be  understand  to   grasp  an  idea  of  how  electric  vehicles  will  be  connected  and  the  effects  that  they  will  have  on   the  network.    The  traditional  grid  can  be  divided  into  generation,  transmission  and  distribution   levels.  The  transmission  network  steps  generator  voltages  up  in  order  to  reduce  the  losses   associated  with  high  currents  over  long  distances,  usually  at  230  kV  to  765  kV  [4].  As  these   high  voltage  feeders  branch  towards  large  populations,  they  are  stepped  down  in  to  the   distribution  network.  Zone  substations  convert  voltages  to  11  kV  for  residential  feeders,   which  then  connect  to  pole  top  or  pad  mount  transformers  that  finally  supply  400  V,  or       230  V  line-­‐to-­‐neutral,  for  use  in  homes  and  businesses  [5].   The  distribution  network  is  the  most  important  section  of  the  grid  to  understand  when   conducting  load  flow  analysis  on  residential  loads,  as  EVs  and  distributed  generation,  such   as  solar  PV,  are  both  connected  at  the  low  voltage  level.  From  zone  substations,  feeders  are   typically  connected  radially  [6][4]  as  they  branch  out  through  streets,  shown  in  Fig  2.1.  This   radial  layout  will  be  used  for  modelling  residential  feeders.     Figure  2.1:  Radial  Feeder  Distribution   2.1.2   Feeder  Voltages     Basic  circuit  theory  states  that  a  voltage  drop  will  result  as  current  flows  through  an   impedance.  Therefore,  as  transformer  loading  is  increased,  the  voltage  drop  along  a   feeder  becomes  greater.  Conversely,  during  periods  of  high  generation,  net  feeder  
4     4     current  is  reduced,  raising  voltage  levels  closer  to  that  of  the  transformer.    During  heavy   loading  or  generation  periods,  voltage  levels  may  surpass  utility  limits.  The  AS/NZS   3000:2007  states  that  in  Australia,  voltage  limits  must  not  move  beyond  +10%  or  -­‐6%  of   nominal  value  to  avoid  damage  to  connected  equipment,  corresponding  to  253  V  and   216  V  line-­‐to-­‐neutral  [5].    Fig.  2.2  shows  the  effects  of  different  load  scenarios  on  feeder   voltage  levels.  Realistically,  these  voltages  would  not  have  a  linear  profile,  even  for   uniform  loading  across  the  feeder,  as  currents,  and  hence  the  rate  of  voltage  drop,  is   greater  closer  to  the  transformer.       Figure  2.2:  Feeder  Voltage  Profiles   Another  consequence  of  voltage  deviations  along  feeders  is  power  loss.   Feeder  power  loss  is  proportional  to  the  square  of  a  voltage  change,  therefore  it  is   important  to  reduce  this  change  in  voltage  along  a  feeder  as  much  as  possible.     2.1.3   Voltage  Correction     Voltage  control  is  important  for  addressing  changes  in  line  voltages.  Network   equipment,  such  as  transformers  and  lines  are  designed  to  operate  within  certain   voltage  limits.  Most  importantly,  however,  are  the  loads  connected  to  LV  feeders,  which   may  become  damaged  while  drawing  power  at  excessive  or  limited  voltage  levels.     In  order  to  maintain  voltage  levels  within  a  specified  range  such  as  this,  a  range   of  network  equipment  is  utilised.  In  distribution  networks,  voltage  control  is  typically   achieved  using  on-­‐load  tap  changers  (OLTC),  step  voltage  regulators  (SVR)  and  switched   capacitors  (SC)  [7].  OLTCs  and  SVRs  are  both  autotransformers  with  automatic  tap   changing.  Normally  the  voltage  regulator  in  a  substation  is  an  OLTC,  while  an  SVR  would   be  located  along  a  feeder,  down  to  LV  levels  [7].    SCs  are  used  for  reactive  power   compensation  in  distribution  networks.  An  SC  reduces  the  displacement  between  real   and  reactive  power  components  to  reduce  voltage  drop  across  lines  that  are  primarily  
5     5     inductive.  In  low  voltage  networks,  the  most  common  voltage  regulators  are  off-­‐load   tap-­‐changers,  located  within  distribution  transformers  [8].  The  transformer  ratio  must   be  changed  manually,  generally  over  a  multiple  year  span  as  network  loading  increases.   Although  SVRs  and  switched  capacitors  can  exist  in  LV  areas,  this  is  uncommon  due  to   the  large  number  of  feeders,  and  the  associated  costs.   Therefore,  on  residential  feeders,  voltage  control  is  limited  to  off-­‐load  tap   changers  on  pole-­‐top  and  pad  mount  transformers.  The  manual  nature  of  this  tap   changing  is  uncoordinated,  therefore  this  is  far  from  being  an  optimal  solution  to   addressing  the  large  scale  integration  of  EVs.   Taking  the  characteristics  of  common  network  equipment  into  account,  the  coordinated   charging  of  EVs  can  be  seen  as  a  worthwhile  solution  to  this  problem  as  the  load  factor   of  a  feeder  may  be  reduced.     2.2   Electric  Vehicles     Electric  vehicles  are  vehicles  that  contain  a  rechargeable  battery  pack,  requiring   charging  by  a  grid  connected  battery  charger.  EVs  are  becoming  popular  as   environmental  awareness  is  increasing  across  the  world,  as  they  produce  little  to  no   emissions.  Improvements  in  battery  technology  are  seeing  prices  fall  rapidly,  allowing   EVs  to  become  a  viable  alternative  to  internal  combustion  (IC)  vehicles.  Penetration  of   EVs  is  beginning  to  increase,  with  over  20  models  due  to  reach  the  markets  in  2012  [9].     2.2.1   EV,  PHEV,  Extended  Range  EV     There  are  four  main  types  of  electric  vehicles  that  currently  exist:  Hybrid,  Plug-­‐in  Hybrid   (PHEV),  Extended-­‐Range  and  Battery  EVs  (BEV)  [10].  Hybrid  and  PHEVs  contain  both   combustion  engines  and  electric  motors  with  battery  storage.  Unlike  hybrids,  however,   PHEVs  can  also  be  charged  through  an  external  battery  charger,  further  reducing   reliance  on  the  combustion  engine  [10]  Extended-­‐Range  EVs  are  similar  to  PHEVs  and   include  vehicles  such  as  the  Holden  Volt  [11].  The  electric  engine  is  used  for  all  driving   speeds  until  the  battery  is  discharged,  and  is  then  replaced  by  the  combustion  engine.   Lastly,  BEVs  are  all  electric  with  no  combustion  engine.  They  contain  large  battery  packs   that  must  be  charged  by  the  grid.    
6     6     In  relation  to  the  topic  of  this  thesis,  hybrid  vehicles  are  considered  irrelevant,  as   they  are  not  charged  by  the  grid.  Therefore,  the  vehicles  of  focus  will  be  PHEVs,   Extended-­‐Range  EVs  and  BEVs,  referred  to  collectively  throughout  this  text  as  ‘EVs’.     2.2.2    Configuration     The  basic  configuration  of  an  EV,  including  an  IC  engine,  which  is  only  applicable  to   PHEVs  and  EREVs,  is  shown  by  the  simplified  block  diagram  in  Fig.  2.3.     Figure  2.3:  Electric  Vehicle  Configuration     Charging  requires  communication  with  the  battery-­‐monitoring  unit  that  measures  the   batteries  state  of  charge  (SoC).  The  inverter  is  used  after  a  DC-­‐DC  converter  to  convert   direct  current  (DC)  into  alternating  current  (AC)  to  power  the  electric  motor.     2.2.3    Battery  system     For  electric  vehicles  to  be  a  viable  alternative  to  IC  vehicles,  their  battery  storage  must   contain  enough  energy  to  ensure  suitable  range  for  drivers.  The  most  important  factor   affecting  this  is  the  energy  to  weight  ratio  of  a  battery  pack,  or  its  energy  density.  This   allows  vehicles  to  be  as  light  as  possible  for  a  given  amount  of  energy  storage,  ensuring   the  greatest  range  possible.   There  exist  three  main  battery  types  for  electric  vehicles:  lead-­‐acid,  nickel-­‐metal   hydride  (NiMH)  and  lithium-­‐ion  (li-­‐ion)  [12].  In  the  past,  EVs  such  as  the  General  Motors   EV1  used  lead-­‐acid  and  nickel-­‐metal  hydride  batteries.  In  recent  years,  however,  the   demand  for  batteries  in  laptops  and  other  portable  devices  has  driven  R&D  in  the  area  of   lithium-­‐ion  batteries,  improving  energy  density  and  charge  time  beyond  other  battery   types.  Due  to  these  improvements,  major  EV  manufacturers  now  use  lithium  ion  battery   packs  [11][13-­‐16].  
7     7     Table  2.1  provides  a  list  of  current  vehicles  and  their  battery  capacities,  showing  a   significant  range  of  battery  capacities  that  will  form  the  basis  for  modelling.     Electric  Vehicle   Battery  Capacity   Tesla  Model  S   40,  60,  85  kWh   Nissan  Leaf   24  kWh   Ford  Focus  Electric   23  kWh   Holden  Volt   8  kWh   Toyota  Prius  Plug-­‐In   4.4  kWh   Table  2.1:  Current  EV  Battery  Capacities  [11][13-­‐16]   2.2.4   Charging     Based  on  standards  by  the  International  Electrotechnical  Commission  (IEC)  [17]  and  the   Society  of  Automotive  Engineers  J1772  [18],  there  exists  three  charging  levels:     Level   Voltage     Current   Power   1   120  V  AC   16  A   1.92  kW   2   208-­‐240  V  AC   12  –  80  A   2.5  –  19.2  kW   3   500  V  DC   125  A   50  kW   Table  2.2:  International  EV  Charging  Standards     The  residential  charger  rating  of  EV  manufacturers  vary  substantially  within  the  Level  2   range.  Nissan  and  Holden’s  chargers  are  rated  3.3  kW  [16][11],  Ford’s  at  7.7  kW  [14],   while  Tesla  manufactures  10  kW  or  20  kW  chargers  [13].  These  ratings  are  significant  in   comparison  to  other  appliances  found  in  the  home.   Fig.  2.4  shows  the  power  demand  and  battery  SoC  profiles  of  a  lithium  ion   battery.  
8     8       Figure  2.4:  Lithium-­‐Ion  Charge  Curve  [26]     Figure  2.4  shows  a  predominantly  constant  charging  power  for  the  duration  of  the   charging  period.  Therefore,  for  modelling  purposes,  a  constant  charge  rate  can  be   considered  accurate  to  assume.     2.2.5   Growth     Due  to  economic  and  technological  factors  surrounding  the  viability  of  electric  vehicles,   their  penetration  levels  are  expected  to  soar  this  decade  [19-­‐21].  Current  estimates   expect  the  price  of  oil  to  rise  by  85%  into  2020  [19],  and  this  rise  is  forecast  to  continue.   By  the  same  time,  lithium  ion  battery  technology  is  expected  to  dramatically  fall  as   economies  of  scale  reduces  manufacturing  costs,  and  technological  improvements  allow   energy  density  to  continually  increase.  Lithium  ion  battery  prices  have  fallen   considerably  from  US$650/kWh  in  2009  to  the  current  price  of  around  US$450/kWh.     Analysts  have  forecasted  prices  to  fall  at  a  7.5%  annual  compound  rate  from  2012   through  2020  to  approximately  US$250/kWh  [19].  EV  manufacturer  Tesla  Motors  is   already  producing  battery  packs  with  480  km  of  range  [13].   Taking  these  factors  into  consideration,  analysts  from  Deutsche  Bank  [19]  have   predicted  that  in  the  US,  around  10%  of  all  vehicles  will  be  hybrid/electric  by  2021,   increasing  to  20%  by  2026,  and  35%  by  2030.  In  terms  of  purchased  vehicles,  EVs  are   expected  to  make  up  3-­‐10%  of  new  car  sales  as  early  as  2015  [20]  and  35%  in  2025,   comprised  of  25%  PHEVs  and  10%  EVs,  according  to  IDtechX  analysts  [21].  These   projections  show  that  a  major  shift  is  about  to  occur,  resulting  in  a  significant   percentage  of  vehicles  becoming  at  least  partially  electric.  This  analysis  raises  questions   about  the  effects  of  a  large  percentage  of  EVs  on  the  distribution  network,  as  well  as  the  
9     9     potential  problems  this  extra  energy  storage  may  solve.     2.3   Impacts  of  Charging     2.3.1     Uncoordinated  Charging     The  introduction  of  EVs  is  expected  to  have  a  significant  effect  on  customer  load  profiles   during  charging  periods.  Studies  in  [2],  [3]  and  [22]  have  concluded  that,  for  high   penetration  levels,  uncoordinated  domestic  charging  will  increase  peak  load  demand   significantly,  resulting  in  transformer  overloading,  poor  feeder  voltage  profiles  and   power  loss.       The  authors  of  [2]  and  [22]  have  conducted  studies  on  uncoordinated  charging   on  residential  radial  feeders,  focusing  on  evening  peaks.  The  modelled  charger  rating   was  4  kW  [2],  and  1.8  kW  in  [22],  both  showing  dramatic  rises  in  peak  load,  clearly   overloading  the  transformer  limitations  for  penetrations  above  20%  in  [22]  and   exceeding  voltage  limits  in  [2]  at  17%.  The  effects  of  peak-­‐time  charging  on  summer  and   winter  load  profiles  are  explored  in  [23]  and  [3].  The  UK  winter  load  profile  in  [23]   showed  a  distinct  evening  peak  compared  to  summer  due  to  electric  heating.  This   caused  the  peak  demand  to  be  increased  by  13.6%  compared  to  10.06%  for  summer  at   10%  EV  penetration.  Although  this  paper  conducts  a  load  study  for  the  entire  UK,  it  is   probable  that  this  would  reflect  the  demand  of  residential  feeders,  as  most  vehicles   would  be  at  home  during  this  period.  A  study  is  conducted  in  [3]  to  determine  the  effects   of  peak  charging  on  power  loss  and  voltage  deviation.  The  voltage  limit  of  0.9  pu  was   found  to  be  exceeded  at  30%  EV  penetration  with  a  4  kW  charger,  with  total  power  loss   at  6%  in  winter  compared  to  5%  in  summer.   These  papers  clearly  show  that  uncoordinated  charging  would  have  a  large   impact,  even  at  low  penetration  levels.  However,  an  analysis  of  these  papers  show  the   large  number  of  variables  associated  with  such  studies.  For  example,  the  voltage  limit  of   0.9pu  in  [3]  differs  to  0.94  used  in  Australia,  as  well  as  the  UK  load  profiles  in  [23].   Another  assumption  made  in  these  studies  is  a  relatively  low  powered  charger,   particularly  in  [22].  A  higher-­‐powered  charger  more  commonly  used  today  would  have  a   significantly  increase  the  peak  demand  determined  by  these  papers.  Of  all  the   assumptions  made,  however,  the  most  important  variable  used  to  determine  the  impacts  
10     10     of  uncoordinated  charging  is  the  time  the  vehicles  arrive  home  to  begin  charging.  In  the   related  papers  [2-­‐3]  [24-­‐26],  and  number  of  assumptions  in  relation  to  charging  times   have  been  made,  while  there  exists  a  significant  degree  of  ambiguity  when  these   assumptions,  such  as  the  data  used,  is  explained.  Papers  [2]  and  [22]  fail  to  explain  how   their  vehicle  arrival  times  are  modelled,  while  [23]  simply  divides  charging  into  three   groups  during  the  evening  peak,  assuming  that  all  vehicles  commence  charging  within   90  minutes  of  one  another.  Papers  [24]  and  [27]  assume  a  more  accurate  normal   distribution,  however  still  disregard  actual  driving  statistics,  such  as  those  provided  by   the  UK  Time  of  Use  survey  noted  in  [23]  and  [3].  Paper  [3]  takes  into  account  the   statistics  from  this  survey  by  dividing  charging  times  according  to  the  morning,  midday   and  late  afternoon  periods,  and  making  assumptions  about  the  percentage  of  cars  that   charge  during  these  times.  Paper  [3]  applies  the  most  accurate  data  regarding  charging   times  as  it  incorporates  the  irregular  and  skewed  peak  provided  by  a  traffic  authority.   Considering  this,  the  majority  of  research  has  been  conducted  with  inaccurate   assumptions,  possibly  causing  significant  variations  in  results  as  the  charging  times,   along  with  the  assumed  charger  rating,  are  the  factors  that  most  influence  the  results  of   loading  simulations.  Charging  times  for  the  uncoordinated  charging  simulations  in  this   thesis  will  be  based  on  local  driving  data  to  ensure  the  most  accurate  modelling  possible.       Therefore,  to  more  accurately  determine  the  effects  of  uncoordinated  charging,  it   is  important  to  use  local  load  profiles,  standards  and  driving  statistics,  with  assumptions   that  are  up  to  date,  or  reflect  expected  future  trends.  These  variables  will  be  taken  in  to   account  in  this  thesis,  to  more  accurately  determine  possible  effects  on  typical   Australian  residential  feeders.     2.3.2    Coordinated  Charging     The  effects  of  uncoordinated  charging  show  the  importance  of  coordinated  or   ‘smart’  charging  in  the  future.  This  would  be  achieved  through  communication   infrastructure  in  a  smart  grid,  by  sending  signals  to  begin  charging  at  times   corresponding  to  uniform  loading  [24].  Coordinated  charging  employs  heuristic   algorithms  and  optimization  techniques  with  the  aim  to  improve  load  factor  and  reduce   network  costs  and  power  losses  by  charging  during  off  peak  periods  [2][24].  As  cars  are   available  for  94.8%  of  the  day  on  average  [23],  coordinated  charging  can  be  considered   viable,  as  a  large  amount  of  flexibility  exists  in  charging  times.    
11     11     A  large  number  of  studies  have  been  conducted  on  novel  approaches  to   coordinating  vehicles,  with  the  aim  to  reduce  evening  peak  demand.  These  range  from   complicated  algorithms  based  on  real-­‐time  market  prices  in  [27]  to  prioritizing  charging   periods  in  [2],  to  simple  delayed  off-­‐peak  charging  in  [23].   Throughout  the  majority  of  coordinated  charging  studies,  the  uncertainties  of  variables,   such  as  load  profiles  and  charging  time,  are  expressed  in  terms  of  probability  density   functions,  allowing  predictions  to  be  made  without  relying  on  fixed-­‐input  variables,  such   as  an  average  past  load  profile  [27].  The  authors  in  [27]  determined  that  coordinated   charging  reduced  load  factor  and  power  losses  by  6-­‐28%  for  penetration  levels  from   10%  to  100%.    In  [27],  a  control  algorithm  was  implemented  for  coordinated  charging   on  an  LV  feeder  in  Belgium,  based  on  a  typical  local  load  profile.  The  results  showed  a   peak  demand  reduction  of  29%  for  a  combination  of  3.6  kW  and  7.4  kW  chargers  at  15%   penetration.    Papers  [2]  and  [27]  take  different  real-­‐time  approaches,  dividing  charging  times   into  red,  blue  and  green  zones,  based  on  the  priority  of  charging.  In  [27],  charging   priority  is  determined  based  on  the  time  vehicles  arrive  home,  as  a  vehicle  that  arrives   late  would  have  a  low  chance  of  being  used  for  the  remainder  of  the  night.  This  paper   found  that  load  demand  could  remain  below  the  evening  peak  for  penetration  levels  of   at  least  63%,  as  low  priority  vehicles  could  be  spread  further  into  the  morning  hours.   Above  this  penetration,  however,  this  paper  found  that  high  and  medium  priority   vehicles  raised  the  peak  demand  above  the  evening  peak,  therefore  stating  there  will   inevitably  be  a  rise  in  peak  demand  as  EV  penetration  reaches  high  levels.   The  study  in  [27]  assumes  a  2  kW  peak,  which  is  relatively  low,  especially  as  this   aims  to  determine  loading  decades  in  to  the  future,  which  is  expected  to  rise  irrespective   of  EVs.  Another  assumption  is  that  low  priority  charging  is  timed  to  finish  at  4  am,   however  this  could  realistically  be  increased  to  6  am,  for  example,  for  the  majority  of   people  who  leave  for  work  after  this  time.  This  would  allow  a  higher  penetration  before   peak  demand  is  raised.   The  authors  in  [23]  have  included  a  study  on  fixed  off-­‐peak  charging,  which  is   implemented  by  simply  charging  in  three  groups,  at  9  pm,  9:30  pm  and  10  pm.  This   avoids  the  evening  peak,  while  allowing  sufficient  time  to  charge  through  to  early   morning.  This  paper  finds  that  the  charging  peak  is  less  than  the  evening  peak  for  low   penetration,  but  states  that  this  may  not  be  the  case  for  penetration  greater  than  10%.   This  is  compared  to  a  study  on  ‘smart’  market  based  charging,  which  shows  a  noticeable  
12     12     reduction  in  charging  peak  load.  From  analysis  of  the  fixed  off-­‐peak  charging  graph,  it   shows  charging  is  finished  by  2  am.  This  shows  a  large  percentage  of  early  morning   hours  with  lower  base  demand  that  are  not  utilized,  therefore  it  could  be  argued  that   this  method  could  support  penetration  much  higher  than  the  10%  stated.  The  simplicity   of  the  fixed  off-­‐peak  method,  and  the  lack  of  research  associated  with  it,  presents  an   opportunity  for  study  in  this  thesis.  This  would  eliminate  the  need  for  complicated   algorithms  at  residential  feeders,  and  may  not  require  smart  infrastructure,  as  signalling   could  be  sent  via  high  frequency  pulses,  as  they  are  today  to  control  off-­‐peak  hot  water   systems.  Lower  electricity  rates  would  provide  the  incentive  for  the  majority  of  owners   to  use  this  method,  while  allowing  a  simple  manual  over-­‐ride  when  required.  However,   in  terms  of  load  levelling,  coordinated  charging  would  be  a  valuable  approach  to  further   reduce  energy  losses.  Initially,  this  method  will  be  tested  by  simulating  a  simple  fixed-­‐ start  delay,  with  preliminary  work  on  staggered  charging  to  focus  on  further  reducing   power  loss.     2.4   Summary     The  results  of  various  studies  related  to  charging  produce  a  wide  range  of  results  due  to   the  number  of  variables  associated  with  distribution  networks  and  electric  vehicles.   From  this  analysis,  a  noticeable  gap  exists  in  research  of  the  impact  of  EVs  applicable  to   Australian  residential  feeders.  Particularly,  there  is  a  lack  of  study  that  incorporates   realistic  driving  pattern  data,  through  either  the  use  of  information  from  traffic   authorities  or  by  conducting  surveys.                  
13     13     3 Methodology     The  study  of  literature  in  Chapter  2  presents  a  number  of  areas  that  can  be  further   studied  to  determine  the  impacts  of  EV  charging.  Further  study  would  gain  valuable   information  for  electricity  distribution  network  service  providers  in  planning  for  future   development,  as  well  determine  the  benefits  for  residents.   3.1 Load  Flow     3.1.1 Load-­‐Flow  Solutions     To  determine  loading  effects  in  the  context  of  an  Australian  residential  feeder,  load  flow   analysis  must  be  conducted.  A  simple  single-­‐line  diagram  can  be  realized  in  Fig.  3.1.       Figure  3.1:  Load  Flow  Analysis  [4]     Figure  3.1  represents  a  simple  power-­‐flow  scenario.  Power-­‐flow  problems  such  as  this   are  separated  in  to  the  following  components:   1. Slack  bus  –  a  reference  bus  for  which  V∠δ°  =  1.0∠0°   2. Load  (PQ)  bus  –   𝑃!  and   𝑄!  are  input  loads,  used  to  compute   𝑉!  and  δ!   3. Voltage  controlled  (PV)  bus  –   𝑃!  and   𝑉!  are  inputs,  includes  voltage  control   devices  such  as  OLTC,  switched  capacitors   The  power  flow  data  listed  is  used  to  calculate  power-­‐flow  solutions  using  methods  such   as  Guass-­‐Seidell  and  Newton-­‐Raphson,  which  solve  nodal  equations  iteratively  [7].     3.1.2 Load  Types   Another  important  consideration  that  must  be  made  during  load  flow  analysis  is  the   type  of  load  connected  to  each  load  bus.  Load  behaviour  is  determined  by  the  
14     14     combination  of  R,  L  and  C  elements  and  power  electronic  circuitry  of  a  load,  and  can  be   divided  into  three  types:   1. Constant  Power  (eg.  LED  TV,  computer)   2. Constant  Current  (eg.  CFL  globe)   3. Constant  Impedance  (eg.  Toaster,  oven)   Therefore,  for  any  given  voltage  a  load  will  conform  to  one  of  these  load  behaviours.  An   appliance  with  a  power  electronics  interface,  for  example,  with  exhibit  constant  power   characteristics  as  the  voltage  is  stepped  down  and  held  at  a  constant  DC  value,  as  this   voltage  will  be  constant  for  all  AC  source  voltage  levels.  A  resistive  load,  on  the  other   hand,  is  regarded  as  constant  impedance  and  will  draw  less  power  as  voltage  levels   drop,  according  to  Ohm’s  law.   The  voltage  dependency  of  loads  can  be  modelled  by  Eqs.    (3.1)  and  (3.2):     𝑃 = 𝑃!(𝑎𝑃 ∙ 𝑣 𝑣! !_!" + 𝑏𝑃 ∙ 𝑣 𝑣! !_!" + (1 − 𝑎𝑃 − 𝑏𝑃) ∙ 𝑣 𝑣! !_!" )   (3.1)     Where  1 − 𝑎𝑃 − 𝑏𝑃 = 𝑐𝑃     𝑄 = 𝑄!(𝑎𝑄 ∙ 𝑣 𝑣! !_!" + 𝑏𝑄 ∙ 𝑣 𝑣! !_!" + (1 − 𝑎𝑄 − 𝑏𝑄) ∙ 𝑣 𝑣! !_!" )   (3.2)     Where  1 − 𝑎𝑄 − 𝑏𝑄 = 𝑐𝑄     When  modelling  a  house  load,  a  number  of  assumptions  have  to  be  made.    For  the   purpose  of  this  simulation,  a  house  will  be  considered  as  a  constant  power  load,  as  each   house  will  be  associated  with  load  profiles  recorded  on  a  hot  day  where  the   predominant  load  type  is  a  constant  power  air  conditioner.  EVs  will  also  be  regarded  as   constant  power  loads,  as  the  charging  profile  of  a  lithium  ion  battery  charger  is  a   constant  power  curve.  From  Eq.  (3.1),  we  simply  require  𝑃 = 𝑃!,  therefore  all   coefficients  and  exponents  have  been  set  to  zero  in  the  voltage  dependence  settings  of   each  load.      
15     15     3.2 Modelling     A  realistic  network  model  is  imperative  for  determining  the  effects  of  EV  charging.   DIgSILENT  PowerFactory  was  chosen  for  this  purpose  due  to  its  flexibility  in  analysis,   incorporating  functions  such  as  unbalanced  power  flow,  and  remote  control  ability   through  DIgSILENT  Engine.  To  ensure  that  loading  results  were  as  accurate  as  possible,   emphasis  was  placed  on  applying  accurate  network  modelling  parameters,  load  profiles   and  vehicle  driving  statistics.   3.2.1 DIgSILENT  PowerFactory  Models       In  order  to  accurately  model  a  typical  low  voltage  network,  data  from  smart  meter-­‐ connected  premises  has  been  accumulated.  The  premises  of  interest  are  connected  to  a   500  kVA  pad-­‐mount  distribution  substation  in  Woodlands  Drive,  Glenmore  Park   (located  in  Western  Sydney),  which  supplies  92  premises  on  four  low  voltage   underground  feeders.  The  network  models  used  for  simulation  are  based  off  sample   DIgSILENT  feeder  models  provided  by  Endeavour  Energy.  Three  models  –  400  V   overhead,  400  V  underground  and  11  kV  overhead  –  were  modified  to  supply  the  same   number  of  loads  as  the  substations  in  Glenmore  Park.   When  implementing  the  LV  models,  each  premise  is  represented  by  a  single-­‐ phase  house  and  EV  load,  with  a  CSV  file  associated  with  each  load  containing  the  load   profile  information  for  a  single  day.  Due  to  limitations  with  the  number  of  possible   nodes  in  a  PowerFactory  student  license,  the  number  of  premises  has  been  halved  to  46   premises  split  across  two  feeders,  supplied  by  a  250  kVA  transformer.  Halving   transformer  ratings  and  feeder  numbers  ensures  an  accurately  scaled  model  for   determining  feeder  voltage  levels  and  transformer  loading.  Modelling  loads  as  single   phase  loads  allows  for  voltage  unbalance  to  be  accounted  for,  which  is  a  primary  cause   of  poor  voltage  regulation.  The  low  voltage  overhead  model  is  shown  in  Fig.  3.2.      
16     16       Figure  3.2:  400  V  Overhead/Underground  DIgSILENT  Model   To  model  the  impacts  of  electric  vehicle  charging  on  a  zone  substation  at  the  11  kV  level,   the  resulting  distribution  transformer  load  profiles  have  been  lumped  and  applied  to   each  of  the  transformer  loads  on  a  single  11  kV  feeder.  The  loading  magnitude  is   doubled  to  account  for  the  halved  number  of  premises  on  the  low  voltage  side,  so  that   each  transformer  is  represented  accurately  at  500  kVA.  There  are  10  11  kV  feeders   supplied  by  Glenmore  Park  Zone  Substation,  which  supplies  a  total  of  7596  premises.   Glenmore  Park  Zone  Substation  has  2  x  45  MVA  transformers  installed,  and   hence  has  an  N-­‐1  capacity  of  45  MVA.  With  an  average  of  760  premises  per  11  kV  feeder,   assuming  92  premises  per  500  kVA  of  installed  capacity,  there  would  be  an  average  of  8   LV  substations  connected  to  each  11  kV  feeder.  Therefore,  8  LV  substation  loads  have   been  modelled  on  the  11  kV  feeder,  and  the  total  zone  substation  load  is  determined  by   multiplying  the  total  feeder  loading  by  10  feeders.  Figure  3.3  shows  the  11  kV  feeder   model.  
17     17       Figure  3.3:  11  kV  Overhead  DIgSILENT  Model       Parameters  such  as  line  and  transformer  impedances,  shown  in  Table  3.1,  were  left   constant  as  they  represent  the  most  common  ratings  used  within  the  Endeavour  Energy   network.     400  V  Overhead   400  V  Underground   11  kV  Overhead   Feeder  Impedance     0.707  +  j0.284   Ω/km   0.162  +  j0.065  Ω/km   0.224  +  j0.224   Ω/km   Feeder  Section   Length     35  m   35  m   570  m   Service  Line   Impedance     1.49  +  j0.097  Ω/km   0.927  +  j0.081  Ω/km   N/A   Service  Line  Length   20  m   20  m   N/A   Transformer  Rating   250  kVA   250  kVA   N/A   Transformer   Impedance     4%   4%   N/A   Voltage  Source   Series  Impedance   0.5  +  j5  Ω     0.5  +  j5  Ω   0.021  +  j0.635  Ω   Table  3.1:  Network  Equipment  Parameters   The  11  kV  model  assumed  a  voltage  source  at  1  pu  voltage,  as  opposed  to  a  transformer,   as  the  transformer’s  OLTC  would  act  to  maintain  this  voltage  in  reality.  The  low  voltage   transformers  modelled  are  equipped  with  offline-­‐tap  changers  with  6  asymmetrical  tap   settings,  ranging  from  -­‐4  to  +1.  At  typical  tap  setting  for  LV  transformers  is  -­‐3,  or  -­‐7.5%,   corresponding  with  a  LV  bus  voltage  of  430  V.  An  increase  in  each  tap  setting  will  raise   the  voltage  by  2.5%,  allowing  for  a  12.5%  voltage  range  (-­‐10%  to  +2.5%).  As  LV  
18     18     transformer  taps  are  offline,  they  must  be  changed  manually  and  hence  would  only  be   changed  over  the  long  term  as  total  loading  increases,  not  in  response  to  a  permanent   increase  in  the  afternoon  peak  caused  by  EV  charging,  for  example,  as  this  would  cause   voltages  to  exceed  their  upper  limits  during  lower  loading  periods.  Instead,  this   regulation  must  be  controlled  using  zone  substation  OLTC’s  which  allow  for  real-­‐time   tap  changing.  As  EV  loading  is  expected  to  only  increase  the  afternoon/evening  peak,  the   tap  setting  is  expected  to  remain  constant  in  the  future.  Although  there  may  be  future   base  load  growth  as  the  penetration  of  air  conditioners  and  other  electrical  appliances   increases,  the  relative  difference  between  low  loading  periods  and  afternoon  EV  loading   will  likely  remain  constant,  therefore  the  actual  future  LV  substation  tap  setting  can  be   disregarded.   3.2.2 DIgSILENT  Programming  Language  (DPL)  Script     A  DIgSILENT  Programming  Language  (DPL)  script  allows  the  automation  of  load  flows   to  extract  specific  data  from  a  network  model.  A  DPL  script  was  provided  by  Endeavour   Energy  which  conducts  time-­‐step  simulation  load  flows  for  house  loads,  saving  power,   losses  and  voltage  data  into  result  objects  at  half  hour  intervals.  This  script  was   modified  to  read  EV  loads,  as  well  as  execute  ‘export  result  objects’  so  that  result  data   would  be  exported  to  text  files  each  time  the  script  was  run.  The  DPL  script  was   associated  with  each  network  model,  and  allowed  load  flow  simulations  to  be  conducted   via  engine  control  of  PowerFactory.     3.2.3 Load  Profiles     3.2.3.1 House  Load  Profiles     Loads  in  PowerFactory  can  be  associated  with  CSV  files  containing  multiple  time  points   for  conducting  time-­‐step  simulations.  Each  of  the  42  house  loads  has  an  associated  CSV   file  containing  the  smart  metering  data  of  a  premise  in  the  Glenmore  Park  trial  area,   chosen  at  random  from  the  92  metered  premises.  The  smart  metering  data  contains  the   power  usage  of  the  premises  over  a  24  hour  period  at  half  hour  intervals.  Each  premise   has  been  assigned  the  same  power  factor,  determined  as  the  average  of  the  premises   power  factor  during  the  evening  hours,  found  to  be  0.9  inductive.  The  selected  load   profiles  correspond  with  the  hottest  day  of  2011,  occurring  on  November  14  at  a  
19     19     maximum  temperature  of  38.7°C.  The  hottest  day  of  2011  was  chosen  as  network   planning  must  take  into  account  the  worst-­‐case  loading  scenarios  that  occur  on  hot  days,   caused  primarily  by  air  conditioners.     3.2.3.2 EV  Charging  Profiles     The  spread  of  EV  charging  start  times  were  determined  by  analysing  driving  statistics   from  the  NSW  Bureau  of  Transport  Statistics  [28],  shown  in  Fig  3.4.     Figure  3.4:  Average  number  of  travellers  in  NSW  on  weekdays  in  2010/11   This  graph  shows  the  average  number  of  travellers  in  NSW  on  weekdays  by  transport   type  in  2010/11.  For  determining  vehicle  arrival  times,  only  the  ‘Vehicle  Driver’  curve   was  considered.  The  time  of  arrival  was  determined  by  shifting  the  afternoon/night   peak,  between  2  pm  and  12  am,  by  20  minutes  -­‐  the  average  vehicle  one-­‐way  trip  time.   This  curve  was  then  normalised  between  2  pm  and  12  am,  and  multiplied  by  46  to   determine  the  number  of  premises  that  would  begin  charging  at  each  half  hour  interval   within  this  period.  The  resulting  scaled  driving  arrival  curve  is  shown  in  Fig  3.5,  shown   to  follow  the  afternoon  driving  trend  displayed  in  Fig  3.4.  
20     20       Figure  3.5:  Scaled  driver  arrival  times   The  number  of  vehicles  arriving  at  each  half  hour  interval  was  recorded,  and  the   vehicles,  having  been  assigned  their  specific  starting  time,  were  allocated  to  premises   using  a  random  function,  so  that  the  feeder  models  were  assigned  a  realistic  variation  in   vehicle  arrival  times.     3.2.4 Loading  Assumptions       To  model  the  effects  of  charging,  the  level  two  residential  chargers  from  Chapter  2  were   considered.  Considering  the  expected  combination  of  chargers  based  on  EV  costs,  an   average  charge  rating  of  4  kW  was  determined  to  provide  a  realistic  charging  power  that   could  be  used  to  model  a  load  of  EV  charging  homes.  The  average  battery  capacity  was   chosen  to  be  25  kWh,  a  mid-­‐range  capacity  in  Table  2.1.   Assuming  a  return  trip  driving  distance  of  18.8  km  [28]  and  a  battery  consumption  of   0.168  kWh/km  [16],  the  average  charging  time  was  found  to  be  approximately  47   minutes.  Due  to  the  time-­‐step  resolution  of  half  an  hour,  however,  this  charging  duration   had  to  be  modelled  as  1  hour.  This  analysis  assumes  that  each  EV  is  charged  only  once   per  day  in  the  afternoon/evening,  and  that  driving  is  split  into  a  morning  and  afternoon   peak.  Vehicles  arriving  home  during  the  late  night  hours  are  probably  drivers  that  have   travelled  previously  during  the  day,  so  charging  has  been  assumed  to  occur  after  the   second  trip.  Vehicle  driving  patterns  have  been  based  on  weekday  statistics,  and  the   vehicles  are  assumed  to  charge  on  a  daily  basis.   In  terms  of  vehicle  penetration,  a  substation  EV  penetration  of  100%   corresponds  to  all  vehicles  being  EVs,  not  100%  of  premises  containing  an  EV.  As  there   is  an  average  of  1.7  motor  vehicles  per  household  in  Australia  [29],  a  penetration  of  59%   would  represent  an  average  of  1  vehicle  per  household.   Another  consideration  made  was  the  percentage  of  travellers  that  drive  vehicles,   as  opposed  to  using  public  transport  or  travelling  as  a  passenger.  Although  we  know  
21     21     that  there  is  an  average  of  1.7  vehicles  per  household,  and  that  47%  of  travellers  drive  a   vehicle  [28],  it  is  impossible  to  discern  the  percentage  of  vehicle  owners  that  drive  a   vehicle  for  the  majority  of  their  travel  during  weekdays.  This  is  because  the  number  of   travellers  includes  school  students,  for  example,  who  may  travel  as  a  passenger  or  on   public  transport,  as  well  as  those  who  own  a  vehicle  but  may  cycle  or  also  use  public   transport  to  travel  to  work.  To  further  complicate  any  assumptions  made,  there  is  no   information  relating  to  the  percentage  of  people  that  actually  travel  significant  distances   during  the  week,  including  the  considerable  proportion  of  vehicle  owners  that  fall  into   this  category  such  as  pensioners  and  those  who  work  or  care  for  children  at  home.   Therefore,  with  the  data  available,  the  most  realistic  assumptions  decided  were   that  every  vehicle  owner  travels  the  average  distance  of  20  km  return-­‐trip  on  weekdays   and  does  the  majority  of  this  travel  in  their  vehicle.  Although  analysis    may  seem  more   accurate  to  apply  a  statistical  spread  of  charger  ratings  across  each  household,  this   would  be  equivalent  to  assuming  an  average  charger  rating  for  each  household,  as  the   total  transformer  loading  would  be  the  same.  A  statistical  variation  in  charger  ratings   would  provide  a  more  accurate  model  of  voltage  regulation,  however  the  limited   number  of  premises  in  the  DIgSILENT  models  prevents  any  statistical  analysis  from   yielding  meaningful  results.  Therefore,  Eqn.  (3.3)  has  been  used  to  determine  the   charging  power  per  premise.     P =  Charger  Rating  (kW)  ∗  (Penetration/100%)  ∗  1.7  vehicles  per  premise   (3.3)     The  assumptions  made  in  this  analysis  present  an  ambiguity  issue  with  the  number  of   drivers  arriving  home  during  the  middle  of  the  day,  and  those  that  may  travel  after   arriving  home  from  work.  The  actual  number  of  drivers,  however,  is  impossible  to   predict  without  conducting  a  large-­‐scale  survey  focusing  on  the  actual  arrival  times  and   driving  patterns  of  vehicle  drivers,  therefore  the  assumptions  made  can  be  considered   as  accurate  as  possible.       3.2.5 Load  Scaling   The  load  profiles  of  premises  supplied  by  the  Woodlands  Drive  substation  represent  the   energy  use  of  premises  in  a  sample  area  of  Glenmore  Park.  These  profiles  provide  an   accurate  load  shape,  however  their  combined  substation  profile  may  not  match  the   magnitude  of  those  substations  located  in  areas  of  lower  or  higher  socio-­‐economic  
