1
00:00:00,480 --> 00:00:03,480
foreign

2
00:00:08,940 --> 00:00:11,900
thank you Peter

3
00:00:14,700 --> 00:00:20,460
quick caveat before I get started this

4
00:00:17,400 --> 00:00:21,960
is a case study about how we're using

5
00:00:20,460 --> 00:00:24,420
python we're not actually going to look

6
00:00:21,960 --> 00:00:26,340
at any pythony's Source the focus is on

7
00:00:24,420 --> 00:00:28,019
the high level problems we are using

8
00:00:26,340 --> 00:00:30,960
python to solve so if you want to look

9
00:00:28,019 --> 00:00:32,759
at some really hard python code you've

10
00:00:30,960 --> 00:00:35,000
got a minute to jump to the room next

11
00:00:32,759 --> 00:00:35,000
door

12
00:00:36,960 --> 00:00:41,940
what do you think of what image comes

13
00:00:39,420 --> 00:00:44,579
into your mind

14
00:00:41,940 --> 00:00:46,920
perhaps it's a lightning strike

15
00:00:44,579 --> 00:00:49,500
on a stormy day

16
00:00:46,920 --> 00:00:51,780
perhaps it's a domestic appliance like a

17
00:00:49,500 --> 00:00:54,440
kettle or a hot plate sorry about the

18
00:00:51,780 --> 00:00:54,440
Retro images

19
00:00:54,899 --> 00:00:59,219
machine from the new

20
00:00:56,100 --> 00:01:01,940
like an EV or a flat screen TV or maybe

21
00:00:59,219 --> 00:01:01,940
a wind farm

22
00:01:02,460 --> 00:01:06,119
where Josh and I work when we think of

23
00:01:04,739 --> 00:01:09,420
electricity

24
00:01:06,119 --> 00:01:12,720
we think of a perfectly formed sine wave

25
00:01:09,420 --> 00:01:15,600
with a frequency of exactly 50 hertz

26
00:01:12,720 --> 00:01:18,840
created by machines as big as a house

27
00:01:15,600 --> 00:01:20,759
spinning at precisely 3000 RPM

28
00:01:18,840 --> 00:01:23,820
dotted around the country

29
00:01:20,759 --> 00:01:26,580
all spinning in precise Harmony

30
00:01:23,820 --> 00:01:27,900
this rotating Mass leads to grid

31
00:01:26,580 --> 00:01:30,119
stability

32
00:01:27,900 --> 00:01:32,220
life was simple in the early days of

33
00:01:30,119 --> 00:01:35,479
what we now call the nem or the national

34
00:01:32,220 --> 00:01:35,479
electricity Market

35
00:01:35,820 --> 00:01:40,020
Josh and I work for a

36
00:01:37,380 --> 00:01:42,659
ustralian energy Market operator Josh

37
00:01:40,020 --> 00:01:44,640
leads amo's data science team within

38
00:01:42,659 --> 00:01:46,500
operational forecasting where they're

39
00:01:44,640 --> 00:01:49,500
responsible for predicting the impact of

40
00:01:46,500 --> 00:01:51,600
the weather on consumer Behavior and

41
00:01:49,500 --> 00:01:54,299
consumer behavior on both energy Supply

42
00:01:51,600 --> 00:01:56,700
and energy demand

43
00:01:54,299 --> 00:01:59,460
I support amo's power system Engineers

44
00:01:56,700 --> 00:02:02,520
with an in-house tool called app the

45
00:01:59,460 --> 00:02:05,820
aemo modeling platform we use this for

46
00:02:02,520 --> 00:02:09,239
modeling the grid for forward looking

47
00:02:05,820 --> 00:02:13,160
analysis and backward looking

48
00:02:09,239 --> 00:02:13,160
false fault studying

49
00:02:13,260 --> 00:02:17,520
the topics we're going to squeeze into

50
00:02:14,819 --> 00:02:21,120
the next 30 minutes together are as

51
00:02:17,520 --> 00:02:24,480
follows what IMO does how generation is

52
00:02:21,120 --> 00:02:26,940
changing why Python and how we use the

53
00:02:24,480 --> 00:02:31,700
weather forecast to help predict energy

54
00:02:26,940 --> 00:02:31,700
Supply and energy demand

55
00:02:32,580 --> 00:02:36,180
couple of bullet points about Amo first

56
00:02:35,280 --> 00:02:39,000
of all

57
00:02:36,180 --> 00:02:41,280
Amo was established by coeg the Council

58
00:02:39,000 --> 00:02:44,459
of Australian governments on the 1st of

59
00:02:41,280 --> 00:02:47,220
July 2009 to manage the national

60
00:02:44,459 --> 00:02:48,420
