Possibility of network voltage control

using demand side management

Prof Jovica V. Milanović Dipl.Ing., MSc, PhD, DSc, CEng, FIET, FIEEE, Foreign member of SAES

Deputy Head of School & Director of External Affairs

Manchester, M13 9PL, United Kingdom

33rd Symposium CIGRÉ Serbia

6th June 2017, Zlatibor, Serbia,

School of Electrical & Electronic Engineering

Outline of presentation

• Network & Voltage controllability challenges

• Advanced demand profiling & management

• How it could be used for voltage control

• Summary

Network & Voltage controllability

challenges

The drivers

• Evolving/new market structures and operation

• Increasingly liberalised market

• Increased cross-boarder bulk power transfers to facilitate effectiveness of market mechanisms

• The participation of controllable (that can be both, controlled and used

for control) plant in generation mix is reducing

• The nature and behaviour of load has changed and is changing

(e.g., spatial in addition to temporal variation)

• New transmission components (still insufficiently understood,

compared to “old” technologies) are being added to the system

Controllability challenge

• The complexity and vulnerability of the system is increasing

• The system control, stability and security are becoming

increasingly important and much more time dependent than before

CONTROL CENTRE

CENTRAL (BULK)

POWER GENERATION ANCILLARY SERVICES

• “Every little helps” is becoming very important

– Additional/Advanced controllability of RES

– Deployment and control of energy storage

– Efficient Demand Side Management (DSM)

Advanced demand profiling &

management

(Advanced) Demand profiling

Identification/estimation and forecasting of static

composition (in terms of load categories and load

controllability) and dynamic response of demand

Why (or) Do we need to consider it ?

Efficient DSM

• To apply efficiently Demand Side Management (shift load from the

hours of high consumption to hours of low consumption) to facilitate

required “change in paradigm” and enable load-follows-generation

approach for efficient integration of RES we need to be able to

forecast (reasonably accurately) the demand from 30 min (or less)

to a day (or more) in advance and to control it.

– At present only demand for real power at bulk supply points is forecasted

• Forecasting only total demand for real power (P) is not sufficient

– the information about the available reactive power (Q) is required as well,

e.g., for voltage regulation

Efficient DSM

• Knowledge of total demand (be it accurate and for both P and Q) is

not sufficient

- the information about demand composition in terms of controllable

(that can be shifted from one hour to the next) and uncontrollable load

is needed

• Knowledge of demand composition is not sufficient

- the information about load category (based on dynamic response to

system disturbances) mix in both controllable and uncontrollable part

of demand is needed

• Knowledge of demand composition in terms of load category is

not sufficient

- the information about dynamic response of aggregate demand at

bulk supply point before and after the shift of controllable load is

needed

Estimation of demand composition

1. Assuming that we have online measurements from

advanced smart meters installed at individual

customer sites

Demand profile

Decomposed Load Curves

based on Load Categories

Decomposed Load Curves

based on Load Type

Decomposed Load Curves

based on Load Catrollability Decomposed Load Curves

based on Load Categories

00:00 06:00 12:00 18:00 23:590

100

200

300

400

Time [h]

Reac

tive p

ower

[kva

r]

Average

Mean

Most probable

Typical

Controllable/uncontrollable load Load categories

Probabilistic estimation of Q

2. Assuming that we don’t have any online

measurements from individual customers

Probabilistic estimation of demand

profile & controllability

On-line measurement

using Smart Meters

Data processing &

Aggregation based

on Load categories

& controlability

Deterministic DDLC

Based on Categories

& Controlability

Participation of Load

Categories in Different

Load Classes

Metering Data Customer Surveys General Knowledge of Demand

Composition

DDLC Based on Load

Classes DDLC Based on

Load Categories

& Controlability

Data Processing

with Parameter

Uncertainty

Probabilistic DDLC

Based on Categories

& Controlability

On-line measurement

using Smart Meters

Data processing &

Aggregation based

on Load categories

& controlability

Deterministic DDLC

Based on Categories

& Controlability

Is this all we need?