22     22     status,  such  as  a  wealthier  area  which  is  more  likely  to  contain  a  greater  number  of  air   conditioners  and  pool  pumps,  for  example.  To  account  for  the  diversity  between  areas   within  suburbs,  it  is  important  that  the  Woodlands  Drive  substation  load  profile  can  be   scaled  before  EV  loading  is  added,  however  non-­‐linear  line  losses  must  also  be   accounted,  therefore  this  scaling  is  not  a  straight  forward  calculation.   As  base  loading  power  increases  linearly,  represented  by  ∆ 𝑃!!"#  in  per  unit,  line  losses   increase  by  the  square  of  this  rate,  or  (∆𝑃!"#$)! .  Therefore,  if  Woodlands  Drive   substation  is  80%  loaded  under  maximum  load,  this  load  profile  cannot  be  scaled  to   represent  substation  that  is  90%  loaded,  for  example,  without  first  separating  the   combined  house  power  and  the  line  losses.   This  would  require  a  scaling  model  in  the  following  form:     𝑃!" = 𝑃! 𝑥 + 𝑃! 𝑥!   (3.4)     Where   𝑃!"  is  the  new  total  power  drawn  by  the  transformer  after  scaling,   𝑃!is  the   combined  house  power  before  scaling,   𝑃!  is  the  line  losses  before  scaling.  For  example,  if   the  total  transformer  loading  was  required  to  be  increased  from  110  kW,  where   𝑃!  =   100  kW  and   𝑃!=  10  kW,  to  240  kW,  the  combined  house  power  would  only  have  to  be   increased  by  a  factor  of  x  =  2,  to  produce  a  transformer  power  increase  of     !"# !!"  =  2.18  pu.   This  formula,  however,  does  not  take  into  account  the  line-­‐loss  increase  as  a   result  of  the  voltage  drop  that  occurs  when  constant  power  loads  are  scaled.  That  is,   when  the  power  consumption  of  a  feeder  with  constant  power  loading  increases,  so  too   does  the  voltage  drop  along  the  feeder,  causing  the  line  current,  and  hence  line  losses,  to   rise  further.  This  is  a  cyclical  response  that  converges  rapidly  due  to  the  large  difference   between  the  percentage  change  in  voltage  and  the  initial  load  power  change,  therefore   any  further  voltage  correction  can  be  considered  negligible.   Figure  3.5  shows  the  voltage  profile  of  a  typical  feeder  with  6  premises  per  phase   per  feeder  in  the  upper  graph,  approximately  the  same  as  the  Woodlands  Drive  feeders,   and  the  profile  of  the  last  4  premises  on  a  feeder  in  the  lower  graph,  with  the  voltage   levels  scaled  for  an  easier  interpretation  of  the  voltage  drop  in  each  feeder  section.    
23     23       Figure  3.6:  Feeder  voltage  profile,  moving  from  last  premise  to  transformer  from  right  to  left   This  feeder  shows,  when  moving  towards  the  transformer  from  right  to  left,  the  voltage   drop  increases  according  to  the  series   𝑉!"(1,  3,  6,  10)  etc.,  where   𝑉!"    is  the  voltage  drop   along  the  last  section  of  feeder,  between  the  second  last  and  last  premises.  This  series   can  be  represented  by  Eq.  (3.5).     𝑖 ! !!! = 𝑛(𝑛 + 1) 2   (3.5)     Voltage  drop  in  the  last  section  of  feeder  can  be  found  using  Eq.  (3.6).     𝑉!" = 2𝑉!" 𝑛(𝑛 + 1)   (3.6)     Where   𝑉!"  is  the  voltage  drop  along  the  entire  feeder.   For  the  feeder  in  the  top  subplot  with  6  premises,  the  total  voltage  drop  is  equal  to   !(!!!) ! 𝑉!" = 21 0.004 = 0.084  pu,  which  is  reflected  on  this  plot.  This  modelling   assumes  that  each  load  draws  the  same  current.   When  a  load  is  increased  by  a  factor   𝑥,  the  line  current  supplying  a  constant   power  load  increases  by  the  same  factor.  As  voltage  drop  is  proportional  to  current,  the   voltage  drop  also  increases  by  this  factor.  When  considering  the  last  section  of  feeder,   𝑉!"#$  is  equal  to   𝑉!"# − 𝑉!"#,  where   𝑉!"#  is  the  voltage  at  the  second  last  premise  and  
24     24     𝑉!"#  is  the  voltage  at  the  last  premise.   𝑉!"#$  can  be  represented  as  a  percentage  by  Eq.   (3.7).     𝑥𝑉!"#$ − 𝑉!"#$ 𝑉!"# = 𝑉!"#$(𝑥 − 1) 𝑉!"#   (3.7)     As  the  load  current  increase  is  directly  proportional  to  the  voltage  drop,  the  line  current   is  increased  by  the  factor  that  is  Eq.  (3.8).     ∆𝐼!! = 1 + 𝑉!"#$(𝑥 − 1) 𝑉!"#   (3.8)     We  now  know  the  factor  that  can  be  squared  to  scale  the  power  losses  in  the  last  section   of  feeder.  As  power  losses  increase  with  the  square  of  the  line  current,  the   𝑛!  series  can   be  used  to  represent  the  increase  in  power  moving  from  the  last  section  of  feeder  to  the   first,  i.e.   𝑃! = 𝑃!(1 + 2 + 4 + 9 + 16 + 25)  for  a  feeder  with  6  premises  per  phase.  The   sum  of  the     𝑛!  series  is  given  by  Eq.  (3.9):     𝑖! ! !!! = 𝑛(𝑛 + 1)(2𝑛 + 1) 6   (3.9)     Therefore,  if  the  total  line  losses  of  a  feeder  are  known,  the  line  losses  can  be  given  by   dividing  the  total  power  by  the  sum  of  the   𝑛!  series,  equal  to  91  for  6  premises.  Once  we   know  the  losses  and  voltage  drop  in  the  last  section  of  feeder,  and  the  scaling  factor   𝑥  by   which  the  house  loads  increase  by,  the  increase  in  line  losses  due  to  the  load  scaling  and   increased  voltage  drop  can  be  determined.  The  power  increase  ∆ 𝑃!!in  the  last  section  of   feeder  therefore  becomes  ∆ 𝑃!!∆𝐼!! ! ,  where  ∆ 𝐼!! !  is  the  increase  in  current  due  to  the   voltage  drop.  The  total  increase  in  power  due  to  voltage  drop  is  shown  in  Eq.  (3.10).   ∆𝑃! = ∆𝑃!!∆𝐼!! ! + 4∆𝑃!!2∆𝐼!! ! 9∆𝑃!!3∆𝐼!! ! + 16∆𝑃!!4∆𝐼!! ! + 25∆𝑃!!5∆𝐼!! ! + 36∆𝑃!!6∆𝐼!! !                    = ∆𝑃!!∆𝐼!! ! (1 + 8 + 27 + 64 + 125 + 216)       (3.10)      
25     25       This  series  represents  the  sum  of  cubes,  which  can  be  expressed  in  the  following  general   equation  form  of  Eq.  (3.11).     𝑖! ! !!! = 𝑛! (𝑛 + 1)! 4   (3.11)     Combining  Eqs.  (3.8),  (3.9)  and  (3.11),  a  general  solution  of  Eq.  (3.12)  can  be  formed  for   line  losses.     𝑃!.!"# = 6𝑃! 𝑥! 𝑛 𝑛 + 1 2𝑛 + 1 ∙ 1 + 𝑉!" 𝑥 − 1 𝑉!"# ! ∙ 𝑛! 𝑛 + 1 ! 4                            = 6𝑃! 𝑥! 𝑛 𝑛 + 1 2𝑛 + 1 ∙ 1 + 2𝑉!" 𝑥 − 1 𝑛(𝑛 + 1)𝑉!"# ! ∙ 𝑛! 𝑛 + 1 ! 4               (3.12)     Replacing  the     𝑃! 𝑥!  term  in  Eq.  (3.4)  with  Eq.  (3.12),  the  complete  transformer  power   formula  Eq.  (3.13)  is  formed.   𝑆!" = 𝑃! 𝑥 + 6𝑃! 𝑥! 𝑛 𝑛 + 1 2𝑛 + 1 ∙ 1 + 2𝑉!" 𝑥 − 1 𝑛(𝑛 + 1)𝑉!"# ! ∙ 𝑛! 𝑛 + 1 ! 4 + 𝑗𝑄! 𝑥       (3.13)       Equation  (3.13)  allows  the  apparent  transformer  power  to  be  determined  when   constant  power  house  loads  are  scaled  by  a  value  𝑥,  taking  into  account  the  non-­‐linear   nature  of  line  losses  caused  by  load  scaling  and  the  subsequent  voltage  drop.  Equation   (3.13)  assumes  that  all  houses  are  loaded  equally,  and  reactive  power  is  constant.  In   reality,  reactive  power  will  increase  slightly  in  response  to  voltage  drop,  depending  on   the  characteristics  of  the  load.  This  equation,  however,  represents  a  relatively  accurate   model  and  provides  an  insight  into  the  complexity  of  load  behaviour,  and  hence  line   losses,  in  response  to  a  change  in  load  magnitude.   Due  to  the  complexity  of  this  4th  degree  polynomial,  solving  for   𝑥  is  difficult,   therefore  loads  will  be  scaled  through  trial  and  error  for  scenarios  where  the   transformer  is  at  a  higher  capacity  than  the  Woodlands  Drive  substation.    
26     26     3.3 Simulation     3.3.1 Graphical  User  Interface     A  graphical  user  interface  is  a  productive  tool  that  allows  the  simple  selection  and   presentation  of  modelling  variables  and  results  in  a  single  window.  An  interface  such  as   a  GUI  is  especially  useful  for  simulating  the  effects  of  EV  charging  on  a  distribution   network,  as  each  EV  load  can  be  automatically  modified  for  all  combinations  of  charging   penetration,  charging  coordination,  charger  rating  and  network  type.  Also,  results  of   interest  can  be  displayed  both  graphically  and  numerically  for  ease  of  analysis,  such  as   lowest  bus  voltages  and  maximum  transformer  loading.   MATLAB  GUIDE,  an  open  GUI  layout  editor,  was  chosen  to  create  an  interface   with  the  DIgSILENT  models  for  this  thesis.  This  GUI,  shown  in  Fig.  3.7,  was  created  to   run  multiple  EV  loading  scenarios  without  the  need  to  manually  change  the  load  profiles   of  each  load.  This  tool  was  used  to  determine  the  loading  results  of  this  thesis,  and  will   be  of  use  to  network  planners  at  Endeavour  Energy  for  forecasting  future  load  demand.         Figure  3.7:  MATLAB  GUI   MATLAB  GUIDE  contains  a  number  of  interactive  controls  such  as  buttons  and  sliders   that  can  be  used  to  create  custom  GUIs.  Each  control  element  has  an  associated  callback   function  that  is  called  when  a  user  presses  or  changes  an  element.  These  callback  
27     27     functions  are  contained  in  an  associated  MATLAB  file,  used  as  the  master  script  for   running  simulations.     3.3.2 GUI  Structure     In  order  to  run  a  simulation  using  the  GUI,  the  network  variables  must  first  be  selected.   Table  3.2  shows  the  possible  variable  options  that  can  be  selected.   Feeder  Type   Overhead     Underground   Charging  Coordination   Uncoordinated     Staggered  10  pm  start   Electric  Vehicle  Penetration   0  to  100  %   Average  Charger  Rating   2  to  10  kW   Table  3.2:  Variable  Options  Structure   Each  time  a  variable  is  selected,  the  callback  of  each  pop-­‐up  menu  or  slider  is  called,  and   the  variable  name  is  stored.  A  flowchart  visualising  the  function  calls  within  the  GUI   script  is  shown  in  Fig.  3.8.     Figure  3.8:  Flowchart  displaying  the  interaction  of  programs  required  for  GUI  simulations  
28     28     The  breakdown  of  this  flowchart  is  as  follows,  beginning  with  the  execution  of  the  GUI   process  after  pressing  ‘Run  Simulation’.   1. The  ‘Run  Simulation’  button’s  callback  function  determines  the  combination  of   variables  selected  using  a  series  of  IF  statements.   2. The  penetration  and  charger  rating  information  is  used  to  call  a  function  that   writes  the  load  profile  to  CSVs  associated  with  the  EV  loads  in  the  LV  DIgSILENT   models.  For  example,  an  average  charger  rating  of  4  kW  at  50%  penetration  will   result  in  an  EV  load  of  2  kW  being  applied  to  each  EV  load  using  the  ‘csvwrite’   function.  The  time  of  this  charging  is  dependent  on  the  charging  coordination   strategy  (explained  in  Section  3.2.4).   3. A  MATLAB  function  changes  the  working  directory  and  calls  a  batch  file.     4. The  called  batch  file  is  used  to  open  a  connection  to  DIgSILENT  Engine,  the   engine  running  behind  DIgSILENT  PowerFactory,  and  call  a  windows  command   file  containing  the  instructions  required  to  open  and  simulate  a  DIgSILENT   model.  When  a  network  is  selected,  a  batch  file  is  called  containing  the  line   ‘digrcom  -­‐d  -­‐p  ncacn_ip_tcp  -­‐n  127.0.0.1  -­‐e  2001  -­‐f="command.cmd"  ‘.  The   windows  command  file  ‘command.cmd’  contains  the  following  lines  that  are   interpreted  by  DIgSILENT  Engine:                    ac  UserProject                  cd  UserProjectLibraryScripts                  Script  Name   This  activates  the  LV  overhead  project  and  calls  the  DPL  script,  which  performs  a   time-­‐step  simulation  and  saves  the  resulting  network  power  and  voltages  to   result  objects.   5. The  DPL  script  then  executes  result  objects  which  write  the  resulting  network   power  and  voltages  to  text  files.   6. A  MATLAB  function  calls  a  windows  command  file  to  convert  the  result  text  files   to  CSVs.  This  is  required  so  that  MATLAB  can  import  the  results  into  arrays.   7. The  command  runs  using  the  command  ‘rename’  to  change  the  result  text  files  to   CSV  files.       8. As  the  data  is  imported  into  a  single  column,  the  text-­‐to-­‐columns  function  must   be  run  to  separate  the  data  into  individual  columns.  Also,  cells  containing  text   and  zeros  must  be  removed  so  that  the  data  can  be  imported  and  analysed.  This   must  be  done  using  an  Excel  VBA  macro,  however  the  macro  had  to  be  saved  in  
29     29     an  excel  add-­‐in  so  that  it  can  be  used  for  multiple  CSV  workbooks.    A  VBScript  file   is  therefore  used  to  open  the  Excel  process,  open  the  voltage  results,  power   results  and  add-­‐in  files,  and  call  the  macro  so  that  the  data  can  be  read  by   MATLAB.  This  VBScript  file  is  called  by  a  windows  command,  which  also  deletes   the  previous  CSV  files  each  time  a  simulation  is  run.   9.  A  MATLAB  function  imports  the  voltage  data  as  a  single  array,  and  determines   the  minimum  bus  voltage.  The  function  then  imports  the  first  two  columns  of  the   power  data,  corresponding  to  the  real  and  reactive  transformer  power  data.  The   maximum  apparent  transformer  power  is  then  determined  from  the  real  and   reactive  power  arrays.  The  row  location  of  the  minimum  voltage  and  maximum   power  values  are  also  saved  and  inputted  into  a  function  that  converts  this  value   into  a  time  string.  The  transformer  load  profile  is  then  written  to  a  single  CSV  file   associated  with  each  load  in  the  11  kV  model.  The  process  is  then  repeated  to   extract  the  11  kV  voltage  and  transformer  data.   10. The  minimum  voltage,  maximum  power  and  the  time  these  values  occur  are  then   displayed  on  a  graph  on  the  GUI,  with  the  400  V  transformer  data  displayed  by   default.  The  user  can  then  change  between  transformer  and  voltage  results,  and   400  V  and  11  kV  results.     The  GUI  script  is  therefore  used  to  call  MATLAB  functions  and  external  batch,   command  and  VBScript  files  to  control  MATLAB,  DIgSILENT  and  Excel.  This  allows   simulation  scenarios  to  be  generated  and  presented  within  a  single  GUI  window.   3.4 Scenarios   3.4.1  Uncoordinated  Charging     To  determine  the  impacts  of  EV  charging  on  low  voltage  networks,  analysis  was   conducted  on  the  Woodlands  Drive  substation,  as  well  as  theoretically  loaded   substations  that  may  represent  other  areas.  The  Woodlands  Drive  substation  base  load   was  scaled  through  trial  and  error,  considering  the  analysis  conducted  in  Section  4.1.2,   so  that  the  maximum  loading  without  added  EV  loads  corresponded  with  80,  85,  90  and   95%  of  the  substations  500  kVA  capacity.      
30     30     The  following  variables  were  considered  in  this  analysis:   1.  Network  type     2.  Average  charger  rating     3.  Charging  time   4.  Phase  balance   Both  underground  and  overhead  LV  networks  were  modelled  to  determine  the   difference  in  EV  loading  caused  by  differences  in  line  impedances.  Average  charger   ratings  of  4,  7  and  10  kW  were  considered,  with  4  kW  the  most  probable  average   charger  rating  expected  in  the  coming  years.  Charging  times  were  found  to  be  19,  27  and   47  minutes  for  4,  7  and  10  kW  chargers  respectively,  using  statistics  from  Section  3.2.4,   modelled  as  half  hour  and  one  hour  charging  periods  due  to  a  half  hour  time-­‐step   resolution  limit.  Scenarios  were  also  simulated  where  this  charging  time  may  be   doubled,  useful  for  representing  areas  where  the  average  driving  distance  may  be   greater  than  the  NSW  average.  Phase  balance  was  the  last  variable  considered,   considering  the  expected  outcome  of  an  unbalanced  network  using  the  provided  smart   metering  data,  and  the  unlikely  scenario  where  a  feeder  may  be  close  to  perfectly   balanced.     All  combinations  of  these  variables  were  modelled  using  MATLAB  functions   through  DIgSILENT  engine.  This  analysis  ensured  that  all  possible  scenarios  were   covered  so  that  results  were  as  accurate  as  possible.   3.4.2 Coordinated  Charging     To  simulate  coordinated  charging,  a  simple  off-­‐peak  staggered  charging  method  was   chosen.  EVs  were  first  split  into  three  groups,  assigned  evenly  across  three  phases  to   minimise  voltage  drop,  and  assigned  a  starting  time  of  10  pm,  11  pm  or  12  am.    The   maximum  EV  penetration  was  determined  for  the  same  variables  considered  in  the   uncoordinated  charging  analysis,  except  for  voltage  unbalance  as  this  was  not  an  issue   during  the  late-­‐night  hours  for  the  smart-­‐metered  houses.     Next,  the  number  of  start  times  was  doubled  to  six,  so  that  EVs  were  assigned  a   start  time  ranging  from  10  pm  to  3  am.  This  allowed  the  EV  loading  at  each  hour  to  be   halved,  while  allowing  a  suitable  amount  of  time  for  drivers  who  may  require  their  car   early  in  the  morning.  
31     31     An  issue  surrounding  the  analysis  of  the  hottest  day  was  the  insignificant  off-­‐ peak  hot  water  loading  during  the  late  night  hours.  This  may  have  been  due  to  a  large   proportion  of  gas/solar  hot  water  systems  in  this  area,  a  reduction  in  showering  times   and  temperatures  due  to  the  hot  weather,  or  both.  Electric  hot  water  systems  are   typically  rated  at  3.6  kW,  drawing  power  as  a  constant  impedance  load,  therefore  the   average  power  consumption  would  be  less  than  this  with  additional  loads  reducing   voltages.  Therefore,  to  consider  an  area  with  all  premises  connected  to  electric  off-­‐peak   water  heating,  the  7  kW  EV  charging  scenario  would  provide  a  better  analysis  of  a   worst-­‐case  scenario  for  hot  water  heating  and  4kW  charging.   3.4.3 11  kV   The  analysis  of  400  V  feeders  provides  an  insight  into  the  effects  of  loading  on  LV   substations,  providing  an  idea  of  the  loading  at  the  zone  substation  level.  A  zone   substation,  however,  is  typically  designed  with  a  greater  focus  on  future  growth  in  the   number  of  loads,  compared  to  a  sole  consideration  on  the  average  power  increase  of   each  load,  as  the  area  covered  by  a  zone  substation  is  significant  and  determined  by   geographical  and  financial  considerations.   To  determine  the  loading  effects  of  EV  charging  on  Glenmore  Park  zone   substation,  the  hottest  day  of  the  2010/11  period  was  selected,  where  capacity  reached   43.5  MVA  of  the  substation’s  45  MVA  N-­‐1  capacity.  This  day’s  loading  was  significantly   greater  than  the  hottest  day  of  2011/12,  chosen  for  modelling  with  the  Woodlands  Drive   smart  metering  data,  where  the  maximum  zone  substation  loading  reached  29  MVA   (however,  this  day  was  the  highest  loaded  day  recorded  by  the  smart  meters  as  they   hadn’t  been  installed  before  the  hottest  day  of  2010/11).   As  Glenmore  Park  ZS  supplies  7596  premises  on  10  11kV  feeders,  there  can  be   assumed  an  average  of  760  premises  per  feeder.  Assuming  92  premises  are  assigned  to   each  500  kVA  of  installed  capacity,  there  would  be  an  average  of  8  transformers  per  11   kV  feeder.  Dividing  the  45  MVA  capacity  by  10  feeders  and  8  transformers,  however,   results  in  an  average  transformer  capacity  of  550  kVA.  This  can  be  explained  by  the   variations  in  socio-­‐economic  status,  gas  supply  availability  and  the  percentage  of   commercial  and  industrial  customers,  which  can  be  disregarded  for  EV  loading  based  on   assumptions  made  in  Section  3.2.4.   To  simulate  the  zone  substation  at  43.5  MVA,  or  96.67%  capacity,  the  400  V   model  was  modified  to  be  on  a  275  kVA  base  (half  of  550  kVA  due  to  node  limitations),  
32     32     and  the  base  house  loads  were  scaled  so  that,  once  lumped  into  the  11  kV  model  (and   doubled  to  represent  550  kVA  transformers),  they  produced  a  feeder  loading  of  4.35   MVA  in  the  11  kV  model  (representing  1/10th  of  the  total  ZS  loading).  A  load  profile  for   both  real  and  reactive  power  was  lumped  into  the  11  kV  model  to  account  for  the  change   in  power  factor  as  EVs  were  added.  The  base  loading  scaling  in  the  400  V  model  was   completed  through  trial  and  error  due  to  the  non-­‐linear  losses  in  both  the  400  V  and  11   kV  networks.     Next,  EV  penetration  was  increased  until  the  11  kV  feeder  loading  reached  4.5   MVA,  marking  100%  transformer  loading.  This  was  completed  for  4,  7  and  10  kW   chargers  for  both  the  expected  driving  distance  of  20  km  and  the  case  that  this  distance   was  doubled.                                                                      
33     33         4 Results     The  implementation  of  a  GUI  has  allowed  for  a  number  of  loading  scenarios  to  be   simulated  and  compared  to  accurately  determine  the  effects  of  EV  charging  on   residential  distribution  feeders.     4.1 Base  Load  Profile   The  base  load  profile,  comprised  of  actual  premise  smart  metering  data,  is  an  important   profile  as  it  forms  the  basis  of  all  EV  loading  analysis.   4.1.1 Effects  of  Temperature  on  Substation  Loading   The  loading  significance  due  to  air  conditioners  can  be  seen  in  Fig.  4.1.  Figure  4.1   compares  the  substation  profile  on  a  38.7  degree  day  compared  to  a  mild  19.9  degree   day,  both  days  being  weekdays.     Figure  4.1:  Woodlands  Drive  substation  loading  for  38.7  and  19.9  degrees  celsius  days   Figure  4.1  clearly  shows  a  substantial  difference  in  loading  between  the  hot  and  mild   day,  with  the  hottest  day  loading  162%  greater  than  the  mild  temperature  day.  This   comparison  confirms  the  need  to  consider  the  hottest  days  of  the  year  when  planning   for  worst-­‐case  loading  scenarios.   4.1.2 Load  Scaling   Another  factor  contributing  to  an  appropriate  base  load  profile  is  the  scaling  of  the   modelled  transformer  power.  As  the  modelled  premises  were  a  50%  sample  of  the  total   number  of  distribution  substation  premises,  the  load  profile  for  this  sample  would  be   desired  to  reflect  the  load  profile  of  the  entire  sample  set  to  ensure  accuracy.  Figure  4.2  
34     34     shows  the  modelled  curve  of  46  premises  once  the  load  profile  magnitude  had  been   doubled,  and  the  actual  substation  meter  load  profile  supplying  92  premises.       Figure  4.2:  Woodlands  Drive  substation  total  load  compared  to  scaled  sample  loads   Figure  4.2  shows  that  the  modelled  sample  (red)  closely  follows  the  actual  substation   profile,  confirming  the  sample  contains  no  significant  outliers  that  would  have  skewed   the  modelled  results  after  scaling.  Although  a  noticeable  loading  difference  exists  at   around  7:30  pm,  the  load  peaks  at  6:00  pm  are  close  to  the  same  value,  this  being  the   most  important  time  as  network  loading  is  greatest  and  during  the  expected  worst   uncoordinated  charging  period.   4.1.3 Network  Type   The  difference  in  transformer  loading  between  underground  and  overhead  low  voltage   networks  becomes  apparent  in  Fig  4.3.       Figure  4.3:  Woodlands  Drive  substation  load  for  overhead  and  underground  networks   As  all  loads  have  been  measured  as  constant  power  loads,  load  currents  must  increase   as  voltages  decrease  down  the  length  of  a  feeder.  As  overhead  lines  typically  have  an   impedance  greater  than  underground  cables,  the  voltage  drop,  and  hence  line  currents,  
35     35     will  be  greater  in  an  overhead  network,  resulting  in  higher  losses  which  must  be   supplied  by  the  transformer.     During  the  maximum  demand  period  the  maximum  power  drawn  is  274  kVA,   representing  56%  of  the  substations  500  kVA  total  capacity.   4.2 Uncoordinated  Charging   Simulations  were  run  for  penetrations  levels  up  to  100%  for  both  overhead  and   underground  low  voltage  networks.  For  overhead  LV  feeders,  a  tap  setting  of  -­‐4  (+10%)   was  found  to  allow  secondary  LV  substation  voltages  to  remain  closest  to  1.1  pu  for  all   EV  penetrations  during  peak  periods,  assuming  at  least  a  1  pu  primary  voltage.   Satisfactory  voltage  regulation  on  a  -­‐4  tap  setting  therefore  requires  the  11  kV  feeder  to   be  capable  of  maintaining  voltages  at  the  end  of  feeder  to  at  least  1  pu.  Hence,  it  is   important  to  determine  the  11  kV  voltage  regulation  capabilities,  to  determine  the   worst-­‐case  voltage  which  will  be  the  limiting  factor  for  LV  voltage  regulation.   4.2.1 11  kV  Voltage  Regulation     A  tap  setting  of  -­‐4  (+10%)  is  a  typical  maximum  tap  setting  for  LV  distribution   transformers.  This  tap  setting  has  been  selected  for  modelling  as  it  provides  the  highest   voltage  at  0%  EV  penetration  during  the  peak  loading  hours.  In  reality,  a  tap  setting  of  -­‐3   (+7.5%)  is  a  typical  setting  for  LV  transformers,  as  this  maintains  a  voltage  less  than  1.1   pu  for  premises  closest  to  the  transformer  during  periods  of  low  loading.     To  ensure  1.1  pu  is  set  at  the  secondary  LV  transformer  side  of  the  LV  overhead   network,  the  11  KV  feeder  voltage  must  be  capable  of  providing  at  least  1  pu  voltage   (exactly  1  pu  required  for  the  highest  tap  setting  of  -­‐4)  to  the  end  of  the  feeder  without   the  start  of  the  feeder  exceeding  the  upper  voltage  limit  or  OLTC  capability.   To  test  the  voltage  regulation  capabilities  of  a  typical  11  kV  feeder  under  worst-­‐case   conditions,  100%  EV  loading  was  applied  to  premises  on  LV  substations,  and  this   resulting  load  profile  was  then  lumped  into  LV  substations  in  the  11  kV  model.   The  11  kV  bus  voltage  was  found  to  be  to  be  1.045  pu  in  order  to  satisfy  the  1  pu  voltage   requirement  at  the  end  of  the  feeder,  confirming  11  kV  voltage  regulation  was  suitable   for  considering  EV  loading.    
36     36     4.2.2 400  V  Transformer  and  Feeder  Loading   When  modelling  400  V  scenarios,  the  overhead  network  LV  tap  changer  was  set  to  -­‐4   (+10%)  and  the  underground  LV  tap  changer  was  set  to  -­‐3  (+7.5%),  allowing   satisfactory  voltage  regulation  during  the  evening  hours.     4.2.2.1 Woodlands  Drive  Substation   The  impacts  of  charging  on  Woodlands  Drive  substation  was  determined  for  EV   penetrations  up  to  100%,  shown  in  Table  4.1.  The  results  in  Table  4.1  show  that  at  zero   percent  EV  penetration,  Woodlands  Drive  substation  is  significantly  under-­‐loaded  for   the  hottest  day  of  2011,  reaching  only  55%  capacity  as  it  is  located  in  an  underground   area.  At  500  kVA,  network  planners  have  allowed  for  5.5  kVA  per  customer,  which  is   typical  for  premises  in  a  medium  socio-­‐economic  area,  therefore  these  premises  must   use  less  energy  than  expected.       Overhead   Underground   Penetration   %   Loading   %   Loading   Increase   Min.  Bus   Voltage   %   Loading   %   Loading   Increase   Min.  Bus   Voltage   0%   56.36%     0.00%   1.006  pu   55.32%   0.00%   1.021  pu   5%   57.60%     2.20%   1.005  pu   56.48%   1.24%   1.020  pu   10%   58.84%     4.40%   1.005  pu   57.72%   2.55%   1.020  pu   15%   60.08%     6.60%   1.005  pu   58.92%   3.87%   1.020  pu   20%   61.36%     8.87%   1.005  pu   60.12%   5.11%   1.020  pu   25%   62.64%     11.14%   1.005  pu   61.36%   6.42%   1.019  pu   30%   63.92%     13.41%   1.005  pu   62.60%   7.74%   1.019  pu   35%   65.25%     15.76%   1.003  pu   63.84%   8.98%   1.019  pu   40%   66.56%     18.10%   0.999  pu   65.12%   10.92%   1.018  pu   45%   67.92%     20.51%   0.994  pu   66.36%   12.55%   1.016  pu   50%   69.28%     22.92%   0.989  pu   67.64%   14.96%   1.014  pu   55%   70.68%     25.41%   0.983  pu   68.92%   17.37%   1.012  pu   60%   72.04%     27.82%   0.978  pu   70.20%   19.78%   1.010  pu   65%   73.48%     30.38%   0.972  pu   71.48%   22.19%   1.009  pu   70%   74.88%     32.86%   0.966  pu   72.76%   24.60%   1.007  pu   75%   76.36%     35.48%   0.960  pu   74.08%   27.00%   1.005  pu   80%   77.80%     38.04%   0.956  pu   75.36%   29.42%   1.003  pu   85%   79.28%     40.67%   0.950  pu   76.68%   31.90%   1.001  pu   90%   80.76%     43.29%   0.943  pu   78.00%   34.31%   0.999  pu   95%   82.28%     45.99%   0.937  pu   79.32%   36.79%   0.997  pu   100%   83.80%     48.69%   0.931  pu   80.64%   39.20%   0.995  pu   Table  4.1:  Woodlands  Drive  substation  transformer  loading  and  voltage  regulation  for  varying  EV   penetrations   The  Woodlands  Drive  substation  loads  would  be  capable  of  supporting  90%  EV   penetration  in  an  overhead  network,  and  100%  in  an  underground  network  with  a  4  kW  
37     37     average  charger  rating.  Transformer  loading  would  be  greater  for  an  overhead  network   due  to  increased  losses  associated  with  higher  feeder  resistances  (1.49  ohms  overhead   compared  to  0.927  ohms  underground  for  the  modelled  cables).  The  impact  of   resistance  can  be  seen  at  EV  penetrations  above  90%  for  an  overhead  feeder,  where  the   minimum  bus  voltage  falls  below  the  lower  limit  of  0.94  pu.  At  this  penetration,  the   transformer  voltage  could  not  be  further  increased  to  compensate  for  the  worst-­‐bus   voltage,  therefore  marking  the  point  at  which  voltage  regulation  would  fail.     Upon  further  analysis,  poor  voltage  regulation  was  found  to  be  caused  by   significant  voltage  unbalance  between  phases,  and  loading  between  feeders.  Where   feeder  two’s  worst  bus  voltage  was  equal  to  0.931  pu  at  100%  EV  penetration,  the   lowest  bus  voltage  on  feeder  one  was  0.987,  as  this  feeder  was  both  more  balanced  and   lightly  loaded  that  feeder  two  after  random  premise  loading  allocation.  On  feeder  two,   voltages  ranged  from   𝑉!=0.931  pu,   𝑉!=1.08  pu  and   𝑉!=0.936  pu  at  the  three  premises   furthest  from  the  transformer,  compared  to   𝑉!=1.04  pu,   𝑉!=1.08  pu  and   𝑉!=1.036  pu  at   the  same  premises  with  no  EV  loading.  At  the  time  of  these  voltage  results,  6:30  pm,   three  EVs  were  charging  on  phases  a  and  c,  with  none  on  phase  b  after  random   allocation.  The  three  EV  loads  charging  on  each  phase  were  also  located  at  the  end  of  the   feeder,  further  contributing  to  the  problem,  resulting  in  an  extreme  loading  scenario   where  charging  occurred  only  on  the  heaviest  loaded  phases  at  the  end  of  the  feeder,   exceeding  voltage  limits  before  the  full  transformer  loading  capacity  was  reached  on  the   overhead  network.   4.2.2.2 Additional  Scenarios     Scenario  1:  4  kW  Charger   Table  4.2  shows  the  maximum  EV  penetration  before  transformer  capacity  is  reached,  or   voltage  regulation  has  failed,  denoted  by  a  (*).  Charging  is  divided  into  overhead  and   underground  networks,  subdivided  into  balanced  and  unbalanced  networks,  and  then   further  subdivided  into  two  charging  times,  where     𝑇!  is  the  expected  charging  time   based  on  an  average  20  km  return  trip,  equal  to  45  minutes  for  a  4  kW  charger   (modelled  as  one  hour).    2 𝑇!  is  equal  to  1.5  hours  in  this  case.  The  column  containing   data  for  an  unbalanced  network  with  charging  time   𝑇!  has  been  highlighted,  as  each   column  contains  the  most  likely  scenarios  based  on  research  in  this  thesis.    
38     38     Existing  Base   Transformer   Loading  on   Hot  Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Unbalanced   Balanced   Unbalanced   Balanced   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   Wood.  Dr.   90%*   80%*   90%*   79%*   100%   100%   100%   100%   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   58%*   38%*   54%*   42%*   77%   54%   76%   53%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   51*%   28%*   47%*   37%*   58%   41%   57%   40%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   39%   28%   39%   27%   38%   27%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   23%   21%   20%   14%   19%   13%   Table  4.2:  Maximum  EV  penetration  for  4  kW  LV  uncoordinated  charging   Each  row  represents  a  different  transformer  base  load  as  a  percentage  of  𝑆!"#$%,   equal  to  500  kVA.  The  Woodlands  Drive  substation  base  loading  is  equal  to  0.55 ∙ 𝑆!"#$%   for  an  underground  network  and  0.56 ∙ 𝑆!"#$%  for  overhead,  as  determined  in  Section   4.2.2.1.  The  remaining  rows  represent  theoretically  loaded  transformers  at  a  worst-­‐case   base  loading  of    0.80 ∙ 𝑆!"#$%  and  0.95 ∙ 𝑆!"#$%  during  the  evening  hours,  determined  by   scaling  the  Woodlands  Drive  smart  meter  loads.  The  degree  of  load  unbalance  is   therefore  the  same  for  all  unbalanced  loading  scenarios,  where  load  unbalance  refers  to   the  difference  between  load  magnitudes  across  phases  at  the  peak  loading  time  of  6  pm,   not  the  loading  unbalance  across  time.   Analysis  of  results  shows  a  stark  contrast  between  overhead  and  underground   networks.  The  higher  overhead  line  impedance  is  shown  to  significantly  reduce  the   maximum  EV  penetration  for  overhead  networks,  limiting  maximum  penetration  with   unbalanced  loads  on  an  80%  loaded  transformer  to  58%  overhead  compared  to  100%   for  an  underground  network,  for  example.  For  an  overhead  unbalanced  network,  scaling   the  base  load  above  85%  was  not  possible  without  exceeding  the  lower  voltage  limit  of   0.94  at  the  worst  bus,  due  to  load  unbalance  as  a  result  of  the  randomly  assigned  smart   meter  load  profiles.     A  noticeable  and  unexpected  relationship  occurred  between  balanced  and   unbalanced  networks,  as  balanced  networks  were  able  to  handle  less  added  EV  loads.   Upon  further  analysis,  the  cause  of  this  relationship  was  found  to  be  the  difference  in   losses  –  a  substation  loaded  at  80%  capacity  with  unbalanced  loads  during  the  evening   peak  would  have  a  greater  proportion  of  losses  than  a  balanced  substation  also  at  80%   capacity.  Hence  the  balanced  substation  could  be  regarded  as  more  efficiently  loaded,  as   the  actual  combined  house  loading  (both  real  and  reactive  power),  disregarding  line  
39     39     losses,  would  be  greater.  A  more  efficient  base  loading,  however,  leaves  less  room  for  EV   loads.     The  difference  in  maximum  EV  penetration  between  balanced  and  unbalanced   networks  is  considerably  less  in  an  underground  network,  with  differences  varying  by   1%  typically,  due  to  the  lower  losses.  The  same  relationship  was  found  between   underground  and  overhead  networks  when  both  were  balanced,  as  the  capacity  of  the   underground  network  dropped  to  less  than  that  of  the  overhead  network  due  to  the   lower  percentage  of  losses.  Lower  EV  penetration  for  balanced  underground  networks   only  occurred  for  the  90  and  95%  loaded  substations  where  maximum  EV  penetration   was  not  limited  by  voltage  regulation  in  the  overhead  network.   Table  4.2  also  clearly  shows  the  significant  reduction  in  possible  EV  loading  in  areas   where  travel  distance,  hence  charging  time,  would  be  twice  the  average,  for  example   100%  EV  penetration  compared  to  73%  for  an  80%  loaded  transformer  supplying  an   underground  network.     Scenario  2:  7kW     Existing  Base   Transformer   Loading  on   Hot  Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Unbalanced   Balanced   Unbalanced   Balanced   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   Wood.  Dr.   53%*   52%*   57%*   52%*   100%   93%*   100%   83%   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   33%*   33%*   38%*   31%*   77%   44%   75%   43%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   29%*   29%*   33%*   27%*   59%   33%   57%   32%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   26%*   22%   39%   22%   38%   21%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   20%*   13%   20%   11%   19%   10%   Table  4.3:  Maximum  EV  penetration  for  7  kW  LV  uncoordinated  charging   At  7  kW,  the  charging  time   𝑇!  reduces  to  half  an  hour.  Maximum  penetration  at   Woodlands  Drive  substation  was  found  to  reduce  from  90%  to  53%  compared  to  a  4  kW   charger,  while  a  penetration  of  100%  was  still  possible  in  an  unbalanced  underground   network.  This  result  draws  further  attention  to  the  unbalance  across  phases  caused  by   the  EVs,  as  there  is  a  greater  difference  between  the  overhead  and  underground   networks  where  overhead  penetration  is  limited  by  voltage  regulation.     The  change  in   𝑇!from  one  hour  for  4  kW  chargers  to  half  an  hour  resulted  in   different  combination  of  vehicles  charging  at  the  same  time,  where  the  same  number  of   vehicles  were  charging  for  both  charging  times.  This  resulted  in  voltage  regulation  
40     40     failing  at  a  penetration  lower  than  that  of  the  balanced  network,  which  contrasts  the   results  of  the  4  kW  scenario.       Scenario  3:  10  kW     Existing  Base   Transformer   Loading  on  Hot   Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Unbalanced   Balanced   Unbalanced   Balanced   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   Wood.  Dr.   37%*   37%*   39%*   36%*   100%   65%*   85%*   58%*   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   23%*   23%*   26%*   22%*   54%   31%   52%*   30%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   20%*   20%*   23%*   19%*   41%   23%   40%   22%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   18%*   16%   27%   15%   26%   15%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   14%*   9%   14%   8%   13%   7%   Table  4.4:  Maximum  EV  penetration  for  10  kW  LV  uncoordinated  charging   At  10  kW,   𝑇!  was  equal  to  20  minutes,  however   𝑇!  was  rounded  to  half  an  hour  as  half  an   hour  was  the  smallest  time-­‐step  resolution  possible.    Woodlands  Drive  substation   loading  was  found  to  be  capable  of  supplying  100%  EV  penetration  for  underground   networks,  while  only  37%  could  be  charged  for  overhead  networks  assuming  a  half  hour   charging  time  and  unbalanced  network.     The  impacts  of  increasing  charging  rating  are  shown  in  Fig.  4.4  for  an  unbalanced   underground  network  at  100%  EV  penetration,  showing  that  complete  EV  penetration  is   possible  up  to  and  including  7  kW  average  charger  ratings.     Figure  4.4:  Impact  of  increasing  charger  rating  on  underground  network  at  100%  EV  penetration        
41     41     Discussion  of  Results   The  analysis  of  400  V  coordinated  charging  results  has  brought  a  number  of  issues  to   light,  some  of  which  were  not  entirely  obvious  without  further  investigation.  These   issues  included:   -­‐  The  difference  in  loading  capabilities  between  underground  and  overhead  networks   was  found  to  be  significant  due  to  the  substantial  difference  in  line  impedances.  The   underground  network  was  able  to  avoid  voltage  regulation  issues  occurring  before  the   transformer  capacity  was  reached,  while  the  overhead  network  could  not,  even  for  the   balanced  network  scenario.   -­‐  Loading  unbalance  during  the  evening  hours  was  the  limiting  factor  in  maximum  EV   penetration  for  overhead  networks.  The  combination  of  unbalanced  base  loads  and  EV   loads  after  random  distribution  caused  a  worst-­‐case  unbalance  scenario  which  is   possible  in  reality.   -­‐  The  location  of  loads  was  also  a  factor  that  contributed  to  poor  voltage  regulation,  as   those  closest  to  the  end  of  the  transformer  forced  additional  current  to  flow  the  entire   length  of  the  feeder,  causing  a  voltage  drop  worse  than  if  they  were  situated  closer  to  the   transformer.   -­‐  A  balanced  overhead  network  is  capable  of  supplying  a  lower  number  of  EVs  than  a   balanced  underground  network  at  the  same  base  loading,  as  the  losses  in  the   underground  substation  loads  make  up  a  smaller  percentage,  hence  supporting  a  larger   combined  house  load,  resulting  in  a  greater  reactive  power  supply.     Of  these  issues,  loading  unbalance  between  houses  and/or  EVs  was  found  to  be   the  most  significant  issue  in  overhead  networks,  causing  poor  voltage  regulation  that   limited  the  penetration  of  EV  charging  in  overhead  networks  to  loading  of  less  than  the   transformers  rating.  Voltage  unbalance  must  be  addressed  by  considering  the  evening   hours,  especially  in  areas  of  higher  economic  status  where  higher  rated  chargers  and   penetration  levels  are  more  probable.  Voltage  unbalance  is  not  expected  to  have  a   critical  effect  on  underground  networks,  as  lower  line  resistances  prevent  voltage  drop   from  being  as  significant.   The  most  important  information  that  can  be  interpreted  from  these  results  is  the   high  percentage  of  possible  EVs  that  may  charge  in  an  uncoordinated  manner  without   overloading  network  equipment.  The  high  percentage  of  EV  penetrations  were  due  to   the  large  variation  in  start  times,  which  resulted  in  only  9  of  the  46  vehicles  charging   during  the  6:00  to  6:30  pm  period  that  caused  peak  loading.  Keeping  in  mind  that  this  
42     42     analysis  disregarded  charging  at  the  workplace,  actual  allowable  EV  penetration  levels   may  be  significantly  higher  than  those  found,  depending  on  the  extent  to  which   workplace  charging  is  integrated.     Based  on  analysis  conducted  by  Deutsche  Bank  [19]  in  Chapter  1,  around  35%  of   vehicles  are  expected  to  be  EVs  by  2030.    Therefore,  transformers  with  a  base  loading  of   85%  in  overhead  networks  and  90%  in  underground  networks  are  expected  to  handle   EVs  until  at  least  2030,  assuming  a  4  kW  average  charger  rating  and  unbalanced  feeders.   However,  if  EV  penetration  increases  beyond  these  projections,  and  workplace  charging   does  not  become  significant,  overhead  networks  may  need  upgrading  as  early  as  this   decade.     4.2.3 11  kV  Transformer  Loading     By  scaling  the  base  house  load  profiles  of  the  550  kVA  overhead  LV  transformer,  a   loading  of  85.8%  was  required  to  cause  a  loading  of  96.67%  in  the  11  kV  model,  due  to   the  significant  proportion  of  losses.  EV  loads  were  then  added  until  the  11  kV  source   power  reached  4.5  MVA,  or  100%.  Modelling  the  LV  network  as  balanced  had  a   significant  impact  on  the  voltage  regulation,  allowing  the  voltage  to  stay  within  its  limits.     Although  this  would  not  be  the  case  in  reality,  this  assumption  had  to  be  made  to   avoid  the  voltage  collapse  that  would  result  when  using  the  smart  meter  load  profiles.   Where  voltage  regulation  was  too  poor  at  the  LV  level,  LV  substations  would  become  the   limiting  factor  for  zone  substation  loading.  The  overhead  network  was  modelled  on  a   tap  setting  of  -­‐4  while  the  underground  network  required  a  tap  setting  of  -­‐3.  The   maximum  penetrations  for  uncoordinated  charging  are  shown  in  Table  4.5.   Charger   Rating   Maximum  EV  Penetration   𝑻 𝒄=  0.5  Hours   𝑻 𝒄=  1  Hour   𝑻 𝒄=  1.5  Hours   𝑅   0.75𝑅   0.5𝑅   𝑅   0.75𝑅   0.5𝑅   𝑅   0.75𝑅   0.5𝑅   4  kW   -­‐   -­‐   -­‐   8%   10%   16%   4%   5%   8%   7  kW   8%   10%   16%   4%   5%   8%   -­‐   -­‐   -­‐   10  kW   5%   8%   10%   3%   4%   6%   -­‐   -­‐   -­‐   Table  4.5:  Maximum  EV  penetration  at  zone  substation  assuming  worst  loading  day  in  2010/11    Table  4.5  is  divided  into  charge  times,  and  sub  divided  into  possible  zone  substation   load  ratios,  where  R  represents  100%  residential  customers,  while  0.5R  may  represent   the  reality  that  50%  of  the  zone  substations  load  is  residential,  while  the  other  50%  may   be  commercial  and/or  industrial.  The  EV  penetrations  were  found  by  simply  multiplying   the  modelled  results,  R,  with  the  reciprocal  of  the  ratio  of  residential  to  non-­‐residential  
43     43     loads.  This  was  possible  because  non-­‐residential  loads  are  not  expected  to  charge  EVs   during  the  evening  peak.   The  maximum  EV  penetration  level  was  constant  for  overhead  and  underground   networks,  showing  that  the  increase  to  losses  by  adding  EVs  was  negligible  when  the   underground  and  overhead  networks  were  loaded  to  the  same  high  capacity,  inclusive   of  the  base  house  losses.       The  penetration  values  displayed  in  Table  4.5  are  significantly  lower  than  the   results  obtained  for  LV  uncoordinated  charging  scenarios.  This  is  partly  due  to  the   maximum  base  substation  loading  simulated  was  95%,  while  this  zone  substation   loading  is  based  on  a  96.67%  loaded  day.  Additional  line  losses  in  the  11  kV  lines  are   also  a  contributing  factor.         4.3 Coordinated  Charging   4.3.1 3-­‐Group  Charging     4  kW  Chargers     With  an  average  charger  rating  of  4  kW,  as  is  expected  in  the  coming  years,  coordinated   charging  was  found  to  keep  transformer  loading  at  a  suitable  level  for  100%   penetration,  for  all  base  loading  levels.  The  base  loading  level  was  found  to  have  a  minor   impact  on  coordinated  charging,  as  this  charging  occurs  after  the  evening  peak.       Figure  4.5:  4  kW  three-­‐group  coordinated  charging  for  different  transformer  base  levels   Figure  4.5  shows  coordinated  charging  with  100%  EV  loading  on  an  overhead  network   for  the  Woodlands  Drive  base  loading,  and  80-­‐95%  loaded  substations.  The  maximum   load  is  seen  to  remain  as  the  evening  peak,  which  does  not  exceed  500  kVA.    
44     44       The  balanced  overhead  network  was  chosen  for  coordinated  modelling  because  this   network  allowed  90  and  95%  base  loads  to  draw  power  during  the  evening  peak   without  voltage  regulation  issues.  The  coordinated  charging  load,  however,  was  the   same  magnitude  for  the  unbalanced  overhead  network  as  no  significant  loading   differences  were  present  in  the  late  night  hours  in  the  unbalanced  model.     7  kW  Chargers     Existing  Base   Transformer   Loading  on  Hot  Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Woodlands  Dr.   96%   100%   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   87%   94%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   86%   92%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   84%   91%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙 82%   89%   Table  4.6:  Maximum  EV  penetration  for  7kW  LV  coordinated  charging   As  EV  charger  ratings  were  increased  to  7  kW  on  average,  Table  4.6  shows  three-­‐group   coordinated  charging  was  able  to  increase  the  maximum  penetration  significantly  over   uncoordinated  charging.  At  85%  base  loading,  for  example,  uncoordinated  charging  is   limited  to  33%  for  a  balanced  overhead  network,  but  can  be  increased  considerably  to   86%  with  three  off-­‐peak  charging  groups.     10  kW  Chargers   The  same  improvement  was  seen  with  10  kW  chargers  in  Table  4.7,  as  a  penetration   maximum  of  13%  can  be  increased  to  57%  for  an  overhead  balanced  network  on  a  95%   loaded  transformer.     Table  4.7:  Maximum  EV  penetration  for  10  kW  LV  coordinated  charging   Existing  Base   Transformer   Loading  on  Hot  Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Woodlands  Dr.   67%   72%   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   60%   66%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   59%   64%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   58%   63%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙 57%   62%  
45     45     4.3.2 Six-­‐Group  Charging   The  change  to  six  groups  of  charging  start  times  saw  a  significant  increase  in  the   maximum  number  of  EVs  able  to  charge  on  a  substation  –  so  much  so  that  100%   penetration  was  possible  for  an  overhead  network  loaded  at  95%  during  the  evening   peak,  with  an  average  charger  rating  of  10  kW.  Figure  4.6  shows  six-­‐group  off-­‐peak   charging  at  100%  EV  penetration  for  4,  7  and  10  kW  charger  ratings.       Figure  4.6:  Six-­‐group  coordinated  charging  for  a  95%  loaded  transformer   Therefore,  the  vast  majority  of  transformers  are  expected  to  be  capable  of  supporting   100%  penetration  of  10  kW  charging.  The  ratio  of  vehicles  could  be  further  tweaked  to   prevent  the  evening  peak  being  exceeded  for  the  same  10  pm  to  4  am  charging  period.   4.3.3 11  kV     A  24  hour  load  profile  was  not  available  for  the  hottest  day  of  2010/11,  when  the   maximum  zone  substation  loading  of  96.67%  occurred.  Based  on  the  results  for  LV   coordinated  charging,  however,  there  is  not  expected  to  be  any  issues  in  the  LV   networks  unless  charging  cannot  be  coordinated  into  six-­‐group  off-­‐peak  staggered   charging,  or  a  more  advanced  coordinated  strategy  is  not  developed.  Also,  the   commercial  and  industrial  substations  will  not  draw  coordinated  charging  loads,  further   reducing  the  likelihood  that  coordinated  charging  will  ever  be  an  issue  at  the  11  kV  level.        
46     46     5 Conclusion       This  thesis  looks  at  the  impacts  of  electric  vehicle  charging  on  the  low  and  medium   voltage  networks  for  both  uncoordinated  and  coordinated  charging  scenarios.  Sample   low  and  medium  voltage  network  models  provided  by  Endeavour  Energy  allowed  for   realistic  network  modelling  of  both  overhead  and  underground  networks,  while  smart   meter  data  from  premises  supplied  by  Woodlands  Drive  LV  substation  ensured  accurate   load  profile  patterns  and  realistic  phase  unbalance  for  simulation  in  the  PowerFactory   models.  Analysis  was  conducted  using  load  profiles  recorded  on  the  hottest  day  of  2011   to  create  a  worst-­‐case  base  loading  scenario  which  network  planning  must  be  based   upon.  A  significant  number  of  variables  have  been  considered  to  ensure  that  all  loading   possible  scenarios  have  been  taken  into  account.   The  development  of  a  GUI  resulted  in  a  powerful  tool  for  simulating  and   analysing  charging  scenarios  for  this  thesis,  and  will  serve  the  same  purpose  for   network  planners  at  Endeavour  Energy.  This  tool  integrates  MATLAB,  DIgSILENT   PowerFactory  and  Excel  for  remotely  controlled  analysis  of  an  unlimited  number  of   charging  scenarios.     The  simulation  of  EV  loading  scenarios  found  that  uncoordinated  charging  may   become  less  of  a  problem  than  expected  for  lightly  loaded  substations  and  those  in   underground  areas,  for  the  expected  charger  rating  of  4  kW.  Phase  unbalance,  however,   was  found  to  be  the  limiting  factor  in  overhead  networks,  limiting  transformer  loading   to  the  voltage  regulation  capability  of  the  network.  Based  on  EV  growth  projections,   transformers  that  reach  85%  or  below  during  the  hottest  days  of  the  year  are  expected   to  handle  EV  loading  until  at  least  2030  for  unbalanced  underground  and  overhead   networks,  however  an  increase  in  the  expected  charger  rating  and  average  driving   distance,  will  reduce  this  time  frame  in  which  overloading  will  occur  to  as  soon  as  this   decade.  The  impacts  of  uncoordinated  charging  were  found  to  be  worse  at  the  medium   voltage  level,  with  zone  substations  possibly  requiring  upgrades  by  the  next  decade  if  an   uncoordinated  strategy  is  not  implemented  at  the  low  voltage  level.     The  study  of  simple  off-­‐peak  coordinated  charging,  however,  determined  that   late-­‐night  charging  is  expected  to  avoid  any  overloading  issues  at  zone  and  LV   substations.  A  three-­‐group  off-­‐peak  coordinated  charging  strategy  allowed  100%   penetration  of  EVs  for  the  expected  charger  rating  of  4  kW,  while  a  six-­‐group  charging  
47     47     method  allowed  complete  EV  penetration  even  for  10  kW  rated  charging  on  substations   up  to  95%  loaded  on  the  hottest  day  of  the  year.      The  analysis  in  this  thesis  provides  an  accurate  guide  to  the  expected  loading   effects  of  EVs  in  the  coming  years,  and  the  ways  in  which  any  undesired  loading  effects   may  be  mitigated.                                        
48     48     References       [1]  C.  Connelly,  (2012,  May).  “Can  petrol  really  burn  out  and  fade  away  for  good?”   [Online].  Available:  http://www.news.com.au/technology/can-­‐petrol-­‐really-­‐burn-­‐out-­‐ and-­‐fade-­‐for-­‐good/story-­‐e6frfro0-­‐1226359173949   [2]  M.  Masoum,  P.  Moses,  and  S.  Hajforoosh,  “Distribution  transformer  stress  in  smart   grid  with  coordinated  charging  of  plug-­‐in  electric  vehicles”,  in  Innovative  Smart  Grid   Technologies  (ISGT),  2012  IEEE  PES,  Jan.  2012,  pp.  1  –  8.       [3]  K.  Clement-­‐Nyns,  E.  Haesen,  and  J.  Driesen,  “The  impact  of  charging  plug-­‐in  hybrid   electric  vehicles  on  a  residential  distribution  grid”,  Power  Systems,  IEEE  Transactions  on,   vol.  25,  no.  1,  pp.  371  –  380,  Feb.  2010.     [4]  M.  S.  S.  J.  D.  Glover  and  T.J.  Overbye,  Power  System  Analysis  and  Design,  5th  ed.   Stamform,  CT:  C.  M.  Shortt,  2012.     [5]  S.  A.  N.  Zealand,  “As/nzs  3000:2007”,  july  2009.   [6]  N.  F.  E.  Command.  (1990)  Electric  power  distribution  system  operations.  [Online].   Available:  http://www.wbdg.org/ccb/NAVFAC/OPER/mo201.pdf.     [7]  X.  Liu,  A.  Aichhorn,  L.  Liu,  and  H.  Li,  “Coordinated  control  of  distributed  energy   storage  system  with  tap  changer  transformers  for  voltage  rise  mitigation  under  high   photovoltaic  penetration”,  Smart  Grid,  IEEE  Transactions  on,  vol.  3,  no.  99,  pp.  1  –  10,   2012.     [8]  P.  Kadurek,  J.  Cobben,  and  W.  Kling,  “Smart  mv/lv  transformer  for  future  grids”,  in   Power  Electronics  Electrical  Drives  Automation  and  Motion  (SPEEDAM),  2010   International  Symposium  on,  June  2010,  pp.  1700  –  1705.     [9]  A.  Keyhani,  Design  of  Smart  Power  Grid  Renewable  Energy  Systems.  Hobo-­‐ken,  NJ:   Wiley  –  IEEE,  2011.     [10]  Tennessee  Valley  Authority.  (N.d.)  Types  of  electric  vehicles.  [Online].  Available:   http://www.tva.com/environment/technology/car_vehicles.htm.     [11]  Holden.  (2012)  Holden  volt  coming  soon.  [Online].  Available:   http://www.holden.com.au/pages/volt-­‐coming-­‐soon.     [12]  M.  T.  Thompson.  (N.d.)  Generic  battery  technology  comparison.  [Online].  Available   http://www.madkatz.com/ev/batteryTechnologyComparison.html.     [13]  T.  Motors.  (2012)  Model  s/options  and  pricing.  [Online].  Available:   http://www.teslamotors.com/models/options.    
49     49     [14]  Ford.  (2012)  2012  focus  electric.  [Online].  Available:   http://www.ford.com/electric/focuselectric/2012/.     [15]  Toyota.  (2012)  2012  Toyota  prius  plug-­‐in  hybrid  performance  and  specs.  [Online].   Available:  http://www.toyota.com/prius-­‐plug-­‐in/specs.html.     [16]  Nissan.  (2012)  Features  and  specifications.  [Online].  Available:   http://www.nissanusa.com/leaf-­‐electric-­‐car/specs-­‐features/index#/leaf-­‐electric-­‐ car/specs-­‐features/index.   [17]  International  Electrotechnical  Commission.  (2003)  International  standard  IEC   62196  –  1.  [Online].  Available:  http://webstore.iec.ch/p-­‐preview/info_iec62196-­‐ 1%7Bed1.0%7Den.pdf.     [18]  Society  of  Automotive  Engineers.  (2012)  Sae  electric  vehicle  and  plug  in  hybrid   electric  vehicle  conductive  charge  coupler.  [Online].  Available:   http://standards.sae.org/j1772_201202/.     [19]  P.  N.  R.  Lache,  D.  Galves,  “Vehicle  electrification”,  Mar.  2010.     [20]  B.  T.  T.  A.  Becker,  I.  Sidhu.  (2009)  Electric  vehicles  in  the  United  States.  [Online].   Available:   http://cet.berkeley.edu/dl/CET_Technical%20Brief_EconomicModel2030_f.pdf.     [21]  R.  D.  P.  Harrup.  (2009)  Hybrid  and  pure  electric  cars  2012  –  2022.  [Online].   Available:  http://www.idtechex.com/research/reports/hybrid-­‐and-­‐pure-­‐electric-­‐cars-­‐ 2012-­‐2022-­‐000287.asp.     [22]  E.  Sortomme,  M.  Hindi,  S.  MacPherson,  and  S.  Venkata,  “Coordinated  charging  of   plug-­‐in  hybrid  electric  vehicles  to  minimize  distribution  system  losses”,  Smart  Grid,  IEEE   Transactions  on,  vol.  2,  no.  1,  pp.  198  –  205,  March  2011.     [23]  K.  Qian,  C.  Zhou,  M.  Allan,  and  Y.  Yuan,  “Load  model  for  prediction  of  electric  vehicle   charging  demand”,  in  Power  System  Technology  (POWERCON),  2010  International   Conference  on,  Oct.  2010,  pp.  1  –  6.   [24]  K.  Qian,  C.  Zhou,  M.  Allan,  Y.  Yuan,  “Modeling  of  load  demand  due  to  ev  battery   charging  in  distribution  systems”,  Power  Systems,  IEEE  Transactions  on,  vol.  26,  no.  2,  pp.   802  –  810,  May  2011.     [25]  S.  Huang  and  D.  Infield,  “The  potential  of  domestic  electric  vehicles  to  contribute  to   power  system  operation  through  vehicle  to  grid  technology”,  in  Universities  Power   Engineering  Conference  (UPEC),  Proceedings  of  the  44th  International,  Sept.  2009,  pp.  1-­‐5.    
50     50     [26]  A.  Masoum,  S.  Deilami,  P.  Moses,  and  A.  Abu-­‐Siada,  “Impact  of  plug-­‐in  electrical   vehicles  on  voltage  profile  and  losses  of  residential  system”,  in  Universities  Power   Engineering  Conference  (AUPEC),  2010  20TH  Australasian,  Dec.  2010,  pp.  1  –  6.     [27]  S.  Deilami,  A.  Masoum,  P.  Moses,  and  M.  Masoum,  “Real-­‐time  coordination  of  plug-­‐in   electric  vehicle  charging  in  smart  grids  to  minimize  power  losses  and  improve  voltage   profile”,  Smart  Grid,  IEEE  Transactions  on,  vol.  2,  no.  3,  pp.  456  –  467,  Sept.  2011.     [28]  Bureau  of  Transport  Statistics  NSW,  “Household  travel  survey  (HTS)”,  Electronic   Publication  No.  E2012-­‐02-­‐HTS-­‐Summary   [29]  Australian  Bureau  of  Statistics  (ABS),  “2011  Census  QuickStats”  [Online],   Availalable:http://www.censusdata.abs.gov.au/census_services/getproduct/census/20 11/quickstat/0      
51     51         Appendix  A     Project  Plan  and  Specification      
52     52     Project  Specification     As  the  penetration  of  electric  vehicles  increases,  charging  of  these  vehicles  is  expected  to   have  significant  loading  effects  on  the  distribution  network,  similar  to  the  effects  caused   by  the  growing  number  of  air  conditioners  over  recent  years.  The  aim  of  this  thesis  is  to   continue  with  analysis  from  ECTE451  to  determine  the  impacts  of  uncoordinated   charging,  as  well  as  coordination  strategies  aimed  at  avoiding  these  loading  effects.  The   coordinated  charging  strategy  of  focus  will  be  staggered  off-­‐peak  charging,  and  novel   coordinated  charging  solutions  explored  in  ETE451,  such  as  charging  during  periods  of   high  solar  energy  generation,  and  supplying  power  to  the  grid  through  vehicle-­‐to-­‐grid   (V2G),  will  no  longer  be  considered.  The  focus  of  the  analysis  will  move  from  a  single   simplified  feeder  that  must  be  modified  manually  for  different  charging  penetrations,  to   an  automated  package  that  will  allow  network  planners  to  select  penetrations  levels,  as   well  as  variables  such  as  network  type  and  temperature,  to  determine  the  effects  of   charging  from  a  simple  user  interface.     Network  Modelling     Prior   study   in   ECTE451   involved   the   investigation   of   transformer   loading   and   line   voltage  levels  for  a  400  V  residential  feeder  using  PowerWorld  Simulator.  This  analysis   assumed  approximated  load  profiles  based  on  crude  peak  demand  estimates  based  on   typical   appliance   use,   and   estimated   network   equipment   ratings.   To   improve   on   this,   analysis   will   be   conducted   using   DIgSILENT   PowerFactory   models   provided   by   Endeavour  Energy.  These  models  cover  both  the  400  V  and  11  kV  distribution  levels,  for   underground,  overhead  and  semi-­‐rural  areas.  As  well,  smart  metering  load  profile  data   from  the  network  area  of  Glenmore  Park  has  been  provided  from  both  residential  and   zone   substation   meters,   containing   both   summer   and   winter   load   profiles.     This   will   allow  an  accurate  investigation  into  the  effects  of  charging  on  the  Glenmore  Park  area,   and  provide  a  reliable  insight  into  the  effects  on  the  remainder  of  the  Endeavour  Energy   network.     Using  these  models  as  a  starting  point,  they  will  be  modified  to  reflect  the  number  of   premises  on  the  Glenmore  Park  distribution  substation  containing  the  residential  smart   meters  of  interest.  The  aim  will  be  to  determine  the  effects  of  electric  vehicle  charging   on   transformer   loading   and   line   voltages,   for   both   uncoordinated   and   coordinated   charging  strategies.     Utility  Planning  Tool     In  order  to  run  a  load  flow  in  PowerFactory,  a  DIgSILENT  Programming  Language  (DPL)   script   must   be   executed.   The   DPL   scripts   provided   by   Endeavour   Energy   allow   line   voltages   and   transformer   loading   to   be   determined.   As   the   effects   of   charging   can   be   determined   with   the   DIgSILENT   projects   currently   provided,   the   focus   of   this   project   will  be  to  integrate  the  existing  scripts  into  a  package  for  network  planners  to  easily  run   a  number  of  penetration  scenarios.  Loads  in  DIgSILENT  are  associated  with  a  single  CSV   file   containing   data   for   specified   time   intervals,   which   must   be   manually   changed   to   determine  a  different  charging  penetration.  This  configuration  would  make  it  tedious  to   determine  the  effects  of  a  number  of  penetration  levels  using  a  single  project.  In  order  to   overcome  this,  MATLAB  will  be  used  as  an  interface  for  selecting  variables,  controlling   DPL  scripts  and  displaying  results.  The  scripting  will  work  as  follows:    
53     53     1.   MATLAB   will   act   as   an   interface   where   variables   are   selected.   These   variables   will   include   location   (to   specify   the   average   charging   time   based   on   average   driving   distance),   charging   power,   charging   penetration   level,   charging   strategy,   temperature   and  network  type.   2.  MATLAB  will  run  a  script  to  copy  specified  csv  files  to  the  master  files  associated  with   each  load  in  the  selected  DIgSILENT  model.   3.  DIgSILENT  will  run  the  existing  400  V  DPL  script  once  called  by  Matlab,  to  determine   low  voltage  transformer  loading  and  line  voltage  data.     4.   MATLAB   will   write   the   400   V   transformer   load   profile   to   the   single   11   kV   csv   file   associated  with  each  distribution  substation  load  in  the  11  kV  model.   5.  DIgSILENT  will  run  the  existing  11  kV  DPL  script  once  called  by  Matlab,  to  determine   medium  voltage  transformer  loading  and  line  voltage  data.     6.  MATLAB  will  display  both  the  400  V  and  11  kV  transformer  loading  and  line  voltage   data.     This  MATLAB-­‐controlled  arrangement  will  allow  an  Endeavour  Energy  network  planner   to  run  scenarios  quickly  and  simply  to  determine  the  impacts  of  charging  for   consideration  in  future  network  planning.                            
54     54     Appendix  B     Logbook  Summary  Signature  Sheet      
55     55         Appendix  C     Software  Documentation      
56     56     MATLAB  GUI  Code     function varargout = secondGUI(varargin) % Initialisation code generated by MATLAB not shown %~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~START~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~% % --- Executes on button press in pushbutton1. function pushbutton1_Callback(hObject, eventdata, handles) network_type=handles.network_type; coordination=handles.coordination; chargepower_val=handles.chargepower_val; penetration_val=handles.penetration_val; disp Variables:; disp ' '; disp(network_type); disp(coordination); disp(chargepower_val); disp(penetration_val); % save variables from callback function where created oldpointer = get(handles.figure1, 'pointer'); set(handles.figure1, 'pointer', 'watch') % display hourglass to show loading drawnow; %determine selected variables and call appropriate matlab functions if strcmp(network_type,'Overhead')&& strcmp(coordination,'Uncoordinated') uncoEVcsv(chargepower_val,penetration_val); OH_batch(); OH_excelcmd(); [dailykva,p,ploc,v,vloc]=OH_getimportant() elseif strcmp(network_type,'Underground')&& strcmp(coordination,'Uncoordinated') uncoEVcsv(chargepower_val,penetration_val); UG_batch(); UG_excelcmd(); [dailykva,p,ploc,v,vloc]=UG_getimportant() elseif strcmp(network_type,'Overhead')&& strcmp(coordination,'Staggered 10 pm start') staggeredEVcsv(chargepower_val,penetration_val); OH_batch(); OH_excelcmd(); [dailykva,p,ploc,v,vloc]=OH_getimportant() elseif strcmp(network_type,'Underground')&& strcmp(coordination,'Staggered 10 pm start') staggeredEVcsv(chargepower_val,penetration_val); UG_batch();
57     57     UG_excelcmd(); [dailykva,p,ploc,v,vloc]=UG_getimportant() else errordlg('Please ensure all network variables have been selected'); end %run 11kv model using matlab functions writeto11kvloads(dailykva); HV_batch(); HV_excelcmd(); [dailykva11,p11,ploc11,v11,vloc11]=HV_getimportant(); zsub=45-p11; pow11=[p11,zsub]; %calculate time of max power and min voltage ptime = gettime(ploc); vtime = gettime(vloc); ptime11 = gettime(ploc11); vtime11 = gettime(vloc11); %get 400 V data and display on graph and in text box dsub=250-p; pow=[p,dsub]; pie3(handles.axes1,pow); handles.list_item1='Transformer Loading'; handles.list_item2='400 V Network'; disp(pow); powerLVtable=['Maximum Substation Power = ', num2str(p), ' KVA at ' num2str(ptime)]; set(handles.edit3,'String',powerLVtable); %save data in handles so can be passed into other functions handles.pow=pow; handles.v=v; handles.p=p; handles.p11=p11; handles.pow11=pow11; handles.v11=v11; handles.ptime=ptime; handles.vtime=vtime; handles.ptime11=ptime11; handles.vtime11=vtime11; guidata(hObject, handles); set(handles.figure1, 'pointer', oldpointer) cd c:UsersOwnerDesktop'Matlab Scripts'; % --- Executes on selection change in popupmenu1. function popupmenu1_Callback(hObject, eventdata, handles) val1 = get(hObject,'Value'); string_list = get(hObject,'String');
58     58     network_type = string_list{val1}; % Convert from cell array to string handles.network_type=network_type; guidata(hObject, handles); % update handles % --- Executes during object creation, after setting all properties. function popupmenu1_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end % --- Executes on selection change in popupmenu2. function popupmenu2_Callback(hObject, eventdata, handles) val3 = get(hObject,'Value'); string_list = get(hObject,'String'); coordination = string_list{val3}; % Convert from cell array to string handles.coordination=coordination; guidata(hObject, handles); % --- Executes during object creation, after setting all properties. function popupmenu2_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end % --- Executes on slider movement. %get penetration value function slider1_Callback(hObject, eventdata, handles) penetration_val = round(get(hObject,'Value')); set(handles.edit1,'String',penetration_val); handles.penetration_val=penetration_val; guidata(hObject, handles); % --- Executes during object creation, after setting all properties. %locates and sizes background image function slider1_CreateFcn(hObject, eventdata, handles) if isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor',[.9 .9 .9]); end function edit1_Callback(hObject, eventdata, handles) % --- Executes during object creation, after setting all properties. % creates slider 1 on open GUI
59     59     function edit1_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end % --- Executes on slider movement. % get charging power function slider2_Callback(hObject, eventdata, handles) chargepower_val = get(hObject,'Value'); set(handles.edit2,'String',chargepower_val); handles.chargepower_val=chargepower_val; guidata(hObject, handles); % --- Executes during object creation, after setting all properties. %creates slider 2 on opening GUI function slider2_CreateFcn(hObject, eventdata, handles) if isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor',[.9 .9 .9]); end % --- Executes during object creation, after setting all properties. % sizes and positions background function figure1_CreateFcn(hObject, eventdata, handles) ha = axes('units','normalized', ... 'position',[0 0 1 1]); % Move the background axes to the bottom uistack(ha,'bottom'); I=imread('bestbanner.png'); hi = imagesc(I); colormap gray % Turn the handlevisibility off, make the axes invisible set(ha,'handlevisibility','off', ... 'visible','off') % --- Executes on selection change in listbox1. % determines which list item is selected function listbox1_Callback(hObject, eventdata, handles) index_selected = get(hObject,'Value'); list = get(hObject,'String'); list_item1 = list{index_selected}; handles.list_item1=list_item1; guidata(hObject, handles); disp(list_item1);
60     60     list_item2=handles.list_item2; pow=handles.pow; p=handles.p; v=handles.v; pow11=handles.pow11; p11=handles.p11; v11=handles.v11; ptime=handles.ptime; vtime=handles.vtime; ptime11=handles.ptime11; vtime11=handles.vtime11; % Set text data in text box voltageLVtable=['Lowest bus voltage = ', num2str(v), ' V at ' ,num2str(vtime)]; powerLVtable=['Maximum Substation Power = ', num2str(p), ' KVA at ' ,num2str(ptime)]; voltageHVtable=['Lowest bus voltage = ', num2str(v11), ' kV ' ,num2str(vtime11)]; powerHVtable=['Maximum Substation Power = ', num2str(p11), ' MVA ',num2str(ptime11)]; %determines variables selected when list box 1 has been selected if strcmp(list_item1,'Transformer Loading') && strcmp(list_item2,'400 V Network') display '400 TX Loading works' pie3(handles.axes1,pow); set(handles.edit3,'String',powerLVtable); elseif strcmp(list_item1,'Line Voltages') && strcmp(list_item2,'400 V Network') display '400 volt Loading works' bar(handles.axes1,v,'stacked'); axis([0.75 1.25 200 250]); set(gca, 'XTickLabelMode', 'Manual'); set(gca, 'XTick', []); set(handles.edit3,'String',voltageLVtable); elseif strcmp(list_item1,'Transformer Loading') && strcmp(list_item2,'11 kV Network') display '11 kv TX Loading works' pie3(handles.axes1,pow11); set(handles.edit3,'String',powerHVtable); elseif strcmp(list_item1,'Line Voltages') && strcmp(list_item2,'11 kV Network') display '11 kv volt Loading works' bar(handles.axes1,v11,'stacked'); axis([0.75 1.25 10 11]); set(gca, 'XTickLabelMode', 'Manual'); set(gca, 'XTick', []); set(handles.edit3,'String',voltageHVtable); end % --- Executes during object creation, after setting all properties. % creates list box on opening GUI function listbox1_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
61     61     set(hObject,'BackgroundColor','white'); end % --- Executes during object creation, after setting all properties. %sets graph axes function axes1_CreateFcn(hObject, eventdata, handles) set(gca, 'YTickLabelMode', 'Manual') set(gca, 'YTick', []) set(gca, 'XTickLabelMode', 'Manual') set(gca, 'XTick', []) % --- Executes on selection change in listbox2. % creates list box on opening GUI function listbox2_Callback(hObject, eventdata, handles) index_selected = get(hObject,'Value'); list = get(hObject,'String'); list_item2 = list{index_selected}; disp(list_item2); handles.list_item2=list_item2; guidata(hObject, handles); list_item1=handles.list_item1; pow=handles.pow; p=handles.p; v=handles.v; pow11=handles.pow11; p11=handles.p11; v11=handles.v11; ptime=handles.ptime; vtime=handles.vtime; ptime11=handles.ptime11; vtime11=handles.vtime11; % Set Text data voltageLVtable=['Lowest bus voltage = ', num2str(v), ' V at ' ,num2str(vtime)]; powerLVtable=['Maximum Substation Power = ', num2str(p), ' KVA at ' ,num2str(ptime)]; voltageHVtable=['Lowest bus voltage = ', num2str(v11), ' kV ' ,num2str(vtime11)]; powerHVtable=['Maximum Substation Power = ', num2str(p11), ' MVA of 45 MVA ',num2str(ptime11)]; %determines variables selected when list box 1 has been selected if strcmp(list_item1,'Transformer Loading') && strcmp(list_item2,'400 V Network') display '400 TX Loading works' pie3(handles.axes1,pow); set(handles.edit3,'String',powerLVtable); elseif strcmp(list_item1,'Line Voltages') && strcmp(list_item2,'400 V Network')
62     62     display '400 volt Loading works' bar(handles.axes1,v,'stacked'); axis([0.75 1.25 200 250]); set(gca, 'XTickLabelMode', 'Manual'); set(gca, 'XTick', []); set(handles.edit3,'String',voltageLVtable); elseif strcmp(list_item1,'Transformer Loading') && strcmp(list_item2,'11 kV Network') display '11 kv TX Loading works' pie3(handles.axes1,pow11); set(handles.edit3,'String',powerHVtable); elseif strcmp(list_item1,'Line Voltages') && strcmp(list_item2,'11 kV Network') display '11 kv volt Loading works' bar(handles.axes1,v11,'stacked'); axis([0.75 1.25 10 11]); set(gca, 'XTickLabelMode', 'Manual'); set(gca, 'XTick', []); set(handles.edit3,'String',voltageHVtable); end % --- Executes during object creation, after setting all properties. function listbox2_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end function edit3_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end Batch  file  to  call  command  file  to  open  DIgSILENT  Engine  connection:     digrcom  -­‐d  -­‐p  ncacn_ip_tcp  -­‐n  127.0.0.1  -­‐e  2001  -­‐f="11kvcmd.cmd"     Command  file:     cls/inp   cls/out   ac/de  all   ac  Matthew  WardhaughOH  11kV  Feeder   cd  Matthew  WardhaughOH  11kV  FeederLibraryScripts   11kV  Losses  and  Voltages  script     Command  file  to  control  Excel  files:  
63     63       cd    c:UsersOwnerDesktopDIgSILENT  Results11kV   del  "11kvvoltresults.csv"   del  "11kvTXresults.csv"   rename  "11kvvoltresults.txt"  "11kvvoltresults.csv"   rename  "11kvTXresults.txt"  "11kvTXresults.csv"   cd    c:DIgSILENTpf141ENGINE   11kvexcelVBScript     VBScript  file  to  format  CSV  files:     Set  objExcel  =  CreateObject("Excel.Application")     objExcel.Visible  =  False   objExcel.Workbooks.Open("C:UsersOwnerAppDataRoamingMicrosoftAddInstxt 2colsmacro.xlam")   objExcel.Workbooks.Open("C:UsersOwnerDesktopDIgSILENT   Results11kv11kvvoltresults.csv")   objExcel.Run  "txt2cols"   objExcel.ActiveWorkbook.Save   objExcel.ActiveWorkbook.Close   objExcel.Workbooks.Open("C:UsersOwnerAppDataRoamingMicrosoftAddInstxt 2colsmacro.xlam")   objExcel.Workbooks.Open("C:UsersOwnerDesktopDIgSILENT   Results11kv11kvTXresults.csv")   objExcel.Run  "txt2cols"   objExcel.ActiveWorkbook.Save   objExcel.ActiveWorkbook.Close   objExcel.Quit     VBA  “txt2cols”  Macro  to  remove  text  and  zero  columns:     Sub  txt2cols()   '   '  txt2cols  Macro   '  Runs  text  to  columns  on  column  A  to  split  digsilent  result  data  to  csv  format   Application.DisplayAlerts  =  False     'Set  up  the  selection  range   Dim  ColumnA  As  Range   Dim  Text  As  Range   Set  ColumnA  =  Range("A:A")   Set  Text  =  Range("1:2")     'Run  Text  to  Columns  function   ColumnA.TextToColumns  _              Destination:=Range("$A$1"),  _              DataType:=xlDelimited,  _              TextQualifier:=xlDoubleQuote,  _              ConsecutiveDelimiter:=False,  _              Tab:=True,  _  
64     64                Semicolon:=False,  _              Comma:=False,  _              Space:=False,  _              Other:=False     Text.Delete   ColumnA.Delete     'Delete  zero  columns   Dim  nLastColumn  As  Long   Set  r  =  ActiveSheet.UsedRange   nLastColumn  =  r.Columns.Count  +  r.Column  -­‐  1   For  i  =  nLastColumn  To  1  Step  -­‐1   i1  =  Application.WorksheetFunction.Sum(Columns(i))   i2  =  Application.WorksheetFunction.Count(Columns(i))   i3  =  Application.WorksheetFunction.CountA(Columns(i))   If  i1  =  0  And  i2  =  i3  Then   Columns(i).Delete   End  If   Next   End  Sub     MATLAB  Functions     Function  to  call  DIgSILENT  Engine  batch:     function OH_batch() cd c:DIgSILENTpf141ENGINE; system('OHbatch.bat'); cd c:UsersOwnerDesktop'Matlab Scripts'; end   Function  to  call  Excel  command  file:     function OH_excelcmd() cd c:DIgSILENTpf141ENGINE; system('OHexcelcmd.cmd'); cd c:UsersOwnerDesktop'Matlab Scripts' end   Function  to  extract  loading  data  from  CSV  files:     function [dailykva,maxkva,ploc,minvolt,vloc]=OH_getimportant() cd c:UsersOwnerDesktop'DIgSILENT Results'Overhead volt=csvread('OHvoltresults.csv'); dailykw=csvread('OHTXresults.csv',0,0,[0,0,47,0]); %47 is 48th row dailykvar=csvread('OHTXresults.csv',0,1,[0,1,47,1]); %47 is 48th row dailykva=2*roundn(sqrt(power(dailykw,2)+ power(dailykvar,2)),-1); [minvolt, location] = min(volt(:)); minvolt=minvolt/230; [vloc, y] = ind2sub(size(volt),location); [maxkva, location] = max(dailykva(:));
65     65     [ploc, y] = ind2sub(size(dailykva),location); minfirstbusvolt=csvread('OHvoltresults.csv',36,3,[36,3,36,3]);%36th row = 6 pm (not 37 with matlab convention) cd c:UsersOwnerDesktop'Matlab Scripts' end Function  to  get  time  of  max  power  and  min  voltage:     function [time]=gettime(loc) loc = loc/2; if loc == 0.5 time = ['12 am']; elseif loc == 1 time = ['12:30 am']; elseif loc<12 if rem(loc,1) == 0 time = [num2str(loc-1),':30 am']; else loc=loc-0.5; time = [num2str(loc),':00 am']; end else if rem(loc,1) == 0 loc = loc-12; time = [num2str(loc-1),':30 pm']; else loc=loc-12.5; time = [num2str(loc),':00 pm']; end end   [Remaining  functions/scripts/files  on  disc]    