0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24

40

60

80

100

Hourtime (sec)

Po

wer, P

(%

)

0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24-20

-10

0

10

20

Hourtime (sec)

Reactiv

e P

ow

er,Q

(%

)

• Demand composition affects demand (load) dynamic response

following small or large voltage (predominantly) or frequency

disturbance in the network

•Changing of transformer tap position

• Fault in the system

• Different dynamic response of the load will affect overall dynamic

response of the system (voltage, frequency or angular stability of

the system)

•The contribution of the load (with given composition) to

system dynamic response depends on

• The size of the load

• The location of the load

Responses of different load categories

0 1 2 3 4 50.65

0.7

0.75

0.8

0.85

0.9

0.95

1

1.05

t (sec)

P (p

.u)

a)

Upper Limit

Lower Limit

Most Probable Response

Average Response

Dish Washer

Refrigerator

Room Air Conditioner

Washing Machine

Dryer

0 1 2 3 4 50.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

t (sec)

Q (p

.u)

b)

a) real power responses, and b) reactive power

response for residential and commercial motors with

most probable and average response specified

0 0.2 0.4 0.6 0.8 1 1.20.92

0.94

0.96

0.98

1

1.02

t (sec)

P (p

.u)

a)

0 0.2 0.4 0.6 0.8 1 1.20.92

0.94

0.96

0.98

1

1.02

t (sec)

P (p

.u)

b)

0 0.2 0.4 0.6 0.8 1 1.20.15

0.3

0.45

0.6

t (sec)

Q (p

.u)

c)

0 0.2 0.4 0.6 0.8 1 1.20.15

0.3

0.45

0.6

t (sec)

Q (p

.u)

d)

Upper Limit

Lower Limit

Average Response

Most Probable Response

a) small industrial IM P response,

and b) large industrial IM P

response to step down voltage; c)

small industrial IM Q response,

and d) large industrial IM Q

response to step down voltage.

(The upper and lower limit and the

average responses are also shown.)

0 0.5 1 1.5 2-0.5

-0.4

-0.3

-0.2

-0.1

0

time(sec)

Q (p

.u.)

a)

500 samples

-0.5 -0.4 -0.3 -0.2 -0.1 00

0.02

0.04

0.06

Steady State Q(p.u)

Prob

abilit

y

b)

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-0.5-0.4-0.3-0.2-0.1

00.10.20.30.40.5

time(sec)

Q (p

.u)

c)

Lower Limit

Most Probable Response

Upper Limit

Real power responses to a step reduction in voltage for

SMPS: a) responses obtained from 500 MC simulations; b)

probability histogram of steady-state power after voltage

drop; c) upper and lower limit of the responses and the

most probable response

The effect of demand profile on dynamic

response of the load

0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24

40

60

80

100

Hourtime (sec)

Po

wer, P

(%

)

0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24-20

-10

0

10

20

Hourtime (sec)

Reactiv

e P

ow

er,Q

(%

)

Comparison of different a) P and b) Q responses at different times of day (solid

line: 03:00; dashed line: 04:00; dash-dot line: 12:00; dotted line: 18:00)

How it can be used for voltage control

Effect of demand management

24

46

17

28

16

14

12

26

21 19

23

223 22

18

15

54

52

53

230

50

75

228

74

229

20221

51 76

13

268

87

48 47

49222

43

42

41

40

39

38

37

269

235 236

231

78

79

80

81

85

88

290

82

83

84

291

91

92

93

95

94

96

97 98

101 99

100

102

103 104

105

106 107

108

109

110

111

112

113

114 115

116

121

122

123124

125

126

127

128

117

118

119 120

86

11

10

8

9

7

6

5

4

55

1 65

64

66 63

57

226 227

58

149

147

154

155

150 153 156

148 146

145

141

143

142

144

140

139

129

130133

131

135

137 136

138

134

157

161

158

186

184

132

160

165

162

163

164

166

167

168169

170

180

181

182 183

185

187

188

189

190 191

192

193

194

197

198

200 199

201

202

203 204

205

206

207

208209

151

152

224

232

77

159

225

215

216

211

212210

213

214

217

218 171

219

220

172

173

174

175

176

177

178

62

60

59

61

2

3

72

70

179

25

27

29

30

32

31 33

3435

36

7168

67 69

73

89

45

44

249 250

266267

242

252

260

289

237

244

261

251

241 245246 243

262272270253 274271 276 275 263 264 273

240

254258 256259

277

247

278

248

280

234

279

233255257

293 292

294 295 297 296

299 298

300

77238

288

269

287 285286

56

BA C D E

F

H I J K L

ON

G

132kV

33kV

11kV

3.3kV

275kV

400kV

G

195

196

Commercial

Industrial

Mixed

IM; 15%

I; 5%

Z; 24% P; 56%

Lights;

16% Cooking;

6%

H Water;

10%

Space

Heating;

18%

Appliaces;

50%

Domestic electricity

consumption broken

down by end use. ~

~ ~

G1

G2 G3

BSP of DN C

2

7

8

93

3005

4

1

10

11

BSP of DN B

BSP of DN A

Change in demand & composition

Exponential load model

00

V

VPP

Z

(K=2)

I (K=1)

P (K=0)

V (

p.u

.)