More Related Content

PDF
D11_SMARTgirl FP Integration Report
Rosemary Taing
 
PDF
MPH van de Weerd 2010
Jennie van de Weerd
 
PDF
Accessing to Two-Dimensional Matrix of Turbine ElmDsl
Francisco Gonzalez-Longatt
 
PDF
Data Management in DIgSILENT PowerFactory 2010
Francisco Gonzalez-Longatt
 
PDF
Creating and Using Study Cases in DIgSILENT PowerFactory. Case P.M. Anderson ...
Francisco Gonzalez-Longatt
 
PDF
WORKSHOP: Frequency Control Schemes and Frequency Response of Power Systems c...
Francisco Gonzalez-Longatt
 
PDF
Modelling Renewables Resources and Storage in PowerFactory V15.2, Universi...
Francisco Gonzalez-Longatt
 
PDF
Importing/Exporting a project using DIgSILENT PowerFactory
Francisco Gonzalez-Longatt
 
D11_SMARTgirl FP Integration Report
Rosemary Taing
 
MPH van de Weerd 2010
Jennie van de Weerd
 
Accessing to Two-Dimensional Matrix of Turbine ElmDsl
Francisco Gonzalez-Longatt
 
Data Management in DIgSILENT PowerFactory 2010
Francisco Gonzalez-Longatt
 
Creating and Using Study Cases in DIgSILENT PowerFactory. Case P.M. Anderson ...
Francisco Gonzalez-Longatt
 
WORKSHOP: Frequency Control Schemes and Frequency Response of Power Systems c...
Francisco Gonzalez-Longatt
 
Modelling Renewables Resources and Storage in PowerFactory V15.2, Universi...
Francisco Gonzalez-Longatt
 
Importing/Exporting a project using DIgSILENT PowerFactory
Francisco Gonzalez-Longatt
 

Viewers also liked (8)

PDF
ELB044 Lecture 18. Introduction to Induction Machines
Francisco Gonzalez-Longatt
 
PDF
Example: Modal analysis using DIgSILENT PowerFactory.
Francisco Gonzalez-Longatt
 
PDF
Tutorial: Modelling and Simulations: Renewable Resources and Storage
Francisco Gonzalez-Longatt
 
PDF
Modelling Renewables Resources and Storage in PowerFactory V15.2, 9 June 2...
Francisco Gonzalez-Longatt
 
PDF
Future Smart-er Grid: Challenges
Francisco Gonzalez-Longatt
 
PDF
PowerFactory Applications for Power System Analysis, 26th October 2015, Por...
Francisco Gonzalez-Longatt
 
PDF
Tutorial: Introduction to Transient Analysis using DIgSILENT PowerFactory.
Francisco Gonzalez-Longatt
 
PDF
Workshop: PowerFactory Applications for Power System Analysis. Kasetsart Un...
Francisco Gonzalez-Longatt
 
ELB044 Lecture 18. Introduction to Induction Machines
Francisco Gonzalez-Longatt
 
Example: Modal analysis using DIgSILENT PowerFactory.
Francisco Gonzalez-Longatt
 
Tutorial: Modelling and Simulations: Renewable Resources and Storage
Francisco Gonzalez-Longatt
 
Modelling Renewables Resources and Storage in PowerFactory V15.2, 9 June 2...
Francisco Gonzalez-Longatt
 
Future Smart-er Grid: Challenges
Francisco Gonzalez-Longatt
 
PowerFactory Applications for Power System Analysis, 26th October 2015, Por...
Francisco Gonzalez-Longatt
 
Tutorial: Introduction to Transient Analysis using DIgSILENT PowerFactory.
Francisco Gonzalez-Longatt
 
Workshop: PowerFactory Applications for Power System Analysis. Kasetsart Un...
Francisco Gonzalez-Longatt
 
Ad

Similar to Thesis458 (20)

PDF
CLOUD CONTROLLED CHARGING STATIONS.pdf
Selvaraj A
 
PPTX
Sysc 5302 101079892
Manishkumar Moorjmalani
 
PPTX
Design of large scale ev charging station
OmarSakib810
 
PDF
IECON Martin Lévesque
martinzeitgeist
 
PDF
APPLICATION OF PHEVs FOR SMART GRID IN INDIAN POWER SECTOR1
Eshwar Pisalkar
 
PDF
thesis_template_FINAL_Dylan_Humenik-2lab94n
Dylan Humenik
 
PDF
Vehicle to Grid Technology
vivatechijri
 
PDF
Reactive power assessment with and without ev charging
Saif Shaikh
 
PDF
Impact of plug-in_electric_vehicles_on_the_distribution_grid-libre
Hemant Singh
 
PDF
“Reactive Power Assessment with And Without Electrical Vehicle Charging”
Saif Shaikh
 
PDF
Design and Implementation of Low-Cost Electric Vehicles (EVs) Supercharger: A...
pijans
 
PDF
Design and Implementation of Low-Cost Electric Vehicles (EVs) Supercharger: A...
pijans
 
PDF
Design and Implementation of Low-Cost Electric Vehicles (EVs) Supercharger: A...
pijans
 
DOCX
Two tier energy compansation framework based on smart electric charging station
SandeshPatil99
 
PDF
Integration of Smart Loads for Demand Side Management Using EV in the Medium ...
IRJET Journal
 
PDF
Power quality disturbance mitigation in grid connected photovoltaic distribu...
IJECEIAES
 
PPTX
below is my attached presentation read it.
MuhammadHassan213034
 
PDF
Hybrid AC/DC microgrid and Electric Vehicles
Md Shamiur Rahman
 
PPTX
V2G and G2V power transfer issues
Mohd Javed
 
PPTX
presntation about smart charging for the vehicles
UMBC
 
CLOUD CONTROLLED CHARGING STATIONS.pdf
Selvaraj A
 
Sysc 5302 101079892
Manishkumar Moorjmalani
 
Design of large scale ev charging station
OmarSakib810
 
IECON Martin Lévesque
martinzeitgeist
 
APPLICATION OF PHEVs FOR SMART GRID IN INDIAN POWER SECTOR1
Eshwar Pisalkar
 
thesis_template_FINAL_Dylan_Humenik-2lab94n
Dylan Humenik
 
Vehicle to Grid Technology
vivatechijri
 
Reactive power assessment with and without ev charging
Saif Shaikh
 
Impact of plug-in_electric_vehicles_on_the_distribution_grid-libre
Hemant Singh
 
“Reactive Power Assessment with And Without Electrical Vehicle Charging”
Saif Shaikh
 
Design and Implementation of Low-Cost Electric Vehicles (EVs) Supercharger: A...
pijans
 
Design and Implementation of Low-Cost Electric Vehicles (EVs) Supercharger: A...
pijans
 
Design and Implementation of Low-Cost Electric Vehicles (EVs) Supercharger: A...
pijans
 
Two tier energy compansation framework based on smart electric charging station
SandeshPatil99
 
Integration of Smart Loads for Demand Side Management Using EV in the Medium ...
IRJET Journal
 
Power quality disturbance mitigation in grid connected photovoltaic distribu...
IJECEIAES
 
below is my attached presentation read it.
MuhammadHassan213034
 
Hybrid AC/DC microgrid and Electric Vehicles
Md Shamiur Rahman
 
V2G and G2V power transfer issues
Mohd Javed
 
presntation about smart charging for the vehicles
UMBC
 
Ad

Thesis458

  • 1.               The  Impacts  of  Electric  Vehicle  Charging   on  Residential  Distribution  Systems         Matthew  Wardhaugh   Bachelor  of  Engineering  (Electrical)             October,  2012               Supervisor:  Dr  Phil  Ciufo                
  • 2. i     i         Abstract     A  significant  increase  in  the  number  of  electric  vehicles  is  expected  over  the  coming   years,  and  this  is  expected  to  create  issues  for  distribution  networks  when  charging   coincides  with  peak  demand  periods.  This  thesis  investigates  the  effects  of   uncoordinated  charging  on  the  residential  distribution  network,  and  looks  at  the   viability  of  coordinated  charging  to  mitigate  these  effects.  A  graphical  user  interface  was   created  to  aid  this  study  and  provide  a  tool  for  network  planners  to  easily  run  electric   vehicle  loading  scenarios.  This  thesis  finds  that  uncoordinated  charging  would  have  an   impact  on  low  voltage  networks,  particularly  for  overhead  networks  where  voltage   unbalance  is  a  greater  issue.  Simple  staggered  off-­‐peak  charging  was  investigated  and   found  to  mitigate  loading  effects  completely,  allowing  up  to  100%  electric  vehicle   penetration  for  the  highest  charger  rating  scenario.  The  impact  of  charging  was  found  to   be  significant  at  the  zone  substation  level  during  uncoordinated  charging  scenarios,   possibly  requiring  upgrades  within  the  next  decade  if  coordinated  charging  strategies   are  not  adopted.                                    
  • 3. ii     ii       Acknowledgements         I  would  like  to  thank  my  supervisors  Dr.  Phil  Ciufo  and  Prof.  Danny  Soetanto  for  their   guidance,  and  Endeavour  Energy  for  providing  network  models  and  data.    
  • 4. iii     iii         Statement  of  Originality         I,  Matthew  Wardhaugh,  declare  that  this  thesis,  submitted  as  part  of  the  requirements   for  the  award  of  Bachelor  of  Engineering,  in  the  School  of  Electrical,  Computer  and   Telecommunications  Engineering,  University  of  Wollongong,  is  wholly  my  own  work   unless  otherwise  referenced  or  acknowledged.  The  document  has  not  been  submitted   for  qualifications  or  assessment  at  any  other  academic  institution.         Signature:                     Print  Name:                     Student  ID  Number:   3667315     Date:                        
  • 5. iv     iv     Contents             Abstract  ....................................................................................................................................................................  i   Acknowledgements  ...........................................................................................................................................  ii   Statement  of  Originality  .................................................................................................................................  iii   Contents  ................................................................................................................................................................  iv   List  of  Figures  ......................................................................................................................................................  vi   List  of  Tables  ......................................................................................................................................................  vii   List  of  Equations  .............................................................................................................................................  viii   Abbreviations  and  Symbols  ..........................................................................................................................  ix   List  of  Changes  ......................................................................................................................................................  x   1   Introduction  ................................................................................................................................................  1   2   Literature  Review  .....................................................................................................................................  3   2.1   Power  system  and  network  configuration  ............................................................................  3   2.1.1   Layout  of  grid  ............................................................................................................................  3   2.1.2   Feeder  Voltages  ........................................................................................................................  3   2.1.3   Voltage  Correction  ..................................................................................................................  4   2.2   Electric  Vehicles  ................................................................................................................................  5   2.2.1   EV,  PHEV,  Extended  Range  EV  ...........................................................................................  5   2.2.2    Configuration  ...........................................................................................................................  6   2.2.3    Battery  system  .........................................................................................................................  6   2.2.4   Charging  ......................................................................................................................................  7   2.2.5   Growth  ..........................................................................................................................................  8   2.3   Impacts  of  Charging  .........................................................................................................................  9   2.3.1     Uncoordinated  Charging  ......................................................................................................  9   2.3.2    Coordinated  Charging  .........................................................................................................  10   2.4   Summary  ............................................................................................................................................  12   3   Methodology  .............................................................................................................................................  13   3.1   Load  Flow  ..........................................................................................................................................  13   3.1.1   Load-­‐Flow  Solutions  .............................................................................................................  13   3.1.2   Load  Types  ...............................................................................................................................  13   3.2   Modelling  ...........................................................................................................................................  15   3.2.1   DIgSILENT  PowerFactory  Models  ..................................................................................  15   3.2.2   DIgSILENT  Programming  Language  (DPL)  Script  ...................................................  18  
  • 6. v       v       3.2.3   Load  Profiles  ............................................................................................................................  18   3.2.4   Loading  Assumptions  ..........................................................................................................  20   3.2.5   Load  Scaling  .............................................................................................................................  21   3.3   Simulation  ..........................................................................................................................................  26   3.3.1   Graphical  User  Interface  .....................................................................................................  26   3.3.2   GUI  Structure  ...........................................................................................................................  27   3.4   Scenarios  ............................................................................................................................................  29   3.4.1   Uncoordinated  Charging  ....................................................................................................  29   3.4.2   Coordinated  Charging  ..........................................................................................................  30   3.4.3   11  kV  ...........................................................................................................................................  31   4   Results  .........................................................................................................................................................  33   4.1   Base  Load  Profile  ............................................................................................................................  33   4.1.1   Effects  of  Temperature  on  Substation  Loading  ........................................................  33   4.1.2   Load  Scaling  .............................................................................................................................  33   4.1.3   Network  Type  .........................................................................................................................  34   4.2   Uncoordinated  Charging  .............................................................................................................  35   4.2.1   11  kV  Voltage  Regulation  ...................................................................................................  35   4.2.2   400  V    Transformer  and  Feeder  Loading  ....................................................................  36   4.2.3   11  kV  Transformer  Loading  ..............................................................................................  42   4.3   Coordinated  Charging  ...................................................................................................................  43   4.3.1   3-­‐Group  Charging  ..................................................................................................................  43   4.3.2   Six-­‐Group  Charging  ...............................................................................................................  45   4.3.3   11  kV  ...........................................................................................................................................  45   5   Conclusion  .................................................................................................................................................  46   References  ...........................................................................................................................................................  48   Appendix  A  ..........................................................................................................................................................  51   Appendix  B  ..........................................................................................................................................................  54   Appendix  C  ..........................................................................................................................................................  55                        
  • 7. vi     vi       List  of  Figures         Figure  2.1:  Radial  Feeder  Distribution  ......................................................................................................  3   Figure  2.2:  Feeder  Voltage  Profiles  ............................................................................................................  4   Figure  2.3:  Electric  Vehicle  Configuration  ...............................................................................................  6   Figure  2.4:  Lithium-­‐Ion  Charge  Curve  [26]  .............................................................................................  8   Figure  3.1:  Load  Flow  Analysis  [4]  ...........................................................................................................  13   Figure  3.2:  400  V  Overhead/Underground  DIgSILENT  Model  .....................................................  16   Figure  3.3:  11  kV  Overhead  DIgSILENT  Model  ....................................................................................  17   Figure  3.4:  Average  number  of  travellers  in  NSW  on  weekdays  in  2010/11  ........................  19   Figure  3.5:  Scaled  driver  arrival  times  ....................................................................................................  20   Figure  3.6:  Feeder  voltage  profile,  moving  from  last  premise  to  transformer  from  right  to   left  ...........................................................................................................................................................................  23   Figure  3.7:  MATLAB  GUI  ...............................................................................................................................  26   Figure  3.8:  Flowchart  displaying  the  interaction  of  programs  required  for  GUI   simulations  ..........................................................................................................................................................  27   Figure  4.1:  Woodlands  Drive  substation  loading  for  38.7  and  19.9  degrees  celsius  days33   Figure  4.2:  Woodlands  Drive  substation  total  load  compared  to  scaled  sample  loads  .....  34   Figure  4.3:  Woodlands  Drive  substation  load  for  overhead  and  underground  networks34   Figure  4.4:  Impact  of  increasing  charger  rating  on  undergroudn  network  at  100%  EV   penetration  .........................................................................................................................................................  40   Figure  4.5:  4  kW  three-­‐group  coordinated  charging  for  different  transformer  base  levels  ..................................................................................................................................................................................  43   Figure  4.6:  Six-­‐group  coordinated  charging  for  a  95%  loaded  transformer  ..........................  45      
  • 8. vii     vii       List  of  Tables       Table  2.1:  Current  EV  Battery  Capacities  [11][13-­‐16]  .......................................................................  7   Table  2.2:  International  EV  Charging  Standards  ..................................................................................  7   Table  3.1:  Network  Equipment  Parameters  .........................................................................................  17   Table  3.2:  Variable  Options  Structure  .....................................................................................................  27   Table  4.1:  Woodlands  Drive  substation  transformer  loading  and  voltage  regulation  for   varying  EV  penetrations  ................................................................................................................................  36   Table  4.2:  Maximum  EV  penetration  for  4  kW  LV  uncoordinated  charging  ...........................  38   Table  4.3:  Maximum  EV  penetration  for  7  kW  LV  uncoordinated  charging  ...........................  39   Table  4.4:  Maximum  EV  penetration  for  10  kW  LV  uncoordinated  charging  ........................  40   Table  4.5:  Maximum  EV  penetration  at  zone  substation  assuming  worst  loading  day  in   2010/11  ...............................................................................................................................................................  42   Table  4.6:  Maximum  EV  penetration  for  7kW  LV  coordinated  charging  .................................  44   Table  4.7:  Maximum  EV  penetration  for  10  kW  LV  coordinated  charging  .............................  44                        
  • 9. viii     viii       List  of  Equations       Equation  3.1  ........................................................................................................................................................  14   Equation  3.2  ........................................................................................................................................................  14   Equation  3.3  ........................................................................................................................................................  21   Equation  3.4  ........................................................................................................................................................  22   Equation  3.5  ........................................................................................................................................................  23   Equation  3.6  ........................................................................................................................................................  23   Equation  3.7  ........................................................................................................................................................  24   Equation  3.8  ........................................................................................................................................................  24   Equation  3.9  ........................................................................................................................................................  24   Equation  3.10  .....................................................................................................................................................  24   Equation  3.11  .....................................................................................................................................................  25   Equation  3.12  .....................................................................................................................................................  25   Equation  3.13  .....................................................................................................................................................  25                          
  • 10. ix     ix         Abbreviations  and  Symbols       EV     Electric  Vehicle   BEV     Battery  Electric  Vehicle   PHEV     Plug-­‐In  Hybrid  Electric  Vehicle   IC     Internal  Combustion   V2G     Vehicle  to  Grid   OLTC     On-­‐load  tap  changer   SC     Switched  capacitor   SoC     State  of  Charge   Li-­‐ion     Lithium  ion   NiMH     Nickel-­‐metal  hydride   PV     Photovoltaic   DC     Direct  current   AC     Alternating  current   pu     per  unit   𝑗𝑋     Reactance,  Ohms   𝑅     Resistance,  Ohms   𝑍     Impedance,  Ohms   𝑃     Power,  Watts   𝑉     Voltage,  Volts        
  • 11. x       x         List  of  Changes         Section   Statement  of  Changes   Page  Number   1   Removed  references  to  solar  and  V2G,  added  description   of  new  work   1,2   2.2   Removed  sentence  relating  to  V2G   5   2   Removed  Solar  section   -­‐   2.3.2   Removed  Solar  sub-­‐subsection   11   2   Removed  ‘V2G  Benefits’  section   -­‐   2.3.1   Added  analysis  of  loading  assumptions  in  literature   10   3   Replaced  Methodology  section   32   4   Replaced  Results  section   13  
  • 12. 1     1     1 Introduction       The  world  is  currently  experiencing  a  major  shift  in  the  way  energy  is  generated  and   consumed.  Pressing  issues  such  as  climate  change  and  declining  fossil  fuel  reserves  are   changing  the  way  people  think  about  the  environment.  Also,  technological  advances  are   allowing  renewable  generation  and  energy  storage  to  become  technically  and   economically  viable,  paving  the  way  for  an  emissions  free  future.   Electric  vehicles  (EV)  and  plug  in  hybrid  electric  vehicles  (PHEV)  (used   interchangeably  in  this  text)  are  becoming  increasingly  popular  due  to  the  impetus  of   these  factors.  Significant  advances  in  battery  storage  capabilities  are  allowing  EVs  to   become  a  viable  alternative  to  internal  combustion  (IC)  vehicles.  Their  storage  of   electricity  allows  energy  to  be  sourced  from  renewable  sources  such  as  wind  and  solar,   allowing  for  zero  emission  driving.  This  is  significant,  as  it  would  play  a  large  role  in   reducing  CO₂  emissions  and  localised  air  pollution  levels  [1].    Without  proper  planning,  however,  EVs  are  expected  to  produce  undesired   impacts  on  the  low  voltage  distribution  network  when  charged  in  an  uncoordinated   manner.  Charging  will  occur  whenever  convenient  for  the  driver,  such  as  on  arrival   home  from  work,  increasing  the  evening  peak  load  and  causing  stress  to  network   equipment,  particularly  at  distribution  levels.  Due  to  the  large  amount  of  energy  drawn   during  charging  periods,  it  is  expected  that  at  high  penetration  levels  this  will  present   serious  power  quality  issues  for  the  grid,  including  potential  transformer  overloading   and  voltage  sags,  resulting  in  outages,  equipment  damage  and  energy  loss  [2][3].   This  outcome  may  be  avoided  if  electric  vehicle  charging  can  be  coordinated  in   such  a  way  to  avoid  the  evening  load,  and  instead  be  automated  for  charging  during  low-­‐ demand  periods,  such  as  late  at  night.  Smart  infrastructure  currently  being   contemplated  will  allow  charging  times  to  be  staggered  between  different  households  to   allow  a  more  evenly  distributed  feeder  load.   The   proposed   focus   of   this   thesis   is   to   investigate   the   impact   of   introducing   a   significant   number   of   EVs   on   the   residential   distribution   system,   particularly   during   uncoordinated  charging  periods  that  coincide  with  peak  load.  The  load  flow  simulation   package  DIgSILENT  PowerFactory  will  be  used  to  carry  out  the  investigations.  Means  of   avoiding   the   undesirable   impacts   of   EV   charging   will   be   investigated,   using   several  
  • 13. 2     2     scenarios   to   determine   the   viability   of   load   levelling.   This   study   will   determine   the   effects   of   charging   on   residential   feeder   voltage   levels,   consequently   discerning   the   associated  impacts  on  transformer  loading  and  energy  loss.   In  order  to  study  the  impacts  of  charging  on  the  residential  distribution  network,   typical  400V  and  11  kV  radial  residential  feeders  have  been  modelled  in  PowerFactory,   using  smart  metering  data  from  premises  in  the  Endeavour  Energy  network  area  of   Glenmore  Park.  Associated  variables  have  been  accounted  for,  including  battery   capacities,  charging  power,  base  load  demand,  load  power  factor  and  phase  unbalance.   To  aid  network  planners  in  making  decisions  based  on  future  electric  vehicle  loading,  a   graphical  user  interface  has  been  developed  using  MATLAB  GUIDE.  This  allows   DIgSILENT  PowerFactory  to  be  controlled  remotely  to  run  various  EV  loading  scenarios,   displaying  transformer  loading  and  voltage  regulation  results  both  numerically  and   graphically  for  analysis.                                          
  • 14. 3     3     2 Literature  Review     2.1 Power  system  and  network  configuration     2.1.1   Layout  of  grid     The  electricity  grid  is  a  complex  network  that  acts  as  a  path  for  electricity  from  generators   to  consumers.  The  layout  of  the  grid  is  an  important  concept  that  must  be  understand  to   grasp  an  idea  of  how  electric  vehicles  will  be  connected  and  the  effects  that  they  will  have  on   the  network.    The  traditional  grid  can  be  divided  into  generation,  transmission  and  distribution   levels.  The  transmission  network  steps  generator  voltages  up  in  order  to  reduce  the  losses   associated  with  high  currents  over  long  distances,  usually  at  230  kV  to  765  kV  [4].  As  these   high  voltage  feeders  branch  towards  large  populations,  they  are  stepped  down  in  to  the   distribution  network.  Zone  substations  convert  voltages  to  11  kV  for  residential  feeders,   which  then  connect  to  pole  top  or  pad  mount  transformers  that  finally  supply  400  V,  or       230  V  line-­‐to-­‐neutral,  for  use  in  homes  and  businesses  [5].   The  distribution  network  is  the  most  important  section  of  the  grid  to  understand  when   conducting  load  flow  analysis  on  residential  loads,  as  EVs  and  distributed  generation,  such   as  solar  PV,  are  both  connected  at  the  low  voltage  level.  From  zone  substations,  feeders  are   typically  connected  radially  [6][4]  as  they  branch  out  through  streets,  shown  in  Fig  2.1.  This   radial  layout  will  be  used  for  modelling  residential  feeders.     Figure  2.1:  Radial  Feeder  Distribution   2.1.2   Feeder  Voltages     Basic  circuit  theory  states  that  a  voltage  drop  will  result  as  current  flows  through  an   impedance.  Therefore,  as  transformer  loading  is  increased,  the  voltage  drop  along  a   feeder  becomes  greater.  Conversely,  during  periods  of  high  generation,  net  feeder  
  • 15. 4     4     current  is  reduced,  raising  voltage  levels  closer  to  that  of  the  transformer.    During  heavy   loading  or  generation  periods,  voltage  levels  may  surpass  utility  limits.  The  AS/NZS   3000:2007  states  that  in  Australia,  voltage  limits  must  not  move  beyond  +10%  or  -­‐6%  of   nominal  value  to  avoid  damage  to  connected  equipment,  corresponding  to  253  V  and   216  V  line-­‐to-­‐neutral  [5].    Fig.  2.2  shows  the  effects  of  different  load  scenarios  on  feeder   voltage  levels.  Realistically,  these  voltages  would  not  have  a  linear  profile,  even  for   uniform  loading  across  the  feeder,  as  currents,  and  hence  the  rate  of  voltage  drop,  is   greater  closer  to  the  transformer.       Figure  2.2:  Feeder  Voltage  Profiles   Another  consequence  of  voltage  deviations  along  feeders  is  power  loss.   Feeder  power  loss  is  proportional  to  the  square  of  a  voltage  change,  therefore  it  is   important  to  reduce  this  change  in  voltage  along  a  feeder  as  much  as  possible.     2.1.3   Voltage  Correction     Voltage  control  is  important  for  addressing  changes  in  line  voltages.  Network   equipment,  such  as  transformers  and  lines  are  designed  to  operate  within  certain   voltage  limits.  Most  importantly,  however,  are  the  loads  connected  to  LV  feeders,  which   may  become  damaged  while  drawing  power  at  excessive  or  limited  voltage  levels.     In  order  to  maintain  voltage  levels  within  a  specified  range  such  as  this,  a  range   of  network  equipment  is  utilised.  In  distribution  networks,  voltage  control  is  typically   achieved  using  on-­‐load  tap  changers  (OLTC),  step  voltage  regulators  (SVR)  and  switched   capacitors  (SC)  [7].  OLTCs  and  SVRs  are  both  autotransformers  with  automatic  tap   changing.  Normally  the  voltage  regulator  in  a  substation  is  an  OLTC,  while  an  SVR  would   be  located  along  a  feeder,  down  to  LV  levels  [7].    SCs  are  used  for  reactive  power   compensation  in  distribution  networks.  An  SC  reduces  the  displacement  between  real   and  reactive  power  components  to  reduce  voltage  drop  across  lines  that  are  primarily  
  • 16. 5     5     inductive.  In  low  voltage  networks,  the  most  common  voltage  regulators  are  off-­‐load   tap-­‐changers,  located  within  distribution  transformers  [8].  The  transformer  ratio  must   be  changed  manually,  generally  over  a  multiple  year  span  as  network  loading  increases.   Although  SVRs  and  switched  capacitors  can  exist  in  LV  areas,  this  is  uncommon  due  to   the  large  number  of  feeders,  and  the  associated  costs.   Therefore,  on  residential  feeders,  voltage  control  is  limited  to  off-­‐load  tap   changers  on  pole-­‐top  and  pad  mount  transformers.  The  manual  nature  of  this  tap   changing  is  uncoordinated,  therefore  this  is  far  from  being  an  optimal  solution  to   addressing  the  large  scale  integration  of  EVs.   Taking  the  characteristics  of  common  network  equipment  into  account,  the  coordinated   charging  of  EVs  can  be  seen  as  a  worthwhile  solution  to  this  problem  as  the  load  factor   of  a  feeder  may  be  reduced.     2.2   Electric  Vehicles     Electric  vehicles  are  vehicles  that  contain  a  rechargeable  battery  pack,  requiring   charging  by  a  grid  connected  battery  charger.  EVs  are  becoming  popular  as   environmental  awareness  is  increasing  across  the  world,  as  they  produce  little  to  no   emissions.  Improvements  in  battery  technology  are  seeing  prices  fall  rapidly,  allowing   EVs  to  become  a  viable  alternative  to  internal  combustion  (IC)  vehicles.  Penetration  of   EVs  is  beginning  to  increase,  with  over  20  models  due  to  reach  the  markets  in  2012  [9].     2.2.1   EV,  PHEV,  Extended  Range  EV     There  are  four  main  types  of  electric  vehicles  that  currently  exist:  Hybrid,  Plug-­‐in  Hybrid   (PHEV),  Extended-­‐Range  and  Battery  EVs  (BEV)  [10].  Hybrid  and  PHEVs  contain  both   combustion  engines  and  electric  motors  with  battery  storage.  Unlike  hybrids,  however,   PHEVs  can  also  be  charged  through  an  external  battery  charger,  further  reducing   reliance  on  the  combustion  engine  [10]  Extended-­‐Range  EVs  are  similar  to  PHEVs  and   include  vehicles  such  as  the  Holden  Volt  [11].  The  electric  engine  is  used  for  all  driving   speeds  until  the  battery  is  discharged,  and  is  then  replaced  by  the  combustion  engine.   Lastly,  BEVs  are  all  electric  with  no  combustion  engine.  They  contain  large  battery  packs   that  must  be  charged  by  the  grid.    
  • 17. 6     6     In  relation  to  the  topic  of  this  thesis,  hybrid  vehicles  are  considered  irrelevant,  as   they  are  not  charged  by  the  grid.  Therefore,  the  vehicles  of  focus  will  be  PHEVs,   Extended-­‐Range  EVs  and  BEVs,  referred  to  collectively  throughout  this  text  as  ‘EVs’.     2.2.2    Configuration     The  basic  configuration  of  an  EV,  including  an  IC  engine,  which  is  only  applicable  to   PHEVs  and  EREVs,  is  shown  by  the  simplified  block  diagram  in  Fig.  2.3.     Figure  2.3:  Electric  Vehicle  Configuration     Charging  requires  communication  with  the  battery-­‐monitoring  unit  that  measures  the   batteries  state  of  charge  (SoC).  The  inverter  is  used  after  a  DC-­‐DC  converter  to  convert   direct  current  (DC)  into  alternating  current  (AC)  to  power  the  electric  motor.     2.2.3    Battery  system     For  electric  vehicles  to  be  a  viable  alternative  to  IC  vehicles,  their  battery  storage  must   contain  enough  energy  to  ensure  suitable  range  for  drivers.  The  most  important  factor   affecting  this  is  the  energy  to  weight  ratio  of  a  battery  pack,  or  its  energy  density.  This   allows  vehicles  to  be  as  light  as  possible  for  a  given  amount  of  energy  storage,  ensuring   the  greatest  range  possible.   There  exist  three  main  battery  types  for  electric  vehicles:  lead-­‐acid,  nickel-­‐metal   hydride  (NiMH)  and  lithium-­‐ion  (li-­‐ion)  [12].  In  the  past,  EVs  such  as  the  General  Motors   EV1  used  lead-­‐acid  and  nickel-­‐metal  hydride  batteries.  In  recent  years,  however,  the   demand  for  batteries  in  laptops  and  other  portable  devices  has  driven  R&D  in  the  area  of   lithium-­‐ion  batteries,  improving  energy  density  and  charge  time  beyond  other  battery   types.  Due  to  these  improvements,  major  EV  manufacturers  now  use  lithium  ion  battery   packs  [11][13-­‐16].  
  • 18. 7     7     Table  2.1  provides  a  list  of  current  vehicles  and  their  battery  capacities,  showing  a   significant  range  of  battery  capacities  that  will  form  the  basis  for  modelling.     Electric  Vehicle   Battery  Capacity   Tesla  Model  S   40,  60,  85  kWh   Nissan  Leaf   24  kWh   Ford  Focus  Electric   23  kWh   Holden  Volt   8  kWh   Toyota  Prius  Plug-­‐In   4.4  kWh   Table  2.1:  Current  EV  Battery  Capacities  [11][13-­‐16]   2.2.4   Charging     Based  on  standards  by  the  International  Electrotechnical  Commission  (IEC)  [17]  and  the   Society  of  Automotive  Engineers  J1772  [18],  there  exists  three  charging  levels:     Level   Voltage     Current   Power   1   120  V  AC   16  A   1.92  kW   2   208-­‐240  V  AC   12  –  80  A   2.5  –  19.2  kW   3   500  V  DC   125  A   50  kW   Table  2.2:  International  EV  Charging  Standards     The  residential  charger  rating  of  EV  manufacturers  vary  substantially  within  the  Level  2   range.  Nissan  and  Holden’s  chargers  are  rated  3.3  kW  [16][11],  Ford’s  at  7.7  kW  [14],   while  Tesla  manufactures  10  kW  or  20  kW  chargers  [13].  These  ratings  are  significant  in   comparison  to  other  appliances  found  in  the  home.   Fig.  2.4  shows  the  power  demand  and  battery  SoC  profiles  of  a  lithium  ion   battery.  
  • 19. 8     8       Figure  2.4:  Lithium-­‐Ion  Charge  Curve  [26]     Figure  2.4  shows  a  predominantly  constant  charging  power  for  the  duration  of  the   charging  period.  Therefore,  for  modelling  purposes,  a  constant  charge  rate  can  be   considered  accurate  to  assume.     2.2.5   Growth     Due  to  economic  and  technological  factors  surrounding  the  viability  of  electric  vehicles,   their  penetration  levels  are  expected  to  soar  this  decade  [19-­‐21].  Current  estimates   expect  the  price  of  oil  to  rise  by  85%  into  2020  [19],  and  this  rise  is  forecast  to  continue.   By  the  same  time,  lithium  ion  battery  technology  is  expected  to  dramatically  fall  as   economies  of  scale  reduces  manufacturing  costs,  and  technological  improvements  allow   energy  density  to  continually  increase.  Lithium  ion  battery  prices  have  fallen   considerably  from  US$650/kWh  in  2009  to  the  current  price  of  around  US$450/kWh.     Analysts  have  forecasted  prices  to  fall  at  a  7.5%  annual  compound  rate  from  2012   through  2020  to  approximately  US$250/kWh  [19].  EV  manufacturer  Tesla  Motors  is   already  producing  battery  packs  with  480  km  of  range  [13].   Taking  these  factors  into  consideration,  analysts  from  Deutsche  Bank  [19]  have   predicted  that  in  the  US,  around  10%  of  all  vehicles  will  be  hybrid/electric  by  2021,   increasing  to  20%  by  2026,  and  35%  by  2030.  In  terms  of  purchased  vehicles,  EVs  are   expected  to  make  up  3-­‐10%  of  new  car  sales  as  early  as  2015  [20]  and  35%  in  2025,   comprised  of  25%  PHEVs  and  10%  EVs,  according  to  IDtechX  analysts  [21].  These   projections  show  that  a  major  shift  is  about  to  occur,  resulting  in  a  significant   percentage  of  vehicles  becoming  at  least  partially  electric.  This  analysis  raises  questions   about  the  effects  of  a  large  percentage  of  EVs  on  the  distribution  network,  as  well  as  the  