Power (p.u.)

A

BC

0

50

100

150

200

250

300

350

400

450

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Po

wer

, M

W

Hours

Aggeregated load profile

Base

Shift Dom. IM+Z

Shift Dom.+Com. IM

Change in demand composition

Effect on voltage levels – TN & DN

Voltage level of transmission (above) and distribution (below) network for shifting domestic load

(Z+IM) at peak time. Left – before the shift; Right – after shifting Z+IM load at domestic buses

Sensitivity analysis of PV curves - DN

ΔV/ ΔP = (V0-Vc)/(Po-Pc) from operating point to

critical point at peak (above) and low (below) time

Summary

0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24

40

60

80

100

Hourtime (sec)

Po

wer, P

(%

)

0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24-20

-10

0

10

20

Hourtime (sec)

Reactiv

e P

ow

er,Q

(%

)

Given

On-line measurements form installed smart meters

General information about demand composition at a bus (e.g., 30%

commercial, 10% industrial and 60% domestic)

Standard P, V and Q (or a combination of those) half-hourly (or at other

time intervals) measurements over some period of time in the past

Weather forecast for the following day

It can be forecasted/estimated

• Total P and Q demand

• Demand composition in terms of categories

• Demand composition in terms of controllable and non-controllable

demand

• Dynamic P and Q response

at that bus (aggregation point) at any given time “now” or in the future)

and how it will change if we “move” part of demand to a different hour

Summary

0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24

40

60

80

100

Hourtime (sec)

Po

wer, P

(%

)

0 1 2 3 4 5 6 7 8 0 3 6 9 12 15 18 21 24-20

-10

0

10

20

Hourtime (sec)

Reactiv

e P

ow

er,Q

(%

)

By embedding information of demand composition in different

network analysis programmes (either as a part of objective

function or as a constraint) greater flexibility and cost

effectiveness of network performance can be achieved with

respect to all steady state and transient network performance

criteria/attributes. The influence of smart meter coverage and missing data on the

accuracy of load decomposition?

DR based on price

24

46

17

28

16

14

12

26

21 19

23

223 22

18

15

54

52

53

230

50

75

228

74

229

20221

51 76

13

268

87

48 47

49222

43

42

41

40

39

38

37

269

235 236

231

78

79

80

81

85

88

290

82

83

84

291

91

92

93

95

94

96

97 98

101 99

100

102

103 104

105

106 107

108

109

110

111

112

113

114 115

116

121

122

123124

125

126

127

128

117

118

119 120

86

11

10

8

9

7

6

5

4

55

1 65

64

66 63

57

226 227

58

149

147

154

155

150 153 156

148 146

145

141

143

142

144

140

139

129

130133

131

135

137 136

138

134

157

161

158

186

184

132

160

165

162

163

164

166

167

168169

170

180

181

182 183

185

187

188

189

190 191

192

193

194

197

198

200 199

201

202

203 204

205

206

207

208209

151

152

224

232

77

159

225

215

216

211

212210

213

214

217

218 171

219

220

172

173

174

175

176

177

178

62

60

59

61

2

3

72

70

179

25

27

29

30

32

31 33

3435

36

7168

67 69

73

89

45

44

249 250

266267

242

252

260

289

237

244

261

251

241 245246 243

262272270253 274271 276 275 263 264 273

240

254258 256259

277

247

278

248

280

234

279

233255257

293 292

294 295 297 296

299 298

300

77238

288

269

287 285286

56

BA C D E

F

H I J K L

ON

G

132kV

33kV

11kV

0.4kV

275kV

400kV

G

A B CD HI

Substation A

JK L

0

1000

2000

3000

4000

5.63 6.11

2487.46

5.9

1692.28

4322.47

1722.30

Buildings

Pro

fit

[£]

JK

CD

HI

EBA

L

the profit resulting from the participation of consumers in

the DSM program based on price