  • 20. 9     9     potential  problems  this  extra  energy  storage  may  solve.     2.3   Impacts  of  Charging     2.3.1     Uncoordinated  Charging     The  introduction  of  EVs  is  expected  to  have  a  significant  effect  on  customer  load  profiles   during  charging  periods.  Studies  in  [2],  [3]  and  [22]  have  concluded  that,  for  high   penetration  levels,  uncoordinated  domestic  charging  will  increase  peak  load  demand   significantly,  resulting  in  transformer  overloading,  poor  feeder  voltage  profiles  and   power  loss.       The  authors  of  [2]  and  [22]  have  conducted  studies  on  uncoordinated  charging   on  residential  radial  feeders,  focusing  on  evening  peaks.  The  modelled  charger  rating   was  4  kW  [2],  and  1.8  kW  in  [22],  both  showing  dramatic  rises  in  peak  load,  clearly   overloading  the  transformer  limitations  for  penetrations  above  20%  in  [22]  and   exceeding  voltage  limits  in  [2]  at  17%.  The  effects  of  peak-­‐time  charging  on  summer  and   winter  load  profiles  are  explored  in  [23]  and  [3].  The  UK  winter  load  profile  in  [23]   showed  a  distinct  evening  peak  compared  to  summer  due  to  electric  heating.  This   caused  the  peak  demand  to  be  increased  by  13.6%  compared  to  10.06%  for  summer  at   10%  EV  penetration.  Although  this  paper  conducts  a  load  study  for  the  entire  UK,  it  is   probable  that  this  would  reflect  the  demand  of  residential  feeders,  as  most  vehicles   would  be  at  home  during  this  period.  A  study  is  conducted  in  [3]  to  determine  the  effects   of  peak  charging  on  power  loss  and  voltage  deviation.  The  voltage  limit  of  0.9  pu  was   found  to  be  exceeded  at  30%  EV  penetration  with  a  4  kW  charger,  with  total  power  loss   at  6%  in  winter  compared  to  5%  in  summer.   These  papers  clearly  show  that  uncoordinated  charging  would  have  a  large   impact,  even  at  low  penetration  levels.  However,  an  analysis  of  these  papers  show  the   large  number  of  variables  associated  with  such  studies.  For  example,  the  voltage  limit  of   0.9pu  in  [3]  differs  to  0.94  used  in  Australia,  as  well  as  the  UK  load  profiles  in  [23].   Another  assumption  made  in  these  studies  is  a  relatively  low  powered  charger,   particularly  in  [22].  A  higher-­‐powered  charger  more  commonly  used  today  would  have  a   significantly  increase  the  peak  demand  determined  by  these  papers.  Of  all  the   assumptions  made,  however,  the  most  important  variable  used  to  determine  the  impacts  
  • 21. 10     10     of  uncoordinated  charging  is  the  time  the  vehicles  arrive  home  to  begin  charging.  In  the   related  papers  [2-­‐3]  [24-­‐26],  and  number  of  assumptions  in  relation  to  charging  times   have  been  made,  while  there  exists  a  significant  degree  of  ambiguity  when  these   assumptions,  such  as  the  data  used,  is  explained.  Papers  [2]  and  [22]  fail  to  explain  how   their  vehicle  arrival  times  are  modelled,  while  [23]  simply  divides  charging  into  three   groups  during  the  evening  peak,  assuming  that  all  vehicles  commence  charging  within   90  minutes  of  one  another.  Papers  [24]  and  [27]  assume  a  more  accurate  normal   distribution,  however  still  disregard  actual  driving  statistics,  such  as  those  provided  by   the  UK  Time  of  Use  survey  noted  in  [23]  and  [3].  Paper  [3]  takes  into  account  the   statistics  from  this  survey  by  dividing  charging  times  according  to  the  morning,  midday   and  late  afternoon  periods,  and  making  assumptions  about  the  percentage  of  cars  that   charge  during  these  times.  Paper  [3]  applies  the  most  accurate  data  regarding  charging   times  as  it  incorporates  the  irregular  and  skewed  peak  provided  by  a  traffic  authority.   Considering  this,  the  majority  of  research  has  been  conducted  with  inaccurate   assumptions,  possibly  causing  significant  variations  in  results  as  the  charging  times,   along  with  the  assumed  charger  rating,  are  the  factors  that  most  influence  the  results  of   loading  simulations.  Charging  times  for  the  uncoordinated  charging  simulations  in  this   thesis  will  be  based  on  local  driving  data  to  ensure  the  most  accurate  modelling  possible.       Therefore,  to  more  accurately  determine  the  effects  of  uncoordinated  charging,  it   is  important  to  use  local  load  profiles,  standards  and  driving  statistics,  with  assumptions   that  are  up  to  date,  or  reflect  expected  future  trends.  These  variables  will  be  taken  in  to   account  in  this  thesis,  to  more  accurately  determine  possible  effects  on  typical   Australian  residential  feeders.     2.3.2    Coordinated  Charging     The  effects  of  uncoordinated  charging  show  the  importance  of  coordinated  or   ‘smart’  charging  in  the  future.  This  would  be  achieved  through  communication   infrastructure  in  a  smart  grid,  by  sending  signals  to  begin  charging  at  times   corresponding  to  uniform  loading  [24].  Coordinated  charging  employs  heuristic   algorithms  and  optimization  techniques  with  the  aim  to  improve  load  factor  and  reduce   network  costs  and  power  losses  by  charging  during  off  peak  periods  [2][24].  As  cars  are   available  for  94.8%  of  the  day  on  average  [23],  coordinated  charging  can  be  considered   viable,  as  a  large  amount  of  flexibility  exists  in  charging  times.    
  • 22. 11     11     A  large  number  of  studies  have  been  conducted  on  novel  approaches  to   coordinating  vehicles,  with  the  aim  to  reduce  evening  peak  demand.  These  range  from   complicated  algorithms  based  on  real-­‐time  market  prices  in  [27]  to  prioritizing  charging   periods  in  [2],  to  simple  delayed  off-­‐peak  charging  in  [23].   Throughout  the  majority  of  coordinated  charging  studies,  the  uncertainties  of  variables,   such  as  load  profiles  and  charging  time,  are  expressed  in  terms  of  probability  density   functions,  allowing  predictions  to  be  made  without  relying  on  fixed-­‐input  variables,  such   as  an  average  past  load  profile  [27].  The  authors  in  [27]  determined  that  coordinated   charging  reduced  load  factor  and  power  losses  by  6-­‐28%  for  penetration  levels  from   10%  to  100%.    In  [27],  a  control  algorithm  was  implemented  for  coordinated  charging   on  an  LV  feeder  in  Belgium,  based  on  a  typical  local  load  profile.  The  results  showed  a   peak  demand  reduction  of  29%  for  a  combination  of  3.6  kW  and  7.4  kW  chargers  at  15%   penetration.    Papers  [2]  and  [27]  take  different  real-­‐time  approaches,  dividing  charging  times   into  red,  blue  and  green  zones,  based  on  the  priority  of  charging.  In  [27],  charging   priority  is  determined  based  on  the  time  vehicles  arrive  home,  as  a  vehicle  that  arrives   late  would  have  a  low  chance  of  being  used  for  the  remainder  of  the  night.  This  paper   found  that  load  demand  could  remain  below  the  evening  peak  for  penetration  levels  of   at  least  63%,  as  low  priority  vehicles  could  be  spread  further  into  the  morning  hours.   Above  this  penetration,  however,  this  paper  found  that  high  and  medium  priority   vehicles  raised  the  peak  demand  above  the  evening  peak,  therefore  stating  there  will   inevitably  be  a  rise  in  peak  demand  as  EV  penetration  reaches  high  levels.   The  study  in  [27]  assumes  a  2  kW  peak,  which  is  relatively  low,  especially  as  this   aims  to  determine  loading  decades  in  to  the  future,  which  is  expected  to  rise  irrespective   of  EVs.  Another  assumption  is  that  low  priority  charging  is  timed  to  finish  at  4  am,   however  this  could  realistically  be  increased  to  6  am,  for  example,  for  the  majority  of   people  who  leave  for  work  after  this  time.  This  would  allow  a  higher  penetration  before   peak  demand  is  raised.   The  authors  in  [23]  have  included  a  study  on  fixed  off-­‐peak  charging,  which  is   implemented  by  simply  charging  in  three  groups,  at  9  pm,  9:30  pm  and  10  pm.  This   avoids  the  evening  peak,  while  allowing  sufficient  time  to  charge  through  to  early   morning.  This  paper  finds  that  the  charging  peak  is  less  than  the  evening  peak  for  low   penetration,  but  states  that  this  may  not  be  the  case  for  penetration  greater  than  10%.   This  is  compared  to  a  study  on  ‘smart’  market  based  charging,  which  shows  a  noticeable  
  • 23. 12     12     reduction  in  charging  peak  load.  From  analysis  of  the  fixed  off-­‐peak  charging  graph,  it   shows  charging  is  finished  by  2  am.  This  shows  a  large  percentage  of  early  morning   hours  with  lower  base  demand  that  are  not  utilized,  therefore  it  could  be  argued  that   this  method  could  support  penetration  much  higher  than  the  10%  stated.  The  simplicity   of  the  fixed  off-­‐peak  method,  and  the  lack  of  research  associated  with  it,  presents  an   opportunity  for  study  in  this  thesis.  This  would  eliminate  the  need  for  complicated   algorithms  at  residential  feeders,  and  may  not  require  smart  infrastructure,  as  signalling   could  be  sent  via  high  frequency  pulses,  as  they  are  today  to  control  off-­‐peak  hot  water   systems.  Lower  electricity  rates  would  provide  the  incentive  for  the  majority  of  owners   to  use  this  method,  while  allowing  a  simple  manual  over-­‐ride  when  required.  However,   in  terms  of  load  levelling,  coordinated  charging  would  be  a  valuable  approach  to  further   reduce  energy  losses.  Initially,  this  method  will  be  tested  by  simulating  a  simple  fixed-­‐ start  delay,  with  preliminary  work  on  staggered  charging  to  focus  on  further  reducing   power  loss.     2.4   Summary     The  results  of  various  studies  related  to  charging  produce  a  wide  range  of  results  due  to   the  number  of  variables  associated  with  distribution  networks  and  electric  vehicles.   From  this  analysis,  a  noticeable  gap  exists  in  research  of  the  impact  of  EVs  applicable  to   Australian  residential  feeders.  Particularly,  there  is  a  lack  of  study  that  incorporates   realistic  driving  pattern  data,  through  either  the  use  of  information  from  traffic   authorities  or  by  conducting  surveys.                  
  • 24. 13     13     3 Methodology     The  study  of  literature  in  Chapter  2  presents  a  number  of  areas  that  can  be  further   studied  to  determine  the  impacts  of  EV  charging.  Further  study  would  gain  valuable   information  for  electricity  distribution  network  service  providers  in  planning  for  future   development,  as  well  determine  the  benefits  for  residents.   3.1 Load  Flow     3.1.1 Load-­‐Flow  Solutions     To  determine  loading  effects  in  the  context  of  an  Australian  residential  feeder,  load  flow   analysis  must  be  conducted.  A  simple  single-­‐line  diagram  can  be  realized  in  Fig.  3.1.       Figure  3.1:  Load  Flow  Analysis  [4]     Figure  3.1  represents  a  simple  power-­‐flow  scenario.  Power-­‐flow  problems  such  as  this   are  separated  in  to  the  following  components:   1. Slack  bus  –  a  reference  bus  for  which  V∠δ°  =  1.0∠0°   2. Load  (PQ)  bus  –   𝑃!  and   𝑄!  are  input  loads,  used  to  compute   𝑉!  and  δ!   3. Voltage  controlled  (PV)  bus  –   𝑃!  and   𝑉!  are  inputs,  includes  voltage  control   devices  such  as  OLTC,  switched  capacitors   The  power  flow  data  listed  is  used  to  calculate  power-­‐flow  solutions  using  methods  such   as  Guass-­‐Seidell  and  Newton-­‐Raphson,  which  solve  nodal  equations  iteratively  [7].     3.1.2 Load  Types   Another  important  consideration  that  must  be  made  during  load  flow  analysis  is  the   type  of  load  connected  to  each  load  bus.  Load  behaviour  is  determined  by  the  
  • 25. 14     14     combination  of  R,  L  and  C  elements  and  power  electronic  circuitry  of  a  load,  and  can  be   divided  into  three  types:   1. Constant  Power  (eg.  LED  TV,  computer)   2. Constant  Current  (eg.  CFL  globe)   3. Constant  Impedance  (eg.  Toaster,  oven)   Therefore,  for  any  given  voltage  a  load  will  conform  to  one  of  these  load  behaviours.  An   appliance  with  a  power  electronics  interface,  for  example,  with  exhibit  constant  power   characteristics  as  the  voltage  is  stepped  down  and  held  at  a  constant  DC  value,  as  this   voltage  will  be  constant  for  all  AC  source  voltage  levels.  A  resistive  load,  on  the  other   hand,  is  regarded  as  constant  impedance  and  will  draw  less  power  as  voltage  levels   drop,  according  to  Ohm’s  law.   The  voltage  dependency  of  loads  can  be  modelled  by  Eqs.    (3.1)  and  (3.2):     𝑃 = 𝑃!(𝑎𝑃 ∙ 𝑣 𝑣! !_!" + 𝑏𝑃 ∙ 𝑣 𝑣! !_!" + (1 − 𝑎𝑃 − 𝑏𝑃) ∙ 𝑣 𝑣! !_!" )   (3.1)     Where  1 − 𝑎𝑃 − 𝑏𝑃 = 𝑐𝑃     𝑄 = 𝑄!(𝑎𝑄 ∙ 𝑣 𝑣! !_!" + 𝑏𝑄 ∙ 𝑣 𝑣! !_!" + (1 − 𝑎𝑄 − 𝑏𝑄) ∙ 𝑣 𝑣! !_!" )   (3.2)     Where  1 − 𝑎𝑄 − 𝑏𝑄 = 𝑐𝑄     When  modelling  a  house  load,  a  number  of  assumptions  have  to  be  made.    For  the   purpose  of  this  simulation,  a  house  will  be  considered  as  a  constant  power  load,  as  each   house  will  be  associated  with  load  profiles  recorded  on  a  hot  day  where  the   predominant  load  type  is  a  constant  power  air  conditioner.  EVs  will  also  be  regarded  as   constant  power  loads,  as  the  charging  profile  of  a  lithium  ion  battery  charger  is  a   constant  power  curve.  From  Eq.  (3.1),  we  simply  require  𝑃 = 𝑃!,  therefore  all   coefficients  and  exponents  have  been  set  to  zero  in  the  voltage  dependence  settings  of   each  load.      
  • 26. 15     15     3.2 Modelling     A  realistic  network  model  is  imperative  for  determining  the  effects  of  EV  charging.   DIgSILENT  PowerFactory  was  chosen  for  this  purpose  due  to  its  flexibility  in  analysis,   incorporating  functions  such  as  unbalanced  power  flow,  and  remote  control  ability   through  DIgSILENT  Engine.  To  ensure  that  loading  results  were  as  accurate  as  possible,   emphasis  was  placed  on  applying  accurate  network  modelling  parameters,  load  profiles   and  vehicle  driving  statistics.   3.2.1 DIgSILENT  PowerFactory  Models       In  order  to  accurately  model  a  typical  low  voltage  network,  data  from  smart  meter-­‐ connected  premises  has  been  accumulated.  The  premises  of  interest  are  connected  to  a   500  kVA  pad-­‐mount  distribution  substation  in  Woodlands  Drive,  Glenmore  Park   (located  in  Western  Sydney),  which  supplies  92  premises  on  four  low  voltage   underground  feeders.  The  network  models  used  for  simulation  are  based  off  sample   DIgSILENT  feeder  models  provided  by  Endeavour  Energy.  Three  models  –  400  V   overhead,  400  V  underground  and  11  kV  overhead  –  were  modified  to  supply  the  same   number  of  loads  as  the  substations  in  Glenmore  Park.   When  implementing  the  LV  models,  each  premise  is  represented  by  a  single-­‐ phase  house  and  EV  load,  with  a  CSV  file  associated  with  each  load  containing  the  load   profile  information  for  a  single  day.  Due  to  limitations  with  the  number  of  possible   nodes  in  a  PowerFactory  student  license,  the  number  of  premises  has  been  halved  to  46   premises  split  across  two  feeders,  supplied  by  a  250  kVA  transformer.  Halving   transformer  ratings  and  feeder  numbers  ensures  an  accurately  scaled  model  for   determining  feeder  voltage  levels  and  transformer  loading.  Modelling  loads  as  single   phase  loads  allows  for  voltage  unbalance  to  be  accounted  for,  which  is  a  primary  cause   of  poor  voltage  regulation.  The  low  voltage  overhead  model  is  shown  in  Fig.  3.2.      
  • 27. 16     16       Figure  3.2:  400  V  Overhead/Underground  DIgSILENT  Model   To  model  the  impacts  of  electric  vehicle  charging  on  a  zone  substation  at  the  11  kV  level,   the  resulting  distribution  transformer  load  profiles  have  been  lumped  and  applied  to   each  of  the  transformer  loads  on  a  single  11  kV  feeder.  The  loading  magnitude  is   doubled  to  account  for  the  halved  number  of  premises  on  the  low  voltage  side,  so  that   each  transformer  is  represented  accurately  at  500  kVA.  There  are  10  11  kV  feeders   supplied  by  Glenmore  Park  Zone  Substation,  which  supplies  a  total  of  7596  premises.   Glenmore  Park  Zone  Substation  has  2  x  45  MVA  transformers  installed,  and   hence  has  an  N-­‐1  capacity  of  45  MVA.  With  an  average  of  760  premises  per  11  kV  feeder,   assuming  92  premises  per  500  kVA  of  installed  capacity,  there  would  be  an  average  of  8   LV  substations  connected  to  each  11  kV  feeder.  Therefore,  8  LV  substation  loads  have   been  modelled  on  the  11  kV  feeder,  and  the  total  zone  substation  load  is  determined  by   multiplying  the  total  feeder  loading  by  10  feeders.  Figure  3.3  shows  the  11  kV  feeder   model.  
  • 28. 17     17       Figure  3.3:  11  kV  Overhead  DIgSILENT  Model       Parameters  such  as  line  and  transformer  impedances,  shown  in  Table  3.1,  were  left   constant  as  they  represent  the  most  common  ratings  used  within  the  Endeavour  Energy   network.     400  V  Overhead   400  V  Underground   11  kV  Overhead   Feeder  Impedance     0.707  +  j0.284   Ω/km   0.162  +  j0.065  Ω/km   0.224  +  j0.224   Ω/km   Feeder  Section   Length     35  m   35  m   570  m   Service  Line   Impedance     1.49  +  j0.097  Ω/km   0.927  +  j0.081  Ω/km   N/A   Service  Line  Length   20  m   20  m   N/A   Transformer  Rating   250  kVA   250  kVA   N/A   Transformer   Impedance     4%   4%   N/A   Voltage  Source   Series  Impedance   0.5  +  j5  Ω     0.5  +  j5  Ω   0.021  +  j0.635  Ω   Table  3.1:  Network  Equipment  Parameters   The  11  kV  model  assumed  a  voltage  source  at  1  pu  voltage,  as  opposed  to  a  transformer,   as  the  transformer’s  OLTC  would  act  to  maintain  this  voltage  in  reality.  The  low  voltage   transformers  modelled  are  equipped  with  offline-­‐tap  changers  with  6  asymmetrical  tap   settings,  ranging  from  -­‐4  to  +1.  At  typical  tap  setting  for  LV  transformers  is  -­‐3,  or  -­‐7.5%,   corresponding  with  a  LV  bus  voltage  of  430  V.  An  increase  in  each  tap  setting  will  raise   the  voltage  by  2.5%,  allowing  for  a  12.5%  voltage  range  (-­‐10%  to  +2.5%).  As  LV  
  • 29. 18     18     transformer  taps  are  offline,  they  must  be  changed  manually  and  hence  would  only  be   changed  over  the  long  term  as  total  loading  increases,  not  in  response  to  a  permanent   increase  in  the  afternoon  peak  caused  by  EV  charging,  for  example,  as  this  would  cause   voltages  to  exceed  their  upper  limits  during  lower  loading  periods.  Instead,  this   regulation  must  be  controlled  using  zone  substation  OLTC’s  which  allow  for  real-­‐time   tap  changing.  As  EV  loading  is  expected  to  only  increase  the  afternoon/evening  peak,  the   tap  setting  is  expected  to  remain  constant  in  the  future.  Although  there  may  be  future   base  load  growth  as  the  penetration  of  air  conditioners  and  other  electrical  appliances   increases,  the  relative  difference  between  low  loading  periods  and  afternoon  EV  loading   will  likely  remain  constant,  therefore  the  actual  future  LV  substation  tap  setting  can  be   disregarded.   3.2.2 DIgSILENT  Programming  Language  (DPL)  Script     A  DIgSILENT  Programming  Language  (DPL)  script  allows  the  automation  of  load  flows   to  extract  specific  data  from  a  network  model.  A  DPL  script  was  provided  by  Endeavour   Energy  which  conducts  time-­‐step  simulation  load  flows  for  house  loads,  saving  power,   losses  and  voltage  data  into  result  objects  at  half  hour  intervals.  This  script  was   modified  to  read  EV  loads,  as  well  as  execute  ‘export  result  objects’  so  that  result  data   would  be  exported  to  text  files  each  time  the  script  was  run.  The  DPL  script  was   associated  with  each  network  model,  and  allowed  load  flow  simulations  to  be  conducted   via  engine  control  of  PowerFactory.     3.2.3 Load  Profiles     3.2.3.1 House  Load  Profiles     Loads  in  PowerFactory  can  be  associated  with  CSV  files  containing  multiple  time  points   for  conducting  time-­‐step  simulations.  Each  of  the  42  house  loads  has  an  associated  CSV   file  containing  the  smart  metering  data  of  a  premise  in  the  Glenmore  Park  trial  area,   chosen  at  random  from  the  92  metered  premises.  The  smart  metering  data  contains  the   power  usage  of  the  premises  over  a  24  hour  period  at  half  hour  intervals.  Each  premise   has  been  assigned  the  same  power  factor,  determined  as  the  average  of  the  premises   power  factor  during  the  evening  hours,  found  to  be  0.9  inductive.  The  selected  load   profiles  correspond  with  the  hottest  day  of  2011,  occurring  on  November  14  at  a  
  • 30. 19     19     maximum  temperature  of  38.7°C.  The  hottest  day  of  2011  was  chosen  as  network   planning  must  take  into  account  the  worst-­‐case  loading  scenarios  that  occur  on  hot  days,   caused  primarily  by  air  conditioners.     3.2.3.2 EV  Charging  Profiles     The  spread  of  EV  charging  start  times  were  determined  by  analysing  driving  statistics   from  the  NSW  Bureau  of  Transport  Statistics  [28],  shown  in  Fig  3.4.     Figure  3.4:  Average  number  of  travellers  in  NSW  on  weekdays  in  2010/11   This  graph  shows  the  average  number  of  travellers  in  NSW  on  weekdays  by  transport   type  in  2010/11.  For  determining  vehicle  arrival  times,  only  the  ‘Vehicle  Driver’  curve   was  considered.  The  time  of  arrival  was  determined  by  shifting  the  afternoon/night   peak,  between  2  pm  and  12  am,  by  20  minutes  -­‐  the  average  vehicle  one-­‐way  trip  time.   This  curve  was  then  normalised  between  2  pm  and  12  am,  and  multiplied  by  46  to   determine  the  number  of  premises  that  would  begin  charging  at  each  half  hour  interval   within  this  period.  The  resulting  scaled  driving  arrival  curve  is  shown  in  Fig  3.5,  shown   to  follow  the  afternoon  driving  trend  displayed  in  Fig  3.4.  
  • 31. 20     20       Figure  3.5:  Scaled  driver  arrival  times   The  number  of  vehicles  arriving  at  each  half  hour  interval  was  recorded,  and  the   vehicles,  having  been  assigned  their  specific  starting  time,  were  allocated  to  premises   using  a  random  function,  so  that  the  feeder  models  were  assigned  a  realistic  variation  in   vehicle  arrival  times.     3.2.4 Loading  Assumptions       To  model  the  effects  of  charging,  the  level  two  residential  chargers  from  Chapter  2  were   considered.  Considering  the  expected  combination  of  chargers  based  on  EV  costs,  an   average  charge  rating  of  4  kW  was  determined  to  provide  a  realistic  charging  power  that   could  be  used  to  model  a  load  of  EV  charging  homes.  The  average  battery  capacity  was   chosen  to  be  25  kWh,  a  mid-­‐range  capacity  in  Table  2.1.   Assuming  a  return  trip  driving  distance  of  18.8  km  [28]  and  a  battery  consumption  of   0.168  kWh/km  [16],  the  average  charging  time  was  found  to  be  approximately  47   minutes.  Due  to  the  time-­‐step  resolution  of  half  an  hour,  however,  this  charging  duration   had  to  be  modelled  as  1  hour.  This  analysis  assumes  that  each  EV  is  charged  only  once   per  day  in  the  afternoon/evening,  and  that  driving  is  split  into  a  morning  and  afternoon   peak.  Vehicles  arriving  home  during  the  late  night  hours  are  probably  drivers  that  have   travelled  previously  during  the  day,  so  charging  has  been  assumed  to  occur  after  the   second  trip.  Vehicle  driving  patterns  have  been  based  on  weekday  statistics,  and  the   vehicles  are  assumed  to  charge  on  a  daily  basis.   In  terms  of  vehicle  penetration,  a  substation  EV  penetration  of  100%   corresponds  to  all  vehicles  being  EVs,  not  100%  of  premises  containing  an  EV.  As  there   is  an  average  of  1.7  motor  vehicles  per  household  in  Australia  [29],  a  penetration  of  59%   would  represent  an  average  of  1  vehicle  per  household.   Another  consideration  made  was  the  percentage  of  travellers  that  drive  vehicles,   as  opposed  to  using  public  transport  or  travelling  as  a  passenger.  Although  we  know  
  • 32. 21     21     that  there  is  an  average  of  1.7  vehicles  per  household,  and  that  47%  of  travellers  drive  a   vehicle  [28],  it  is  impossible  to  discern  the  percentage  of  vehicle  owners  that  drive  a   vehicle  for  the  majority  of  their  travel  during  weekdays.  This  is  because  the  number  of   travellers  includes  school  students,  for  example,  who  may  travel  as  a  passenger  or  on   public  transport,  as  well  as  those  who  own  a  vehicle  but  may  cycle  or  also  use  public   transport  to  travel  to  work.  To  further  complicate  any  assumptions  made,  there  is  no   information  relating  to  the  percentage  of  people  that  actually  travel  significant  distances   during  the  week,  including  the  considerable  proportion  of  vehicle  owners  that  fall  into   this  category  such  as  pensioners  and  those  who  work  or  care  for  children  at  home.   Therefore,  with  the  data  available,  the  most  realistic  assumptions  decided  were   that  every  vehicle  owner  travels  the  average  distance  of  20  km  return-­‐trip  on  weekdays   and  does  the  majority  of  this  travel  in  their  vehicle.  Although  analysis    may  seem  more   accurate  to  apply  a  statistical  spread  of  charger  ratings  across  each  household,  this   would  be  equivalent  to  assuming  an  average  charger  rating  for  each  household,  as  the   total  transformer  loading  would  be  the  same.  A  statistical  variation  in  charger  ratings   would  provide  a  more  accurate  model  of  voltage  regulation,  however  the  limited   number  of  premises  in  the  DIgSILENT  models  prevents  any  statistical  analysis  from   yielding  meaningful  results.  Therefore,  Eqn.  (3.3)  has  been  used  to  determine  the   charging  power  per  premise.     P =  Charger  Rating  (kW)  ∗  (Penetration/100%)  ∗  1.7  vehicles  per  premise   (3.3)     The  assumptions  made  in  this  analysis  present  an  ambiguity  issue  with  the  number  of   drivers  arriving  home  during  the  middle  of  the  day,  and  those  that  may  travel  after   arriving  home  from  work.  The  actual  number  of  drivers,  however,  is  impossible  to   predict  without  conducting  a  large-­‐scale  survey  focusing  on  the  actual  arrival  times  and   driving  patterns  of  vehicle  drivers,  therefore  the  assumptions  made  can  be  considered   as  accurate  as  possible.       3.2.5 Load  Scaling   The  load  profiles  of  premises  supplied  by  the  Woodlands  Drive  substation  represent  the   energy  use  of  premises  in  a  sample  area  of  Glenmore  Park.  These  profiles  provide  an   accurate  load  shape,  however  their  combined  substation  profile  may  not  match  the   magnitude  of  those  substations  located  in  areas  of  lower  or  higher  socio-­‐economic  
  • 33. 22     22     status,  such  as  a  wealthier  area  which  is  more  likely  to  contain  a  greater  number  of  air   conditioners  and  pool  pumps,  for  example.  To  account  for  the  diversity  between  areas   within  suburbs,  it  is  important  that  the  Woodlands  Drive  substation  load  profile  can  be   scaled  before  EV  loading  is  added,  however  non-­‐linear  line  losses  must  also  be   accounted,  therefore  this  scaling  is  not  a  straight  forward  calculation.   As  base  loading  power  increases  linearly,  represented  by  ∆ 𝑃!!"#  in  per  unit,  line  losses   increase  by  the  square  of  this  rate,  or  (∆𝑃!"#$)! .  Therefore,  if  Woodlands  Drive   substation  is  80%  loaded  under  maximum  load,  this  load  profile  cannot  be  scaled  to   represent  substation  that  is  90%  loaded,  for  example,  without  first  separating  the   combined  house  power  and  the  line  losses.   This  would  require  a  scaling  model  in  the  following  form:     𝑃!" = 𝑃! 𝑥 + 𝑃! 𝑥!   (3.4)     Where   𝑃!"  is  the  new  total  power  drawn  by  the  transformer  after  scaling,   𝑃!is  the   combined  house  power  before  scaling,   𝑃!  is  the  line  losses  before  scaling.  For  example,  if   the  total  transformer  loading  was  required  to  be  increased  from  110  kW,  where   𝑃!  =   100  kW  and   𝑃!=  10  kW,  to  240  kW,  the  combined  house  power  would  only  have  to  be   increased  by  a  factor  of  x  =  2,  to  produce  a  transformer  power  increase  of     !"# !!"  =  2.18  pu.   This  formula,  however,  does  not  take  into  account  the  line-­‐loss  increase  as  a   result  of  the  voltage  drop  that  occurs  when  constant  power  loads  are  scaled.  That  is,   when  the  power  consumption  of  a  feeder  with  constant  power  loading  increases,  so  too   does  the  voltage  drop  along  the  feeder,  causing  the  line  current,  and  hence  line  losses,  to   rise  further.  This  is  a  cyclical  response  that  converges  rapidly  due  to  the  large  difference   between  the  percentage  change  in  voltage  and  the  initial  load  power  change,  therefore   any  further  voltage  correction  can  be  considered  negligible.   Figure  3.5  shows  the  voltage  profile  of  a  typical  feeder  with  6  premises  per  phase   per  feeder  in  the  upper  graph,  approximately  the  same  as  the  Woodlands  Drive  feeders,   and  the  profile  of  the  last  4  premises  on  a  feeder  in  the  lower  graph,  with  the  voltage   levels  scaled  for  an  easier  interpretation  of  the  voltage  drop  in  each  feeder  section.    
  • 34. 23     23       Figure  3.6:  Feeder  voltage  profile,  moving  from  last  premise  to  transformer  from  right  to  left   This  feeder  shows,  when  moving  towards  the  transformer  from  right  to  left,  the  voltage   drop  increases  according  to  the  series   𝑉!"(1,  3,  6,  10)  etc.,  where   𝑉!"    is  the  voltage  drop   along  the  last  section  of  feeder,  between  the  second  last  and  last  premises.  This  series   can  be  represented  by  Eq.  (3.5).     𝑖 ! !!! = 𝑛(𝑛 + 1) 2   (3.5)     Voltage  drop  in  the  last  section  of  feeder  can  be  found  using  Eq.  (3.6).     𝑉!" = 2𝑉!" 𝑛(𝑛 + 1)   (3.6)     Where   𝑉!"  is  the  voltage  drop  along  the  entire  feeder.   For  the  feeder  in  the  top  subplot  with  6  premises,  the  total  voltage  drop  is  equal  to   !(!!!) ! 𝑉!" = 21 0.004 = 0.084  pu,  which  is  reflected  on  this  plot.  This  modelling   assumes  that  each  load  draws  the  same  current.   When  a  load  is  increased  by  a  factor   𝑥,  the  line  current  supplying  a  constant   power  load  increases  by  the  same  factor.  As  voltage  drop  is  proportional  to  current,  the   voltage  drop  also  increases  by  this  factor.  When  considering  the  last  section  of  feeder,   𝑉!"#$  is  equal  to   𝑉!"# − 𝑉!"#,  where   𝑉!"#  is  the  voltage  at  the  second  last  premise  and  
  • 35. 24     24     𝑉!"#  is  the  voltage  at  the  last  premise.   𝑉!"#$  can  be  represented  as  a  percentage  by  Eq.   (3.7).     𝑥𝑉!"#$ − 𝑉!"#$ 𝑉!"# = 𝑉!"#$(𝑥 − 1) 𝑉!"#   (3.7)     As  the  load  current  increase  is  directly  proportional  to  the  voltage  drop,  the  line  current   is  increased  by  the  factor  that  is  Eq.  (3.8).     ∆𝐼!! = 1 + 𝑉!"#$(𝑥 − 1) 𝑉!"#   (3.8)     We  now  know  the  factor  that  can  be  squared  to  scale  the  power  losses  in  the  last  section   of  feeder.  As  power  losses  increase  with  the  square  of  the  line  current,  the   𝑛!  series  can   be  used  to  represent  the  increase  in  power  moving  from  the  last  section  of  feeder  to  the   first,  i.e.   𝑃! = 𝑃!(1 + 2 + 4 + 9 + 16 + 25)  for  a  feeder  with  6  premises  per  phase.  The   sum  of  the     𝑛!  series  is  given  by  Eq.  (3.9):     𝑖! ! !!! = 𝑛(𝑛 + 1)(2𝑛 + 1) 6   (3.9)     Therefore,  if  the  total  line  losses  of  a  feeder  are  known,  the  line  losses  can  be  given  by   dividing  the  total  power  by  the  sum  of  the   𝑛!  series,  equal  to  91  for  6  premises.  Once  we   know  the  losses  and  voltage  drop  in  the  last  section  of  feeder,  and  the  scaling  factor   𝑥  by   which  the  house  loads  increase  by,  the  increase  in  line  losses  due  to  the  load  scaling  and   increased  voltage  drop  can  be  determined.  The  power  increase  ∆ 𝑃!!in  the  last  section  of   feeder  therefore  becomes  ∆ 𝑃!!∆𝐼!! ! ,  where  ∆ 𝐼!! !  is  the  increase  in  current  due  to  the   voltage  drop.  The  total  increase  in  power  due  to  voltage  drop  is  shown  in  Eq.  (3.10).   ∆𝑃! = ∆𝑃!!∆𝐼!! ! + 4∆𝑃!!2∆𝐼!! ! 9∆𝑃!!3∆𝐼!! ! + 16∆𝑃!!4∆𝐼!! ! + 25∆𝑃!!5∆𝐼!! ! + 36∆𝑃!!6∆𝐼!! !                    = ∆𝑃!!∆𝐼!! ! (1 + 8 + 27 + 64 + 125 + 216)       (3.10)      
  • 36. 25     25       This  series  represents  the  sum  of  cubes,  which  can  be  expressed  in  the  following  general   equation  form  of  Eq.  (3.11).     𝑖! ! !!! = 𝑛! (𝑛 + 1)! 4   (3.11)     Combining  Eqs.  (3.8),  (3.9)  and  (3.11),  a  general  solution  of  Eq.  (3.12)  can  be  formed  for   line  losses.     𝑃!.!"# = 6𝑃! 𝑥! 𝑛 𝑛 + 1 2𝑛 + 1 ∙ 1 + 𝑉!" 𝑥 − 1 𝑉!"# ! ∙ 𝑛! 𝑛 + 1 ! 4                            = 6𝑃! 𝑥! 𝑛 𝑛 + 1 2𝑛 + 1 ∙ 1 + 2𝑉!" 𝑥 − 1 𝑛(𝑛 + 1)𝑉!"# ! ∙ 𝑛! 𝑛 + 1 ! 4               (3.12)     Replacing  the     𝑃! 𝑥!  term  in  Eq.  (3.4)  with  Eq.  (3.12),  the  complete  transformer  power   formula  Eq.  (3.13)  is  formed.   𝑆!" = 𝑃! 𝑥 + 6𝑃! 𝑥! 𝑛 𝑛 + 1 2𝑛 + 1 ∙ 1 + 2𝑉!" 𝑥 − 1 𝑛(𝑛 + 1)𝑉!"# ! ∙ 𝑛! 𝑛 + 1 ! 4 + 𝑗𝑄! 𝑥       (3.13)       Equation  (3.13)  allows  the  apparent  transformer  power  to  be  determined  when   constant  power  house  loads  are  scaled  by  a  value  𝑥,  taking  into  account  the  non-­‐linear   nature  of  line  losses  caused  by  load  scaling  and  the  subsequent  voltage  drop.  Equation   (3.13)  assumes  that  all  houses  are  loaded  equally,  and  reactive  power  is  constant.  In   reality,  reactive  power  will  increase  slightly  in  response  to  voltage  drop,  depending  on   the  characteristics  of  the  load.  This  equation,  however,  represents  a  relatively  accurate   model  and  provides  an  insight  into  the  complexity  of  load  behaviour,  and  hence  line   losses,  in  response  to  a  change  in  load  magnitude.   Due  to  the  complexity  of  this  4th  degree  polynomial,  solving  for   𝑥  is  difficult,   therefore  loads  will  be  scaled  through  trial  and  error  for  scenarios  where  the   transformer  is  at  a  higher  capacity  than  the  Woodlands  Drive  substation.    
  • 37. 26     26     3.3 Simulation     3.3.1 Graphical  User  Interface     A  graphical  user  interface  is  a  productive  tool  that  allows  the  simple  selection  and   presentation  of  modelling  variables  and  results  in  a  single  window.  An  interface  such  as   a  GUI  is  especially  useful  for  simulating  the  effects  of  EV  charging  on  a  distribution   network,  as  each  EV  load  can  be  automatically  modified  for  all  combinations  of  charging   penetration,  charging  coordination,  charger  rating  and  network  type.  Also,  results  of   interest  can  be  displayed  both  graphically  and  numerically  for  ease  of  analysis,  such  as   lowest  bus  voltages  and  maximum  transformer  loading.   MATLAB  GUIDE,  an  open  GUI  layout  editor,  was  chosen  to  create  an  interface   with  the  DIgSILENT  models  for  this  thesis.  This  GUI,  shown  in  Fig.  3.7,  was  created  to   run  multiple  EV  loading  scenarios  without  the  need  to  manually  change  the  load  profiles   of  each  load.  This  tool  was  used  to  determine  the  loading  results  of  this  thesis,  and  will   be  of  use  to  network  planners  at  Endeavour  Energy  for  forecasting  future  load  demand.         Figure  3.7:  MATLAB  GUI   MATLAB  GUIDE  contains  a  number  of  interactive  controls  such  as  buttons  and  sliders   that  can  be  used  to  create  custom  GUIs.  Each  control  element  has  an  associated  callback   function  that  is  called  when  a  user  presses  or  changes  an  element.  These  callback  
  • 38. 27     27     functions  are  contained  in  an  associated  MATLAB  file,  used  as  the  master  script  for   running  simulations.     3.3.2 GUI  Structure     In  order  to  run  a  simulation  using  the  GUI,  the  network  variables  must  first  be  selected.   Table  3.2  shows  the  possible  variable  options  that  can  be  selected.   Feeder  Type   Overhead     Underground   Charging  Coordination   Uncoordinated     Staggered  10  pm  start   Electric  Vehicle  Penetration   0  to  100  %   Average  Charger  Rating   2  to  10  kW   Table  3.2:  Variable  Options  Structure   Each  time  a  variable  is  selected,  the  callback  of  each  pop-­‐up  menu  or  slider  is  called,  and   the  variable  name  is  stored.  A  flowchart  visualising  the  function  calls  within  the  GUI   script  is  shown  in  Fig.  3.8.     Figure  3.8:  Flowchart  displaying  the  interaction  of  programs  required  for  GUI  simulations  
  • 39. 28     28     The  breakdown  of  this  flowchart  is  as  follows,  beginning  with  the  execution  of  the  GUI   process  after  pressing  ‘Run  Simulation’.   1. The  ‘Run  Simulation’  button’s  callback  function  determines  the  combination  of   variables  selected  using  a  series  of  IF  statements.   2. The  penetration  and  charger  rating  information  is  used  to  call  a  function  that   writes  the  load  profile  to  CSVs  associated  with  the  EV  loads  in  the  LV  DIgSILENT   models.  For  example,  an  average  charger  rating  of  4  kW  at  50%  penetration  will   result  in  an  EV  load  of  2  kW  being  applied  to  each  EV  load  using  the  ‘csvwrite’   function.  The  time  of  this  charging  is  dependent  on  the  charging  coordination   strategy  (explained  in  Section  3.2.4).   3. A  MATLAB  function  changes  the  working  directory  and  calls  a  batch  file.     4. The  called  batch  file  is  used  to  open  a  connection  to  DIgSILENT  Engine,  the   engine  running  behind  DIgSILENT  PowerFactory,  and  call  a  windows  command   file  containing  the  instructions  required  to  open  and  simulate  a  DIgSILENT   model.  When  a  network  is  selected,  a  batch  file  is  called  containing  the  line   ‘digrcom  -­‐d  -­‐p  ncacn_ip_tcp  -­‐n  127.0.0.1  -­‐e  2001  -­‐f="command.cmd"  ‘.  The   windows  command  file  ‘command.cmd’  contains  the  following  lines  that  are   interpreted  by  DIgSILENT  Engine:                    ac  UserProject                  cd  UserProjectLibraryScripts                  Script  Name   This  activates  the  LV  overhead  project  and  calls  the  DPL  script,  which  performs  a   time-­‐step  simulation  and  saves  the  resulting  network  power  and  voltages  to   result  objects.   5. The  DPL  script  then  executes  result  objects  which  write  the  resulting  network   power  and  voltages  to  text  files.   6. A  MATLAB  function  calls  a  windows  command  file  to  convert  the  result  text  files   to  CSVs.  This  is  required  so  that  MATLAB  can  import  the  results  into  arrays.   7. The  command  runs  using  the  command  ‘rename’  to  change  the  result  text  files  to   CSV  files.       8. As  the  data  is  imported  into  a  single  column,  the  text-­‐to-­‐columns  function  must   be  run  to  separate  the  data  into  individual  columns.  Also,  cells  containing  text   and  zeros  must  be  removed  so  that  the  data  can  be  imported  and  analysed.  This   must  be  done  using  an  Excel  VBA  macro,  however  the  macro  had  to  be  saved  in  
  • 40. 29     29     an  excel  add-­‐in  so  that  it  can  be  used  for  multiple  CSV  workbooks.    A  VBScript  file   is  therefore  used  to  open  the  Excel  process,  open  the  voltage  results,  power   results  and  add-­‐in  files,  and  call  the  macro  so  that  the  data  can  be  read  by   MATLAB.  This  VBScript  file  is  called  by  a  windows  command,  which  also  deletes   the  previous  CSV  files  each  time  a  simulation  is  run.   9.  A  MATLAB  function  imports  the  voltage  data  as  a  single  array,  and  determines   the  minimum  bus  voltage.  The  function  then  imports  the  first  two  columns  of  the   power  data,  corresponding  to  the  real  and  reactive  transformer  power  data.  The   maximum  apparent  transformer  power  is  then  determined  from  the  real  and   reactive  power  arrays.  The  row  location  of  the  minimum  voltage  and  maximum   power  values  are  also  saved  and  inputted  into  a  function  that  converts  this  value   into  a  time  string.  The  transformer  load  profile  is  then  written  to  a  single  CSV  file   associated  with  each  load  in  the  11  kV  model.  The  process  is  then  repeated  to   extract  the  11  kV  voltage  and  transformer  data.   10. The  minimum  voltage,  maximum  power  and  the  time  these  values  occur  are  then   displayed  on  a  graph  on  the  GUI,  with  the  400  V  transformer  data  displayed  by   default.  The  user  can  then  change  between  transformer  and  voltage  results,  and   400  V  and  11  kV  results.     The  GUI  script  is  therefore  used  to  call  MATLAB  functions  and  external  batch,   command  and  VBScript  files  to  control  MATLAB,  DIgSILENT  and  Excel.  This  allows   simulation  scenarios  to  be  generated  and  presented  within  a  single  GUI  window.   3.4 Scenarios   3.4.1  Uncoordinated  Charging     To  determine  the  impacts  of  EV  charging  on  low  voltage  networks,  analysis  was   conducted  on  the  Woodlands  Drive  substation,  as  well  as  theoretically  loaded   substations  that  may  represent  other  areas.  The  Woodlands  Drive  substation  base  load   was  scaled  through  trial  and  error,  considering  the  analysis  conducted  in  Section  4.1.2,   so  that  the  maximum  loading  without  added  EV  loads  corresponded  with  80,  85,  90  and   95%  of  the  substations  500  kVA  capacity.      
  • 41. 30     30     The  following  variables  were  considered  in  this  analysis:   1.  Network  type     2.  Average  charger  rating     3.  Charging  time   4.  Phase  balance   Both  underground  and  overhead  LV  networks  were  modelled  to  determine  the   difference  in  EV  loading  caused  by  differences  in  line  impedances.  Average  charger   ratings  of  4,  7  and  10  kW  were  considered,  with  4  kW  the  most  probable  average   charger  rating  expected  in  the  coming  years.  Charging  times  were  found  to  be  19,  27  and   47  minutes  for  4,  7  and  10  kW  chargers  respectively,  using  statistics  from  Section  3.2.4,   modelled  as  half  hour  and  one  hour  charging  periods  due  to  a  half  hour  time-­‐step   resolution  limit.  Scenarios  were  also  simulated  where  this  charging  time  may  be   doubled,  useful  for  representing  areas  where  the  average  driving  distance  may  be   greater  than  the  NSW  average.  Phase  balance  was  the  last  variable  considered,   considering  the  expected  outcome  of  an  unbalanced  network  using  the  provided  smart   metering  data,  and  the  unlikely  scenario  where  a  feeder  may  be  close  to  perfectly   balanced.     All  combinations  of  these  variables  were  modelled  using  MATLAB  functions   through  DIgSILENT  engine.  This  analysis  ensured  that  all  possible  scenarios  were   covered  so  that  results  were  as  accurate  as  possible.   3.4.2 Coordinated  Charging     To  simulate  coordinated  charging,  a  simple  off-­‐peak  staggered  charging  method  was   chosen.  EVs  were  first  split  into  three  groups,  assigned  evenly  across  three  phases  to   minimise  voltage  drop,  and  assigned  a  starting  time  of  10  pm,  11  pm  or  12  am.    The   maximum  EV  penetration  was  determined  for  the  same  variables  considered  in  the   uncoordinated  charging  analysis,  except  for  voltage  unbalance  as  this  was  not  an  issue   during  the  late-­‐night  hours  for  the  smart-­‐metered  houses.     Next,  the  number  of  start  times  was  doubled  to  six,  so  that  EVs  were  assigned  a   start  time  ranging  from  10  pm  to  3  am.  This  allowed  the  EV  loading  at  each  hour  to  be   halved,  while  allowing  a  suitable  amount  of  time  for  drivers  who  may  require  their  car   early  in  the  morning.  
  • 42. 31     31     An  issue  surrounding  the  analysis  of  the  hottest  day  was  the  insignificant  off-­‐ peak  hot  water  loading  during  the  late  night  hours.  This  may  have  been  due  to  a  large   proportion  of  gas/solar  hot  water  systems  in  this  area,  a  reduction  in  showering  times   and  temperatures  due  to  the  hot  weather,  or  both.  Electric  hot  water  systems  are   typically  rated  at  3.6  kW,  drawing  power  as  a  constant  impedance  load,  therefore  the   average  power  consumption  would  be  less  than  this  with  additional  loads  reducing   voltages.  Therefore,  to  consider  an  area  with  all  premises  connected  to  electric  off-­‐peak   water  heating,  the  7  kW  EV  charging  scenario  would  provide  a  better  analysis  of  a   worst-­‐case  scenario  for  hot  water  heating  and  4kW  charging.   3.4.3 11  kV   The  analysis  of  400  V  feeders  provides  an  insight  into  the  effects  of  loading  on  LV   substations,  providing  an  idea  of  the  loading  at  the  zone  substation  level.  A  zone   substation,  however,  is  typically  designed  with  a  greater  focus  on  future  growth  in  the   number  of  loads,  compared  to  a  sole  consideration  on  the  average  power  increase  of   each  load,  as  the  area  covered  by  a  zone  substation  is  significant  and  determined  by   geographical  and  financial  considerations.   To  determine  the  loading  effects  of  EV  charging  on  Glenmore  Park  zone   substation,  the  hottest  day  of  the  2010/11  period  was  selected,  where  capacity  reached   43.5  MVA  of  the  substation’s  45  MVA  N-­‐1  capacity.  This  day’s  loading  was  significantly   greater  than  the  hottest  day  of  2011/12,  chosen  for  modelling  with  the  Woodlands  Drive   smart  metering  data,  where  the  maximum  zone  substation  loading  reached  29  MVA   (however,  this  day  was  the  highest  loaded  day  recorded  by  the  smart  meters  as  they   hadn’t  been  installed  before  the  hottest  day  of  2010/11).   As  Glenmore  Park  ZS  supplies  7596  premises  on  10  11kV  feeders,  there  can  be   assumed  an  average  of  760  premises  per  feeder.  Assuming  92  premises  are  assigned  to   each  500  kVA  of  installed  capacity,  there  would  be  an  average  of  8  transformers  per  11   kV  feeder.  Dividing  the  45  MVA  capacity  by  10  feeders  and  8  transformers,  however,   results  in  an  average  transformer  capacity  of  550  kVA.  This  can  be  explained  by  the   variations  in  socio-­‐economic  status,  gas  supply  availability  and  the  percentage  of   commercial  and  industrial  customers,  which  can  be  disregarded  for  EV  loading  based  on   assumptions  made  in  Section  3.2.4.   To  simulate  the  zone  substation  at  43.5  MVA,  or  96.67%  capacity,  the  400  V   model  was  modified  to  be  on  a  275  kVA  base  (half  of  550  kVA  due  to  node  limitations),  
  • 43. 32     32     and  the  base  house  loads  were  scaled  so  that,  once  lumped  into  the  11  kV  model  (and   doubled  to  represent  550  kVA  transformers),  they  produced  a  feeder  loading  of  4.35   MVA  in  the  11  kV  model  (representing  1/10th  of  the  total  ZS  loading).  A  load  profile  for   both  real  and  reactive  power  was  lumped  into  the  11  kV  model  to  account  for  the  change   in  power  factor  as  EVs  were  added.  The  base  loading  scaling  in  the  400  V  model  was   completed  through  trial  and  error  due  to  the  non-­‐linear  losses  in  both  the  400  V  and  11   kV  networks.     Next,  EV  penetration  was  increased  until  the  11  kV  feeder  loading  reached  4.5   MVA,  marking  100%  transformer  loading.  This  was  completed  for  4,  7  and  10  kW   chargers  for  both  the  expected  driving  distance  of  20  km  and  the  case  that  this  distance   was  doubled.                                                                      
  • 44. 33     33         4 Results     The  implementation  of  a  GUI  has  allowed  for  a  number  of  loading  scenarios  to  be   simulated  and  compared  to  accurately  determine  the  effects  of  EV  charging  on   residential  distribution  feeders.     4.1 Base  Load  Profile   The  base  load  profile,  comprised  of  actual  premise  smart  metering  data,  is  an  important   profile  as  it  forms  the  basis  of  all  EV  loading  analysis.   4.1.1 Effects  of  Temperature  on  Substation  Loading   The  loading  significance  due  to  air  conditioners  can  be  seen  in  Fig.  4.1.  Figure  4.1   compares  the  substation  profile  on  a  38.7  degree  day  compared  to  a  mild  19.9  degree   day,  both  days  being  weekdays.     Figure  4.1:  Woodlands  Drive  substation  loading  for  38.7  and  19.9  degrees  celsius  days   Figure  4.1  clearly  shows  a  substantial  difference  in  loading  between  the  hot  and  mild   day,  with  the  hottest  day  loading  162%  greater  than  the  mild  temperature  day.  This   comparison  confirms  the  need  to  consider  the  hottest  days  of  the  year  when  planning   for  worst-­‐case  loading  scenarios.   4.1.2 Load  Scaling   Another  factor  contributing  to  an  appropriate  base  load  profile  is  the  scaling  of  the   modelled  transformer  power.  As  the  modelled  premises  were  a  50%  sample  of  the  total   number  of  distribution  substation  premises,  the  load  profile  for  this  sample  would  be   desired  to  reflect  the  load  profile  of  the  entire  sample  set  to  ensure  accuracy.  Figure  4.2  
  • 45. 34     34     shows  the  modelled  curve  of  46  premises  once  the  load  profile  magnitude  had  been   doubled,  and  the  actual  substation  meter  load  profile  supplying  92  premises.       Figure  4.2:  Woodlands  Drive  substation  total  load  compared  to  scaled  sample  loads   Figure  4.2  shows  that  the  modelled  sample  (red)  closely  follows  the  actual  substation   profile,  confirming  the  sample  contains  no  significant  outliers  that  would  have  skewed   the  modelled  results  after  scaling.  Although  a  noticeable  loading  difference  exists  at   around  7:30  pm,  the  load  peaks  at  6:00  pm  are  close  to  the  same  value,  this  being  the   most  important  time  as  network  loading  is  greatest  and  during  the  expected  worst   uncoordinated  charging  period.   4.1.3 Network  Type   The  difference  in  transformer  loading  between  underground  and  overhead  low  voltage   networks  becomes  apparent  in  Fig  4.3.       Figure  4.3:  Woodlands  Drive  substation  load  for  overhead  and  underground  networks   As  all  loads  have  been  measured  as  constant  power  loads,  load  currents  must  increase   as  voltages  decrease  down  the  length  of  a  feeder.  As  overhead  lines  typically  have  an   impedance  greater  than  underground  cables,  the  voltage  drop,  and  hence  line  currents,  
  • 46. 35     35     will  be  greater  in  an  overhead  network,  resulting  in  higher  losses  which  must  be   supplied  by  the  transformer.     During  the  maximum  demand  period  the  maximum  power  drawn  is  274  kVA,   representing  56%  of  the  substations  500  kVA  total  capacity.   4.2 Uncoordinated  Charging   Simulations  were  run  for  penetrations  levels  up  to  100%  for  both  overhead  and   underground  low  voltage  networks.  For  overhead  LV  feeders,  a  tap  setting  of  -­‐4  (+10%)   was  found  to  allow  secondary  LV  substation  voltages  to  remain  closest  to  1.1  pu  for  all   EV  penetrations  during  peak  periods,  assuming  at  least  a  1  pu  primary  voltage.   Satisfactory  voltage  regulation  on  a  -­‐4  tap  setting  therefore  requires  the  11  kV  feeder  to   be  capable  of  maintaining  voltages  at  the  end  of  feeder  to  at  least  1  pu.  Hence,  it  is   important  to  determine  the  11  kV  voltage  regulation  capabilities,  to  determine  the   worst-­‐case  voltage  which  will  be  the  limiting  factor  for  LV  voltage  regulation.   4.2.1 11  kV  Voltage  Regulation     A  tap  setting  of  -­‐4  (+10%)  is  a  typical  maximum  tap  setting  for  LV  distribution   transformers.  This  tap  setting  has  been  selected  for  modelling  as  it  provides  the  highest   voltage  at  0%  EV  penetration  during  the  peak  loading  hours.  In  reality,  a  tap  setting  of  -­‐3   (+7.5%)  is  a  typical  setting  for  LV  transformers,  as  this  maintains  a  voltage  less  than  1.1   pu  for  premises  closest  to  the  transformer  during  periods  of  low  loading.     To  ensure  1.1  pu  is  set  at  the  secondary  LV  transformer  side  of  the  LV  overhead   network,  the  11  KV  feeder  voltage  must  be  capable  of  providing  at  least  1  pu  voltage   (exactly  1  pu  required  for  the  highest  tap  setting  of  -­‐4)  to  the  end  of  the  feeder  without   the  start  of  the  feeder  exceeding  the  upper  voltage  limit  or  OLTC  capability.   To  test  the  voltage  regulation  capabilities  of  a  typical  11  kV  feeder  under  worst-­‐case   conditions,  100%  EV  loading  was  applied  to  premises  on  LV  substations,  and  this   resulting  load  profile  was  then  lumped  into  LV  substations  in  the  11  kV  model.   The  11  kV  bus  voltage  was  found  to  be  to  be  1.045  pu  in  order  to  satisfy  the  1  pu  voltage   requirement  at  the  end  of  the  feeder,  confirming  11  kV  voltage  regulation  was  suitable   for  considering  EV  loading.    
  • 47. 36     36     4.2.2 400  V  Transformer  and  Feeder  Loading   When  modelling  400  V  scenarios,  the  overhead  network  LV  tap  changer  was  set  to  -­‐4   (+10%)  and  the  underground  LV  tap  changer  was  set  to  -­‐3  (+7.5%),  allowing   satisfactory  voltage  regulation  during  the  evening  hours.     4.2.2.1 Woodlands  Drive  Substation   The  impacts  of  charging  on  Woodlands  Drive  substation  was  determined  for  EV   penetrations  up  to  100%,  shown  in  Table  4.1.  The  results  in  Table  4.1  show  that  at  zero   percent  EV  penetration,  Woodlands  Drive  substation  is  significantly  under-­‐loaded  for   the  hottest  day  of  2011,  reaching  only  55%  capacity  as  it  is  located  in  an  underground   area.  At  500  kVA,  network  planners  have  allowed  for  5.5  kVA  per  customer,  which  is   typical  for  premises  in  a  medium  socio-­‐economic  area,  therefore  these  premises  must   use  less  energy  than  expected.       Overhead   Underground   Penetration   %   Loading   %   Loading   Increase   Min.  Bus   Voltage   %   Loading   %   Loading   Increase   Min.  Bus   Voltage   0%   56.36%     0.00%   1.006  pu   55.32%   0.00%   1.021  pu   5%   57.60%     2.20%   1.005  pu   56.48%   1.24%   1.020  pu   10%   58.84%     4.40%   1.005  pu   57.72%   2.55%   1.020  pu   15%   60.08%     6.60%   1.005  pu   58.92%   3.87%   1.020  pu   20%   61.36%     8.87%   1.005  pu   60.12%   5.11%   1.020  pu   25%   62.64%     11.14%   1.005  pu   61.36%   6.42%   1.019  pu   30%   63.92%     13.41%   1.005  pu   62.60%   7.74%   1.019  pu   35%   65.25%     15.76%   1.003  pu   63.84%   8.98%   1.019  pu   40%   66.56%     18.10%   0.999  pu   65.12%   10.92%   1.018  pu   45%   67.92%     20.51%   0.994  pu   66.36%   12.55%   1.016  pu   50%   69.28%     22.92%   0.989  pu   67.64%   14.96%   1.014  pu   55%   70.68%     25.41%   0.983  pu   68.92%   17.37%   1.012  pu   60%   72.04%     27.82%   0.978  pu   70.20%   19.78%   1.010  pu   65%   73.48%     30.38%   0.972  pu   71.48%   22.19%   1.009  pu   70%   74.88%     32.86%   0.966  pu   72.76%   24.60%   1.007  pu   75%   76.36%     35.48%   0.960  pu   74.08%   27.00%   1.005  pu   80%   77.80%     38.04%   0.956  pu   75.36%   29.42%   1.003  pu   85%   79.28%     40.67%   0.950  pu   76.68%   31.90%   1.001  pu   90%   80.76%     43.29%   0.943  pu   78.00%   34.31%   0.999  pu   95%   82.28%     45.99%   0.937  pu   79.32%   36.79%   0.997  pu   100%   83.80%     48.69%   0.931  pu   80.64%   39.20%   0.995  pu   Table  4.1:  Woodlands  Drive  substation  transformer  loading  and  voltage  regulation  for  varying  EV   penetrations   The  Woodlands  Drive  substation  loads  would  be  capable  of  supporting  90%  EV   penetration  in  an  overhead  network,  and  100%  in  an  underground  network  with  a  4  kW  
  • 48. 37     37     average  charger  rating.  Transformer  loading  would  be  greater  for  an  overhead  network   due  to  increased  losses  associated  with  higher  feeder  resistances  (1.49  ohms  overhead   compared  to  0.927  ohms  underground  for  the  modelled  cables).  The  impact  of   resistance  can  be  seen  at  EV  penetrations  above  90%  for  an  overhead  feeder,  where  the   minimum  bus  voltage  falls  below  the  lower  limit  of  0.94  pu.  At  this  penetration,  the   transformer  voltage  could  not  be  further  increased  to  compensate  for  the  worst-­‐bus   voltage,  therefore  marking  the  point  at  which  voltage  regulation  would  fail.     Upon  further  analysis,  poor  voltage  regulation  was  found  to  be  caused  by   significant  voltage  unbalance  between  phases,  and  loading  between  feeders.  Where   feeder  two’s  worst  bus  voltage  was  equal  to  0.931  pu  at  100%  EV  penetration,  the   lowest  bus  voltage  on  feeder  one  was  0.987,  as  this  feeder  was  both  more  balanced  and   lightly  loaded  that  feeder  two  after  random  premise  loading  allocation.  On  feeder  two,   voltages  ranged  from   𝑉!=0.931  pu,   𝑉!=1.08  pu  and   𝑉!=0.936  pu  at  the  three  premises   furthest  from  the  transformer,  compared  to   𝑉!=1.04  pu,   𝑉!=1.08  pu  and   𝑉!=1.036  pu  at   the  same  premises  with  no  EV  loading.  At  the  time  of  these  voltage  results,  6:30  pm,   three  EVs  were  charging  on  phases  a  and  c,  with  none  on  phase  b  after  random   allocation.  The  three  EV  loads  charging  on  each  phase  were  also  located  at  the  end  of  the   feeder,  further  contributing  to  the  problem,  resulting  in  an  extreme  loading  scenario   where  charging  occurred  only  on  the  heaviest  loaded  phases  at  the  end  of  the  feeder,   exceeding  voltage  limits  before  the  full  transformer  loading  capacity  was  reached  on  the   overhead  network.   4.2.2.2 Additional  Scenarios     Scenario  1:  4  kW  Charger   Table  4.2  shows  the  maximum  EV  penetration  before  transformer  capacity  is  reached,  or   voltage  regulation  has  failed,  denoted  by  a  (*).  Charging  is  divided  into  overhead  and   underground  networks,  subdivided  into  balanced  and  unbalanced  networks,  and  then   further  subdivided  into  two  charging  times,  where     𝑇!  is  the  expected  charging  time   based  on  an  average  20  km  return  trip,  equal  to  45  minutes  for  a  4  kW  charger   (modelled  as  one  hour).    2 𝑇!  is  equal  to  1.5  hours  in  this  case.  The  column  containing   data  for  an  unbalanced  network  with  charging  time   𝑇!  has  been  highlighted,  as  each   column  contains  the  most  likely  scenarios  based  on  research  in  this  thesis.    
  • 49. 38     38     Existing  Base   Transformer   Loading  on   Hot  Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Unbalanced   Balanced   Unbalanced   Balanced   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   Wood.  Dr.   90%*   80%*   90%*   79%*   100%   100%   100%   100%   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   58%*   38%*   54%*   42%*   77%   54%   76%   53%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   51*%   28%*   47%*   37%*   58%   41%   57%   40%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   39%   28%   39%   27%   38%   27%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   23%   21%   20%   14%   19%   13%   Table  4.2:  Maximum  EV  penetration  for  4  kW  LV  uncoordinated  charging   Each  row  represents  a  different  transformer  base  load  as  a  percentage  of  𝑆!"#$%,   equal  to  500  kVA.  The  Woodlands  Drive  substation  base  loading  is  equal  to  0.55 ∙ 𝑆!"#$%   for  an  underground  network  and  0.56 ∙ 𝑆!"#$%  for  overhead,  as  determined  in  Section   4.2.2.1.  The  remaining  rows  represent  theoretically  loaded  transformers  at  a  worst-­‐case   base  loading  of    0.80 ∙ 𝑆!"#$%  and  0.95 ∙ 𝑆!"#$%  during  the  evening  hours,  determined  by   scaling  the  Woodlands  Drive  smart  meter  loads.  The  degree  of  load  unbalance  is   therefore  the  same  for  all  unbalanced  loading  scenarios,  where  load  unbalance  refers  to   the  difference  between  load  magnitudes  across  phases  at  the  peak  loading  time  of  6  pm,   not  the  loading  unbalance  across  time.   Analysis  of  results  shows  a  stark  contrast  between  overhead  and  underground   networks.  The  higher  overhead  line  impedance  is  shown  to  significantly  reduce  the   maximum  EV  penetration  for  overhead  networks,  limiting  maximum  penetration  with   unbalanced  loads  on  an  80%  loaded  transformer  to  58%  overhead  compared  to  100%   for  an  underground  network,  for  example.  For  an  overhead  unbalanced  network,  scaling   the  base  load  above  85%  was  not  possible  without  exceeding  the  lower  voltage  limit  of   0.94  at  the  worst  bus,  due  to  load  unbalance  as  a  result  of  the  randomly  assigned  smart   meter  load  profiles.     A  noticeable  and  unexpected  relationship  occurred  between  balanced  and   unbalanced  networks,  as  balanced  networks  were  able  to  handle  less  added  EV  loads.   Upon  further  analysis,  the  cause  of  this  relationship  was  found  to  be  the  difference  in   losses  –  a  substation  loaded  at  80%  capacity  with  unbalanced  loads  during  the  evening   peak  would  have  a  greater  proportion  of  losses  than  a  balanced  substation  also  at  80%   capacity.  Hence  the  balanced  substation  could  be  regarded  as  more  efficiently  loaded,  as   the  actual  combined  house  loading  (both  real  and  reactive  power),  disregarding  line  
  • 50. 39     39     losses,  would  be  greater.  A  more  efficient  base  loading,  however,  leaves  less  room  for  EV   loads.     The  difference  in  maximum  EV  penetration  between  balanced  and  unbalanced   networks  is  considerably  less  in  an  underground  network,  with  differences  varying  by   1%  typically,  due  to  the  lower  losses.  The  same  relationship  was  found  between   underground  and  overhead  networks  when  both  were  balanced,  as  the  capacity  of  the   underground  network  dropped  to  less  than  that  of  the  overhead  network  due  to  the   lower  percentage  of  losses.  Lower  EV  penetration  for  balanced  underground  networks   only  occurred  for  the  90  and  95%  loaded  substations  where  maximum  EV  penetration   was  not  limited  by  voltage  regulation  in  the  overhead  network.   Table  4.2  also  clearly  shows  the  significant  reduction  in  possible  EV  loading  in  areas   where  travel  distance,  hence  charging  time,  would  be  twice  the  average,  for  example   100%  EV  penetration  compared  to  73%  for  an  80%  loaded  transformer  supplying  an   underground  network.     Scenario  2:  7kW     Existing  Base   Transformer   Loading  on   Hot  Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Unbalanced   Balanced   Unbalanced   Balanced   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   Wood.  Dr.   53%*   52%*   57%*   52%*   100%   93%*   100%   83%   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   33%*   33%*   38%*   31%*   77%   44%   75%   43%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   29%*   29%*   33%*   27%*   59%   33%   57%   32%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   26%*   22%   39%   22%   38%   21%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   20%*   13%   20%   11%   19%   10%   Table  4.3:  Maximum  EV  penetration  for  7  kW  LV  uncoordinated  charging   At  7  kW,  the  charging  time   𝑇!  reduces  to  half  an  hour.  Maximum  penetration  at   Woodlands  Drive  substation  was  found  to  reduce  from  90%  to  53%  compared  to  a  4  kW   charger,  while  a  penetration  of  100%  was  still  possible  in  an  unbalanced  underground   network.  This  result  draws  further  attention  to  the  unbalance  across  phases  caused  by   the  EVs,  as  there  is  a  greater  difference  between  the  overhead  and  underground   networks  where  overhead  penetration  is  limited  by  voltage  regulation.     The  change  in   𝑇!from  one  hour  for  4  kW  chargers  to  half  an  hour  resulted  in   different  combination  of  vehicles  charging  at  the  same  time,  where  the  same  number  of   vehicles  were  charging  for  both  charging  times.  This  resulted  in  voltage  regulation  
  • 51. 40     40     failing  at  a  penetration  lower  than  that  of  the  balanced  network,  which  contrasts  the   results  of  the  4  kW  scenario.       Scenario  3:  10  kW     Existing  Base   Transformer   Loading  on  Hot   Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Unbalanced   Balanced   Unbalanced   Balanced   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   𝑻 𝒄   𝟐𝑻 𝒄   Wood.  Dr.   37%*   37%*   39%*   36%*   100%   65%*   85%*   58%*   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   23%*   23%*   26%*   22%*   54%   31%   52%*   30%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   20%*   20%*   23%*   19%*   41%   23%   40%   22%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   18%*   16%   27%   15%   26%   15%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   0%*   0%*   14%*   9%   14%   8%   13%   7%   Table  4.4:  Maximum  EV  penetration  for  10  kW  LV  uncoordinated  charging   At  10  kW,   𝑇!  was  equal  to  20  minutes,  however   𝑇!  was  rounded  to  half  an  hour  as  half  an   hour  was  the  smallest  time-­‐step  resolution  possible.    Woodlands  Drive  substation   loading  was  found  to  be  capable  of  supplying  100%  EV  penetration  for  underground   networks,  while  only  37%  could  be  charged  for  overhead  networks  assuming  a  half  hour   charging  time  and  unbalanced  network.     The  impacts  of  increasing  charging  rating  are  shown  in  Fig.  4.4  for  an  unbalanced   underground  network  at  100%  EV  penetration,  showing  that  complete  EV  penetration  is   possible  up  to  and  including  7  kW  average  charger  ratings.     Figure  4.4:  Impact  of  increasing  charger  rating  on  underground  network  at  100%  EV  penetration        
  • 52. 41     41     Discussion  of  Results   The  analysis  of  400  V  coordinated  charging  results  has  brought  a  number  of  issues  to   light,  some  of  which  were  not  entirely  obvious  without  further  investigation.  These   issues  included:   -­‐  The  difference  in  loading  capabilities  between  underground  and  overhead  networks   was  found  to  be  significant  due  to  the  substantial  difference  in  line  impedances.  The   underground  network  was  able  to  avoid  voltage  regulation  issues  occurring  before  the   transformer  capacity  was  reached,  while  the  overhead  network  could  not,  even  for  the   balanced  network  scenario.   -­‐  Loading  unbalance  during  the  evening  hours  was  the  limiting  factor  in  maximum  EV   penetration  for  overhead  networks.  The  combination  of  unbalanced  base  loads  and  EV   loads  after  random  distribution  caused  a  worst-­‐case  unbalance  scenario  which  is   possible  in  reality.   -­‐  The  location  of  loads  was  also  a  factor  that  contributed  to  poor  voltage  regulation,  as   those  closest  to  the  end  of  the  transformer  forced  additional  current  to  flow  the  entire   length  of  the  feeder,  causing  a  voltage  drop  worse  than  if  they  were  situated  closer  to  the   transformer.   -­‐  A  balanced  overhead  network  is  capable  of  supplying  a  lower  number  of  EVs  than  a   balanced  underground  network  at  the  same  base  loading,  as  the  losses  in  the   underground  substation  loads  make  up  a  smaller  percentage,  hence  supporting  a  larger   combined  house  load,  resulting  in  a  greater  reactive  power  supply.     Of  these  issues,  loading  unbalance  between  houses  and/or  EVs  was  found  to  be   the  most  significant  issue  in  overhead  networks,  causing  poor  voltage  regulation  that   limited  the  penetration  of  EV  charging  in  overhead  networks  to  loading  of  less  than  the   transformers  rating.  Voltage  unbalance  must  be  addressed  by  considering  the  evening   hours,  especially  in  areas  of  higher  economic  status  where  higher  rated  chargers  and   penetration  levels  are  more  probable.  Voltage  unbalance  is  not  expected  to  have  a   critical  effect  on  underground  networks,  as  lower  line  resistances  prevent  voltage  drop   from  being  as  significant.   The  most  important  information  that  can  be  interpreted  from  these  results  is  the   high  percentage  of  possible  EVs  that  may  charge  in  an  uncoordinated  manner  without   overloading  network  equipment.  The  high  percentage  of  EV  penetrations  were  due  to   the  large  variation  in  start  times,  which  resulted  in  only  9  of  the  46  vehicles  charging   during  the  6:00  to  6:30  pm  period  that  caused  peak  loading.  Keeping  in  mind  that  this  
  • 53. 42     42     analysis  disregarded  charging  at  the  workplace,  actual  allowable  EV  penetration  levels   may  be  significantly  higher  than  those  found,  depending  on  the  extent  to  which   workplace  charging  is  integrated.     Based  on  analysis  conducted  by  Deutsche  Bank  [19]  in  Chapter  1,  around  35%  of   vehicles  are  expected  to  be  EVs  by  2030.    Therefore,  transformers  with  a  base  loading  of   85%  in  overhead  networks  and  90%  in  underground  networks  are  expected  to  handle   EVs  until  at  least  2030,  assuming  a  4  kW  average  charger  rating  and  unbalanced  feeders.   However,  if  EV  penetration  increases  beyond  these  projections,  and  workplace  charging   does  not  become  significant,  overhead  networks  may  need  upgrading  as  early  as  this   decade.     4.2.3 11  kV  Transformer  Loading     By  scaling  the  base  house  load  profiles  of  the  550  kVA  overhead  LV  transformer,  a   loading  of  85.8%  was  required  to  cause  a  loading  of  96.67%  in  the  11  kV  model,  due  to   the  significant  proportion  of  losses.  EV  loads  were  then  added  until  the  11  kV  source   power  reached  4.5  MVA,  or  100%.  Modelling  the  LV  network  as  balanced  had  a   significant  impact  on  the  voltage  regulation,  allowing  the  voltage  to  stay  within  its  limits.     Although  this  would  not  be  the  case  in  reality,  this  assumption  had  to  be  made  to   avoid  the  voltage  collapse  that  would  result  when  using  the  smart  meter  load  profiles.   Where  voltage  regulation  was  too  poor  at  the  LV  level,  LV  substations  would  become  the   limiting  factor  for  zone  substation  loading.  The  overhead  network  was  modelled  on  a   tap  setting  of  -­‐4  while  the  underground  network  required  a  tap  setting  of  -­‐3.  The   maximum  penetrations  for  uncoordinated  charging  are  shown  in  Table  4.5.   Charger   Rating   Maximum  EV  Penetration   𝑻 𝒄=  0.5  Hours   𝑻 𝒄=  1  Hour   𝑻 𝒄=  1.5  Hours   𝑅   0.75𝑅   0.5𝑅   𝑅   0.75𝑅   0.5𝑅   𝑅   0.75𝑅   0.5𝑅   4  kW   -­‐   -­‐   -­‐   8%   10%   16%   4%   5%   8%   7  kW   8%   10%   16%   4%   5%   8%   -­‐   -­‐   -­‐   10  kW   5%   8%   10%   3%   4%   6%   -­‐   -­‐   -­‐   Table  4.5:  Maximum  EV  penetration  at  zone  substation  assuming  worst  loading  day  in  2010/11    Table  4.5  is  divided  into  charge  times,  and  sub  divided  into  possible  zone  substation   load  ratios,  where  R  represents  100%  residential  customers,  while  0.5R  may  represent   the  reality  that  50%  of  the  zone  substations  load  is  residential,  while  the  other  50%  may   be  commercial  and/or  industrial.  The  EV  penetrations  were  found  by  simply  multiplying   the  modelled  results,  R,  with  the  reciprocal  of  the  ratio  of  residential  to  non-­‐residential  
  • 54. 43     43     loads.  This  was  possible  because  non-­‐residential  loads  are  not  expected  to  charge  EVs   during  the  evening  peak.   The  maximum  EV  penetration  level  was  constant  for  overhead  and  underground   networks,  showing  that  the  increase  to  losses  by  adding  EVs  was  negligible  when  the   underground  and  overhead  networks  were  loaded  to  the  same  high  capacity,  inclusive   of  the  base  house  losses.       The  penetration  values  displayed  in  Table  4.5  are  significantly  lower  than  the   results  obtained  for  LV  uncoordinated  charging  scenarios.  This  is  partly  due  to  the   maximum  base  substation  loading  simulated  was  95%,  while  this  zone  substation   loading  is  based  on  a  96.67%  loaded  day.  Additional  line  losses  in  the  11  kV  lines  are   also  a  contributing  factor.         4.3 Coordinated  Charging   4.3.1 3-­‐Group  Charging     4  kW  Chargers     With  an  average  charger  rating  of  4  kW,  as  is  expected  in  the  coming  years,  coordinated   charging  was  found  to  keep  transformer  loading  at  a  suitable  level  for  100%   penetration,  for  all  base  loading  levels.  The  base  loading  level  was  found  to  have  a  minor   impact  on  coordinated  charging,  as  this  charging  occurs  after  the  evening  peak.       Figure  4.5:  4  kW  three-­‐group  coordinated  charging  for  different  transformer  base  levels   Figure  4.5  shows  coordinated  charging  with  100%  EV  loading  on  an  overhead  network   for  the  Woodlands  Drive  base  loading,  and  80-­‐95%  loaded  substations.  The  maximum   load  is  seen  to  remain  as  the  evening  peak,  which  does  not  exceed  500  kVA.    
  • 55. 44     44       The  balanced  overhead  network  was  chosen  for  coordinated  modelling  because  this   network  allowed  90  and  95%  base  loads  to  draw  power  during  the  evening  peak   without  voltage  regulation  issues.  The  coordinated  charging  load,  however,  was  the   same  magnitude  for  the  unbalanced  overhead  network  as  no  significant  loading   differences  were  present  in  the  late  night  hours  in  the  unbalanced  model.     7  kW  Chargers     Existing  Base   Transformer   Loading  on  Hot  Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Woodlands  Dr.   96%   100%   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   87%   94%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   86%   92%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   84%   91%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙 82%   89%   Table  4.6:  Maximum  EV  penetration  for  7kW  LV  coordinated  charging   As  EV  charger  ratings  were  increased  to  7  kW  on  average,  Table  4.6  shows  three-­‐group   coordinated  charging  was  able  to  increase  the  maximum  penetration  significantly  over   uncoordinated  charging.  At  85%  base  loading,  for  example,  uncoordinated  charging  is   limited  to  33%  for  a  balanced  overhead  network,  but  can  be  increased  considerably  to   86%  with  three  off-­‐peak  charging  groups.     10  kW  Chargers   The  same  improvement  was  seen  with  10  kW  chargers  in  Table  4.7,  as  a  penetration   maximum  of  13%  can  be  increased  to  57%  for  an  overhead  balanced  network  on  a  95%   loaded  transformer.     Table  4.7:  Maximum  EV  penetration  for  10  kW  LV  coordinated  charging   Existing  Base   Transformer   Loading  on  Hot  Day   Maximum  EV  Penetration  Before  Transformer/Feeder  Overload   Overhead   Underground   Woodlands  Dr.   67%   72%   𝟎. 𝟖𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   60%   66%   𝟎. 𝟖𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   59%   64%   𝟎. 𝟗𝟎 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙   58%   63%   𝟎. 𝟗𝟓 ∙ 𝑺 𝑻𝑿𝒎𝒂𝒙 57%   62%  
  • 56. 45     45     4.3.2 Six-­‐Group  Charging   The  change  to  six  groups  of  charging  start  times  saw  a  significant  increase  in  the   maximum  number  of  EVs  able  to  charge  on  a  substation  –  so  much  so  that  100%   penetration  was  possible  for  an  overhead  network  loaded  at  95%  during  the  evening   peak,  with  an  average  charger  rating  of  10  kW.  Figure  4.6  shows  six-­‐group  off-­‐peak   charging  at  100%  EV  penetration  for  4,  7  and  10  kW  charger  ratings.       Figure  4.6:  Six-­‐group  coordinated  charging  for  a  95%  loaded  transformer   Therefore,  the  vast  majority  of  transformers  are  expected  to  be  capable  of  supporting   100%  penetration  of  10  kW  charging.  The  ratio  of  vehicles  could  be  further  tweaked  to   prevent  the  evening  peak  being  exceeded  for  the  same  10  pm  to  4  am  charging  period.   4.3.3 11  kV     A  24  hour  load  profile  was  not  available  for  the  hottest  day  of  2010/11,  when  the   maximum  zone  substation  loading  of  96.67%  occurred.  Based  on  the  results  for  LV   coordinated  charging,  however,  there  is  not  expected  to  be  any  issues  in  the  LV   networks  unless  charging  cannot  be  coordinated  into  six-­‐group  off-­‐peak  staggered   charging,  or  a  more  advanced  coordinated  strategy  is  not  developed.  Also,  the   commercial  and  industrial  substations  will  not  draw  coordinated  charging  loads,  further   reducing  the  likelihood  that  coordinated  charging  will  ever  be  an  issue  at  the  11  kV  level.        
  • 57. 46     46     5 Conclusion       This  thesis  looks  at  the  impacts  of  electric  vehicle  charging  on  the  low  and  medium   voltage  networks  for  both  uncoordinated  and  coordinated  charging  scenarios.  Sample   low  and  medium  voltage  network  models  provided  by  Endeavour  Energy  allowed  for   realistic  network  modelling  of  both  overhead  and  underground  networks,  while  smart   meter  data  from  premises  supplied  by  Woodlands  Drive  LV  substation  ensured  accurate   load  profile  patterns  and  realistic  phase  unbalance  for  simulation  in  the  PowerFactory   models.  Analysis  was  conducted  using  load  profiles  recorded  on  the  hottest  day  of  2011   to  create  a  worst-­‐case  base  loading  scenario  which  network  planning  must  be  based   upon.  A  significant  number  of  variables  have  been  considered  to  ensure  that  all  loading   possible  scenarios  have  been  taken  into  account.   The  development  of  a  GUI  resulted  in  a  powerful  tool  for  simulating  and   analysing  charging  scenarios  for  this  thesis,  and  will  serve  the  same  purpose  for   network  planners  at  Endeavour  Energy.  This  tool  integrates  MATLAB,  DIgSILENT   PowerFactory  and  Excel  for  remotely  controlled  analysis  of  an  unlimited  number  of   charging  scenarios.     The  simulation  of  EV  loading  scenarios  found  that  uncoordinated  charging  may   become  less  of  a  problem  than  expected  for  lightly  loaded  substations  and  those  in   underground  areas,  for  the  expected  charger  rating  of  4  kW.  Phase  unbalance,  however,   was  found  to  be  the  limiting  factor  in  overhead  networks,  limiting  transformer  loading   to  the  voltage  regulation  capability  of  the  network.  Based  on  EV  growth  projections,   transformers  that  reach  85%  or  below  during  the  hottest  days  of  the  year  are  expected   to  handle  EV  loading  until  at  least  2030  for  unbalanced  underground  and  overhead   networks,  however  an  increase  in  the  expected  charger  rating  and  average  driving   distance,  will  reduce  this  time  frame  in  which  overloading  will  occur  to  as  soon  as  this   decade.  The  impacts  of  uncoordinated  charging  were  found  to  be  worse  at  the  medium   voltage  level,  with  zone  substations  possibly  requiring  upgrades  by  the  next  decade  if  an   uncoordinated  strategy  is  not  implemented  at  the  low  voltage  level.     The  study  of  simple  off-­‐peak  coordinated  charging,  however,  determined  that   late-­‐night  charging  is  expected  to  avoid  any  overloading  issues  at  zone  and  LV   substations.  A  three-­‐group  off-­‐peak  coordinated  charging  strategy  allowed  100%   penetration  of  EVs  for  the  expected  charger  rating  of  4  kW,  while  a  six-­‐group  charging  
  • 58. 47     47     method  allowed  complete  EV  penetration  even  for  10  kW  rated  charging  on  substations   up  to  95%  loaded  on  the  hottest  day  of  the  year.      The  analysis  in  this  thesis  provides  an  accurate  guide  to  the  expected  loading   effects  of  EVs  in  the  coming  years,  and  the  ways  in  which  any  undesired  loading  effects   may  be  mitigated.                                        
  • 59. 48     48     References       [1]  C.  Connelly,  (2012,  May).  “Can  petrol  really  burn  out  and  fade  away  for  good?”   [Online].  Available:  http://www.news.com.au/technology/can-­‐petrol-­‐really-­‐burn-­‐out-­‐ and-­‐fade-­‐for-­‐good/story-­‐e6frfro0-­‐1226359173949   [2]  M.  Masoum,  P.  Moses,  and  S.  Hajforoosh,  “Distribution  transformer  stress  in  smart   grid  with  coordinated  charging  of  plug-­‐in  electric  vehicles”,  in  Innovative  Smart  Grid   Technologies  (ISGT),  2012  IEEE  PES,  Jan.  2012,  pp.  1  –  8.       [3]  K.  Clement-­‐Nyns,  E.  Haesen,  and  J.  Driesen,  “The  impact  of  charging  plug-­‐in  hybrid   electric  vehicles  on  a  residential  distribution  grid”,  Power  Systems,  IEEE  Transactions  on,   vol.  25,  no.  1,  pp.  371  –  380,  Feb.  2010.     [4]  M.  S.  S.  J.  D.  Glover  and  T.J.  Overbye,  Power  System  Analysis  and  Design,  5th  ed.   Stamform,  CT:  C.  M.  Shortt,  2012.     [5]  S.  A.  N.  Zealand,  “As/nzs  3000:2007”,  july  2009.   [6]  N.  F.  E.  Command.  (1990)  Electric  power  distribution  system  operations.  [Online].   Available:  http://www.wbdg.org/ccb/NAVFAC/OPER/mo201.pdf.     [7]  X.  Liu,  A.  Aichhorn,  L.  Liu,  and  H.  Li,  “Coordinated  control  of  distributed  energy   storage  system  with  tap  changer  transformers  for  voltage  rise  mitigation  under  high   photovoltaic  penetration”,  Smart  Grid,  IEEE  Transactions  on,  vol.  3,  no.  99,  pp.  1  –  10,   2012.     [8]  P.  Kadurek,  J.  Cobben,  and  W.  Kling,  “Smart  mv/lv  transformer  for  future  grids”,  in   Power  Electronics  Electrical  Drives  Automation  and  Motion  (SPEEDAM),  2010   International  Symposium  on,  June  2010,  pp.  1700  –  1705.     [9]  A.  Keyhani,  Design  of  Smart  Power  Grid  Renewable  Energy  Systems.  Hobo-­‐ken,  NJ:   Wiley  –  IEEE,  2011.     [10]  Tennessee  Valley  Authority.  (N.d.)  Types  of  electric  vehicles.  [Online].  Available:   http://www.tva.com/environment/technology/car_vehicles.htm.     [11]  Holden.  (2012)  Holden  volt  coming  soon.  [Online].  Available:   http://www.holden.com.au/pages/volt-­‐coming-­‐soon.     [12]  M.  T.  Thompson.  (N.d.)  Generic  battery  technology  comparison.  [Online].  Available   http://www.madkatz.com/ev/batteryTechnologyComparison.html.     [13]  T.  Motors.  (2012)  Model  s/options  and  pricing.  [Online].  Available:   http://www.teslamotors.com/models/options.    
  • 60. 49     49     [14]  Ford.  (2012)  2012  focus  electric.  [Online].  Available:   http://www.ford.com/electric/focuselectric/2012/.     [15]  Toyota.  (2012)  2012  Toyota  prius  plug-­‐in  hybrid  performance  and  specs.  [Online].   Available:  http://www.toyota.com/prius-­‐plug-­‐in/specs.html.     [16]  Nissan.  (2012)  Features  and  specifications.  [Online].  Available:   http://www.nissanusa.com/leaf-­‐electric-­‐car/specs-­‐features/index#/leaf-­‐electric-­‐ car/specs-­‐features/index.   [17]  International  Electrotechnical  Commission.  (2003)  International  standard  IEC   62196  –  1.  [Online].  Available:  http://webstore.iec.ch/p-­‐preview/info_iec62196-­‐ 1%7Bed1.0%7Den.pdf.     [18]  Society  of  Automotive  Engineers.  (2012)  Sae  electric  vehicle  and  plug  in  hybrid   electric  vehicle  conductive  charge  coupler.  [Online].  Available:   http://standards.sae.org/j1772_201202/.     [19]  P.  N.  R.  Lache,  D.  Galves,  “Vehicle  electrification”,  Mar.  2010.     [20]  B.  T.  T.  A.  Becker,  I.  Sidhu.  (2009)  Electric  vehicles  in  the  United  States.  [Online].   Available:   http://cet.berkeley.edu/dl/CET_Technical%20Brief_EconomicModel2030_f.pdf.     [21]  R.  D.  P.  Harrup.  (2009)  Hybrid  and  pure  electric  cars  2012  –  2022.  [Online].   Available:  http://www.idtechex.com/research/reports/hybrid-­‐and-­‐pure-­‐electric-­‐cars-­‐ 2012-­‐2022-­‐000287.asp.     [22]  E.  Sortomme,  M.  Hindi,  S.  MacPherson,  and  S.  Venkata,  “Coordinated  charging  of   plug-­‐in  hybrid  electric  vehicles  to  minimize  distribution  system  losses”,  Smart  Grid,  IEEE   Transactions  on,  vol.  2,  no.  1,  pp.  198  –  205,  March  2011.     [23]  K.  Qian,  C.  Zhou,  M.  Allan,  and  Y.  Yuan,  “Load  model  for  prediction  of  electric  vehicle   charging  demand”,  in  Power  System  Technology  (POWERCON),  2010  International   Conference  on,  Oct.  2010,  pp.  1  –  6.   [24]  K.  Qian,  C.  Zhou,  M.  Allan,  Y.  Yuan,  “Modeling  of  load  demand  due  to  ev  battery   charging  in  distribution  systems”,  Power  Systems,  IEEE  Transactions  on,  vol.  26,  no.  2,  pp.   802  –  810,  May  2011.     [25]  S.  Huang  and  D.  Infield,  “The  potential  of  domestic  electric  vehicles  to  contribute  to   power  system  operation  through  vehicle  to  grid  technology”,  in  Universities  Power   Engineering  Conference  (UPEC),  Proceedings  of  the  44th  International,  Sept.  2009,  pp.  1-­‐5.    
  • 61. 50     50     [26]  A.  Masoum,  S.  Deilami,  P.  Moses,  and  A.  Abu-­‐Siada,  “Impact  of  plug-­‐in  electrical   vehicles  on  voltage  profile  and  losses  of  residential  system”,  in  Universities  Power   Engineering  Conference  (AUPEC),  2010  20TH  Australasian,  Dec.  2010,  pp.  1  –  6.     [27]  S.  Deilami,  A.  Masoum,  P.  Moses,  and  M.  Masoum,  “Real-­‐time  coordination  of  plug-­‐in   electric  vehicle  charging  in  smart  grids  to  minimize  power  losses  and  improve  voltage   profile”,  Smart  Grid,  IEEE  Transactions  on,  vol.  2,  no.  3,  pp.  456  –  467,  Sept.  2011.     [28]  Bureau  of  Transport  Statistics  NSW,  “Household  travel  survey  (HTS)”,  Electronic   Publication  No.  E2012-­‐02-­‐HTS-­‐Summary   [29]  Australian  Bureau  of  Statistics  (ABS),  “2011  Census  QuickStats”  [Online],   Availalable:http://www.censusdata.abs.gov.au/census_services/getproduct/census/20 11/quickstat/0      
  • 62. 51     51         Appendix  A     Project  Plan  and  Specification      
  • 63. 52     52     Project  Specification     As  the  penetration  of  electric  vehicles  increases,  charging  of  these  vehicles  is  expected  to   have  significant  loading  effects  on  the  distribution  network,  similar  to  the  effects  caused   by  the  growing  number  of  air  conditioners  over  recent  years.  The  aim  of  this  thesis  is  to   continue  with  analysis  from  ECTE451  to  determine  the  impacts  of  uncoordinated   charging,  as  well  as  coordination  strategies  aimed  at  avoiding  these  loading  effects.  The   coordinated  charging  strategy  of  focus  will  be  staggered  off-­‐peak  charging,  and  novel   coordinated  charging  solutions  explored  in  ETE451,  such  as  charging  during  periods  of   high  solar  energy  generation,  and  supplying  power  to  the  grid  through  vehicle-­‐to-­‐grid   (V2G),  will  no  longer  be  considered.  The  focus  of  the  analysis  will  move  from  a  single   simplified  feeder  that  must  be  modified  manually  for  different  charging  penetrations,  to   an  automated  package  that  will  allow  network  planners  to  select  penetrations  levels,  as   well  as  variables  such  as  network  type  and  temperature,  to  determine  the  effects  of   charging  from  a  simple  user  interface.     Network  Modelling     Prior   study   in   ECTE451   involved   the   investigation   of   transformer   loading   and   line   voltage  levels  for  a  400  V  residential  feeder  using  PowerWorld  Simulator.  This  analysis   assumed  approximated  load  profiles  based  on  crude  peak  demand  estimates  based  on   typical   appliance   use,   and   estimated   network   equipment   ratings.   To   improve   on   this,   analysis   will   be   conducted   using   DIgSILENT   PowerFactory   models   provided   by   Endeavour  Energy.  These  models  cover  both  the  400  V  and  11  kV  distribution  levels,  for   underground,  overhead  and  semi-­‐rural  areas.  As  well,  smart  metering  load  profile  data   from  the  network  area  of  Glenmore  Park  has  been  provided  from  both  residential  and   zone   substation   meters,   containing   both   summer   and   winter   load   profiles.     This   will   allow  an  accurate  investigation  into  the  effects  of  charging  on  the  Glenmore  Park  area,   and  provide  a  reliable  insight  into  the  effects  on  the  remainder  of  the  Endeavour  Energy   network.     Using  these  models  as  a  starting  point,  they  will  be  modified  to  reflect  the  number  of   premises  on  the  Glenmore  Park  distribution  substation  containing  the  residential  smart   meters  of  interest.  The  aim  will  be  to  determine  the  effects  of  electric  vehicle  charging   on   transformer   loading   and   line   voltages,   for   both   uncoordinated   and   coordinated   charging  strategies.     Utility  Planning  Tool     In  order  to  run  a  load  flow  in  PowerFactory,  a  DIgSILENT  Programming  Language  (DPL)   script   must   be   executed.   The   DPL   scripts   provided   by   Endeavour   Energy   allow   line   voltages   and   transformer   loading   to   be   determined.   As   the   effects   of   charging   can   be   determined   with   the   DIgSILENT   projects   currently   provided,   the   focus   of   this   project   will  be  to  integrate  the  existing  scripts  into  a  package  for  network  planners  to  easily  run   a  number  of  penetration  scenarios.  Loads  in  DIgSILENT  are  associated  with  a  single  CSV   file   containing   data   for   specified   time   intervals,   which   must   be   manually   changed   to   determine  a  different  charging  penetration.  This  configuration  would  make  it  tedious  to   determine  the  effects  of  a  number  of  penetration  levels  using  a  single  project.  In  order  to   overcome  this,  MATLAB  will  be  used  as  an  interface  for  selecting  variables,  controlling   DPL  scripts  and  displaying  results.  The  scripting  will  work  as  follows:    
  • 64. 53     53     1.   MATLAB   will   act   as   an   interface   where   variables   are   selected.   These   variables   will   include   location   (to   specify   the   average   charging   time   based   on   average   driving   distance),   charging   power,   charging   penetration   level,   charging   strategy,   temperature   and  network  type.   2.  MATLAB  will  run  a  script  to  copy  specified  csv  files  to  the  master  files  associated  with   each  load  in  the  selected  DIgSILENT  model.   3.  DIgSILENT  will  run  the  existing  400  V  DPL  script  once  called  by  Matlab,  to  determine   low  voltage  transformer  loading  and  line  voltage  data.     4.   MATLAB   will   write   the   400   V   transformer   load   profile   to   the   single   11   kV   csv   file   associated  with  each  distribution  substation  load  in  the  11  kV  model.   5.  DIgSILENT  will  run  the  existing  11  kV  DPL  script  once  called  by  Matlab,  to  determine   medium  voltage  transformer  loading  and  line  voltage  data.     6.  MATLAB  will  display  both  the  400  V  and  11  kV  transformer  loading  and  line  voltage   data.     This  MATLAB-­‐controlled  arrangement  will  allow  an  Endeavour  Energy  network  planner   to  run  scenarios  quickly  and  simply  to  determine  the  impacts  of  charging  for   consideration  in  future  network  planning.                            
  • 65. 54     54     Appendix  B     Logbook  Summary  Signature  Sheet      
  • 66. 55     55         Appendix  C     Software  Documentation      
  • 67. 56     56     MATLAB  GUI  Code     function varargout = secondGUI(varargin) % Initialisation code generated by MATLAB not shown %~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~START~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~% % --- Executes on button press in pushbutton1. function pushbutton1_Callback(hObject, eventdata, handles) network_type=handles.network_type; coordination=handles.coordination; chargepower_val=handles.chargepower_val; penetration_val=handles.penetration_val; disp Variables:; disp ' '; disp(network_type); disp(coordination); disp(chargepower_val); disp(penetration_val); % save variables from callback function where created oldpointer = get(handles.figure1, 'pointer'); set(handles.figure1, 'pointer', 'watch') % display hourglass to show loading drawnow; %determine selected variables and call appropriate matlab functions if strcmp(network_type,'Overhead')&& strcmp(coordination,'Uncoordinated') uncoEVcsv(chargepower_val,penetration_val); OH_batch(); OH_excelcmd(); [dailykva,p,ploc,v,vloc]=OH_getimportant() elseif strcmp(network_type,'Underground')&& strcmp(coordination,'Uncoordinated') uncoEVcsv(chargepower_val,penetration_val); UG_batch(); UG_excelcmd(); [dailykva,p,ploc,v,vloc]=UG_getimportant() elseif strcmp(network_type,'Overhead')&& strcmp(coordination,'Staggered 10 pm start') staggeredEVcsv(chargepower_val,penetration_val); OH_batch(); OH_excelcmd(); [dailykva,p,ploc,v,vloc]=OH_getimportant() elseif strcmp(network_type,'Underground')&& strcmp(coordination,'Staggered 10 pm start') staggeredEVcsv(chargepower_val,penetration_val); UG_batch();
  • 68. 57     57     UG_excelcmd(); [dailykva,p,ploc,v,vloc]=UG_getimportant() else errordlg('Please ensure all network variables have been selected'); end %run 11kv model using matlab functions writeto11kvloads(dailykva); HV_batch(); HV_excelcmd(); [dailykva11,p11,ploc11,v11,vloc11]=HV_getimportant(); zsub=45-p11; pow11=[p11,zsub]; %calculate time of max power and min voltage ptime = gettime(ploc); vtime = gettime(vloc); ptime11 = gettime(ploc11); vtime11 = gettime(vloc11); %get 400 V data and display on graph and in text box dsub=250-p; pow=[p,dsub]; pie3(handles.axes1,pow); handles.list_item1='Transformer Loading'; handles.list_item2='400 V Network'; disp(pow); powerLVtable=['Maximum Substation Power = ', num2str(p), ' KVA at ' num2str(ptime)]; set(handles.edit3,'String',powerLVtable); %save data in handles so can be passed into other functions handles.pow=pow; handles.v=v; handles.p=p; handles.p11=p11; handles.pow11=pow11; handles.v11=v11; handles.ptime=ptime; handles.vtime=vtime; handles.ptime11=ptime11; handles.vtime11=vtime11; guidata(hObject, handles); set(handles.figure1, 'pointer', oldpointer) cd c:UsersOwnerDesktop'Matlab Scripts'; % --- Executes on selection change in popupmenu1. function popupmenu1_Callback(hObject, eventdata, handles) val1 = get(hObject,'Value'); string_list = get(hObject,'String');
  • 69. 58     58     network_type = string_list{val1}; % Convert from cell array to string handles.network_type=network_type; guidata(hObject, handles); % update handles % --- Executes during object creation, after setting all properties. function popupmenu1_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end % --- Executes on selection change in popupmenu2. function popupmenu2_Callback(hObject, eventdata, handles) val3 = get(hObject,'Value'); string_list = get(hObject,'String'); coordination = string_list{val3}; % Convert from cell array to string handles.coordination=coordination; guidata(hObject, handles); % --- Executes during object creation, after setting all properties. function popupmenu2_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end % --- Executes on slider movement. %get penetration value function slider1_Callback(hObject, eventdata, handles) penetration_val = round(get(hObject,'Value')); set(handles.edit1,'String',penetration_val); handles.penetration_val=penetration_val; guidata(hObject, handles); % --- Executes during object creation, after setting all properties. %locates and sizes background image function slider1_CreateFcn(hObject, eventdata, handles) if isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor',[.9 .9 .9]); end function edit1_Callback(hObject, eventdata, handles) % --- Executes during object creation, after setting all properties. % creates slider 1 on open GUI
  • 70. 59     59     function edit1_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end % --- Executes on slider movement. % get charging power function slider2_Callback(hObject, eventdata, handles) chargepower_val = get(hObject,'Value'); set(handles.edit2,'String',chargepower_val); handles.chargepower_val=chargepower_val; guidata(hObject, handles); % --- Executes during object creation, after setting all properties. %creates slider 2 on opening GUI function slider2_CreateFcn(hObject, eventdata, handles) if isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor',[.9 .9 .9]); end % --- Executes during object creation, after setting all properties. % sizes and positions background function figure1_CreateFcn(hObject, eventdata, handles) ha = axes('units','normalized', ... 'position',[0 0 1 1]); % Move the background axes to the bottom uistack(ha,'bottom'); I=imread('bestbanner.png'); hi = imagesc(I); colormap gray % Turn the handlevisibility off, make the axes invisible set(ha,'handlevisibility','off', ... 'visible','off') % --- Executes on selection change in listbox1. % determines which list item is selected function listbox1_Callback(hObject, eventdata, handles) index_selected = get(hObject,'Value'); list = get(hObject,'String'); list_item1 = list{index_selected}; handles.list_item1=list_item1; guidata(hObject, handles); disp(list_item1);
  • 71. 60     60     list_item2=handles.list_item2; pow=handles.pow; p=handles.p; v=handles.v; pow11=handles.pow11; p11=handles.p11; v11=handles.v11; ptime=handles.ptime; vtime=handles.vtime; ptime11=handles.ptime11; vtime11=handles.vtime11; % Set text data in text box voltageLVtable=['Lowest bus voltage = ', num2str(v), ' V at ' ,num2str(vtime)]; powerLVtable=['Maximum Substation Power = ', num2str(p), ' KVA at ' ,num2str(ptime)]; voltageHVtable=['Lowest bus voltage = ', num2str(v11), ' kV ' ,num2str(vtime11)]; powerHVtable=['Maximum Substation Power = ', num2str(p11), ' MVA ',num2str(ptime11)]; %determines variables selected when list box 1 has been selected if strcmp(list_item1,'Transformer Loading') && strcmp(list_item2,'400 V Network') display '400 TX Loading works' pie3(handles.axes1,pow); set(handles.edit3,'String',powerLVtable); elseif strcmp(list_item1,'Line Voltages') && strcmp(list_item2,'400 V Network') display '400 volt Loading works' bar(handles.axes1,v,'stacked'); axis([0.75 1.25 200 250]); set(gca, 'XTickLabelMode', 'Manual'); set(gca, 'XTick', []); set(handles.edit3,'String',voltageLVtable); elseif strcmp(list_item1,'Transformer Loading') && strcmp(list_item2,'11 kV Network') display '11 kv TX Loading works' pie3(handles.axes1,pow11); set(handles.edit3,'String',powerHVtable); elseif strcmp(list_item1,'Line Voltages') && strcmp(list_item2,'11 kV Network') display '11 kv volt Loading works' bar(handles.axes1,v11,'stacked'); axis([0.75 1.25 10 11]); set(gca, 'XTickLabelMode', 'Manual'); set(gca, 'XTick', []); set(handles.edit3,'String',voltageHVtable); end % --- Executes during object creation, after setting all properties. % creates list box on opening GUI function listbox1_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
  • 72. 61     61     set(hObject,'BackgroundColor','white'); end % --- Executes during object creation, after setting all properties. %sets graph axes function axes1_CreateFcn(hObject, eventdata, handles) set(gca, 'YTickLabelMode', 'Manual') set(gca, 'YTick', []) set(gca, 'XTickLabelMode', 'Manual') set(gca, 'XTick', []) % --- Executes on selection change in listbox2. % creates list box on opening GUI function listbox2_Callback(hObject, eventdata, handles) index_selected = get(hObject,'Value'); list = get(hObject,'String'); list_item2 = list{index_selected}; disp(list_item2); handles.list_item2=list_item2; guidata(hObject, handles); list_item1=handles.list_item1; pow=handles.pow; p=handles.p; v=handles.v; pow11=handles.pow11; p11=handles.p11; v11=handles.v11; ptime=handles.ptime; vtime=handles.vtime; ptime11=handles.ptime11; vtime11=handles.vtime11; % Set Text data voltageLVtable=['Lowest bus voltage = ', num2str(v), ' V at ' ,num2str(vtime)]; powerLVtable=['Maximum Substation Power = ', num2str(p), ' KVA at ' ,num2str(ptime)]; voltageHVtable=['Lowest bus voltage = ', num2str(v11), ' kV ' ,num2str(vtime11)]; powerHVtable=['Maximum Substation Power = ', num2str(p11), ' MVA of 45 MVA ',num2str(ptime11)]; %determines variables selected when list box 1 has been selected if strcmp(list_item1,'Transformer Loading') && strcmp(list_item2,'400 V Network') display '400 TX Loading works' pie3(handles.axes1,pow); set(handles.edit3,'String',powerLVtable); elseif strcmp(list_item1,'Line Voltages') && strcmp(list_item2,'400 V Network')
  • 73. 62     62     display '400 volt Loading works' bar(handles.axes1,v,'stacked'); axis([0.75 1.25 200 250]); set(gca, 'XTickLabelMode', 'Manual'); set(gca, 'XTick', []); set(handles.edit3,'String',voltageLVtable); elseif strcmp(list_item1,'Transformer Loading') && strcmp(list_item2,'11 kV Network') display '11 kv TX Loading works' pie3(handles.axes1,pow11); set(handles.edit3,'String',powerHVtable); elseif strcmp(list_item1,'Line Voltages') && strcmp(list_item2,'11 kV Network') display '11 kv volt Loading works' bar(handles.axes1,v11,'stacked'); axis([0.75 1.25 10 11]); set(gca, 'XTickLabelMode', 'Manual'); set(gca, 'XTick', []); set(handles.edit3,'String',voltageHVtable); end % --- Executes during object creation, after setting all properties. function listbox2_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end function edit3_CreateFcn(hObject, eventdata, handles) if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor')) set(hObject,'BackgroundColor','white'); end Batch  file  to  call  command  file  to  open  DIgSILENT  Engine  connection:     digrcom  -­‐d  -­‐p  ncacn_ip_tcp  -­‐n  127.0.0.1  -­‐e  2001  -­‐f="11kvcmd.cmd"     Command  file:     cls/inp   cls/out   ac/de  all   ac  Matthew  WardhaughOH  11kV  Feeder   cd  Matthew  WardhaughOH  11kV  FeederLibraryScripts   11kV  Losses  and  Voltages  script     Command  file  to  control  Excel  files:  
  • 74. 63     63       cd    c:UsersOwnerDesktopDIgSILENT  Results11kV   del  "11kvvoltresults.csv"   del  "11kvTXresults.csv"   rename  "11kvvoltresults.txt"  "11kvvoltresults.csv"   rename  "11kvTXresults.txt"  "11kvTXresults.csv"   cd    c:DIgSILENTpf141ENGINE   11kvexcelVBScript     VBScript  file  to  format  CSV  files:     Set  objExcel  =  CreateObject("Excel.Application")     objExcel.Visible  =  False   objExcel.Workbooks.Open("C:UsersOwnerAppDataRoamingMicrosoftAddInstxt 2colsmacro.xlam")   objExcel.Workbooks.Open("C:UsersOwnerDesktopDIgSILENT   Results11kv11kvvoltresults.csv")   objExcel.Run  "txt2cols"   objExcel.ActiveWorkbook.Save   objExcel.ActiveWorkbook.Close   objExcel.Workbooks.Open("C:UsersOwnerAppDataRoamingMicrosoftAddInstxt 2colsmacro.xlam")   objExcel.Workbooks.Open("C:UsersOwnerDesktopDIgSILENT   Results11kv11kvTXresults.csv")   objExcel.Run  "txt2cols"   objExcel.ActiveWorkbook.Save   objExcel.ActiveWorkbook.Close   objExcel.Quit     VBA  “txt2cols”  Macro  to  remove  text  and  zero  columns:     Sub  txt2cols()   '   '  txt2cols  Macro   '  Runs  text  to  columns  on  column  A  to  split  digsilent  result  data  to  csv  format   Application.DisplayAlerts  =  False     'Set  up  the  selection  range   Dim  ColumnA  As  Range   Dim  Text  As  Range   Set  ColumnA  =  Range("A:A")   Set  Text  =  Range("1:2")     'Run  Text  to  Columns  function   ColumnA.TextToColumns  _              Destination:=Range("$A$1"),  _              DataType:=xlDelimited,  _              TextQualifier:=xlDoubleQuote,  _              ConsecutiveDelimiter:=False,  _              Tab:=True,  _  
  • 75. 64     64                Semicolon:=False,  _              Comma:=False,  _              Space:=False,  _              Other:=False     Text.Delete   ColumnA.Delete     'Delete  zero  columns   Dim  nLastColumn  As  Long   Set  r  =  ActiveSheet.UsedRange   nLastColumn  =  r.Columns.Count  +  r.Column  -­‐  1   For  i  =  nLastColumn  To  1  Step  -­‐1   i1  =  Application.WorksheetFunction.Sum(Columns(i))   i2  =  Application.WorksheetFunction.Count(Columns(i))   i3  =  Application.WorksheetFunction.CountA(Columns(i))   If  i1  =  0  And  i2  =  i3  Then   Columns(i).Delete   End  If   Next   End  Sub     MATLAB  Functions     Function  to  call  DIgSILENT  Engine  batch:     function OH_batch() cd c:DIgSILENTpf141ENGINE; system('OHbatch.bat'); cd c:UsersOwnerDesktop'Matlab Scripts'; end   Function  to  call  Excel  command  file:     function OH_excelcmd() cd c:DIgSILENTpf141ENGINE; system('OHexcelcmd.cmd'); cd c:UsersOwnerDesktop'Matlab Scripts' end   Function  to  extract  loading  data  from  CSV  files:     function [dailykva,maxkva,ploc,minvolt,vloc]=OH_getimportant() cd c:UsersOwnerDesktop'DIgSILENT Results'Overhead volt=csvread('OHvoltresults.csv'); dailykw=csvread('OHTXresults.csv',0,0,[0,0,47,0]); %47 is 48th row dailykvar=csvread('OHTXresults.csv',0,1,[0,1,47,1]); %47 is 48th row dailykva=2*roundn(sqrt(power(dailykw,2)+ power(dailykvar,2)),-1); [minvolt, location] = min(volt(:)); minvolt=minvolt/230; [vloc, y] = ind2sub(size(volt),location); [maxkva, location] = max(dailykva(:));
  • 76. 65     65     [ploc, y] = ind2sub(size(dailykva),location); minfirstbusvolt=csvread('OHvoltresults.csv',36,3,[36,3,36,3]);%36th row = 6 pm (not 37 with matlab convention) cd c:UsersOwnerDesktop'Matlab Scripts' end Function  to  get  time  of  max  power  and  min  voltage:     function [time]=gettime(loc) loc = loc/2; if loc == 0.5 time = ['12 am']; elseif loc == 1 time = ['12:30 am']; elseif loc<12 if rem(loc,1) == 0 time = [num2str(loc-1),':30 am']; else loc=loc-0.5; time = [num2str(loc),':00 am']; end else if rem(loc,1) == 0 loc = loc-12; time = [num2str(loc-1),':30 pm']; else loc=loc-12.5; time = [num2str(loc),':00 pm']; end end   [Remaining  functions/scripts/files  on  disc]