electricity Market or nem in the eastern

61
00:02:47,220 --> 00:02:51,180
states

62
00:02:48,420 --> 00:02:55,260
our predecessor nemco began operating

63
00:02:51,180 --> 00:02:56,700
the nem in December 98. I started with

64
00:02:55,260 --> 00:02:59,580
nemco

65
00:02:56,700 --> 00:03:02,940
August the following year

66
00:02:59,580 --> 00:03:05,519
AMA has around about 1200 staff from six

67
00:03:02,940 --> 00:03:07,620
offices around five states

68
00:03:05,519 --> 00:03:10,860
we're broadly divided into teams that

69
00:03:07,620 --> 00:03:13,140
support the markets mostly c-sharp Folk

70
00:03:10,860 --> 00:03:16,260
teams that support the grids or the

71
00:03:13,140 --> 00:03:17,879
physics mostly Python and our folk and

72
00:03:16,260 --> 00:03:21,300
other supporting roles that you'd expect

73
00:03:17,879 --> 00:03:23,340
of an Enterprise our size

74
00:03:21,300 --> 00:03:26,099
we support the nem with two identical

75
00:03:23,340 --> 00:03:28,500
control rooms in two states operating in

76
00:03:26,099 --> 00:03:29,640
parallel each consuming full control of

77
00:03:28,500 --> 00:03:32,580
the grid

78
00:03:29,640 --> 00:03:34,500
the whim or Western Australian Market is

79
00:03:32,580 --> 00:03:38,940
operated out of a control room in Perth

80
00:03:34,500 --> 00:03:41,459
while gas is controlled from Melbourne

81
00:03:38,940 --> 00:03:43,560
the nem delivers around about 200

82
00:03:41,459 --> 00:03:45,420
terawatt hours of electricity to

83
00:03:43,560 --> 00:03:48,299
customers every year

84
00:03:45,420 --> 00:03:50,099
with around about 11.5 billion dollars

85
00:03:48,299 --> 00:03:52,680
being traded

86
00:03:50,099 --> 00:03:54,720
now I have no idea what 200 terawatt

87
00:03:52,680 --> 00:03:57,599
hours of electricity looks like or what

88
00:03:54,720 --> 00:03:58,980
11.5 billion looks like so I did a

89
00:03:57,599 --> 00:04:01,860
comparison

90
00:03:58,980 --> 00:04:03,540
Australia's defense budget is about 52

91
00:04:01,860 --> 00:04:05,819
billion per annum so our trade in

92
00:04:03,540 --> 00:04:07,860
wholesale electricity is about a quarter

93
00:04:05,819 --> 00:04:09,780
of what we spend on defense

94
00:04:07,860 --> 00:04:12,379
I don't know if that helps get the scale

95
00:04:09,780 --> 00:04:12,379
in your head

96
00:04:12,659 --> 00:04:17,880
the day the transition began

97
00:04:15,599 --> 00:04:22,019
Australia's transitioned to renewals

98
00:04:17,880 --> 00:04:24,660
began on the 17th of February 2005 when

99
00:04:22,019 --> 00:04:27,120
our first grid scale wind farm became

100
00:04:24,660 --> 00:04:28,979
visible to Australia to aimo's control

101
00:04:27,120 --> 00:04:31,520
systems

102
00:04:28,979 --> 00:04:34,259
wind farms have very different physical

103
00:04:31,520 --> 00:04:38,540
characteristics than those rotating

104
00:04:34,259 --> 00:04:38,540
masses I spoke about earlier

105
00:04:39,240 --> 00:04:44,940
prior to February 2005 the grid looked

106
00:04:42,780 --> 00:04:48,120
like this

107
00:04:44,940 --> 00:04:51,060
black circles indicate the location of

108
00:04:48,120 --> 00:04:53,940
black coal powered generators

109
00:04:51,060 --> 00:04:57,360
brown circles indicate the location of

110
00:04:53,940 --> 00:05:01,380
brown Coal Fired generators blue circles

111
00:04:57,360 --> 00:05:02,639
hydro and red other gas diesel and

112
00:05:01,380 --> 00:05:05,220
biofuel

113
00:05:02,639 --> 00:05:08,100
the diameter of the circle indicates the

114
00:05:05,220 --> 00:05:13,080
installed capacity of the generator

115
00:05:08,100 --> 00:05:16,759
prior to Feb 2005 we had 305 units

116
00:05:13,080 --> 00:05:20,940
to interact with and around about 42

117
00:05:16,759 --> 00:05:23,220
gigawatts of installed capacity

118
00:05:20,940 --> 00:05:24,539
roll forward to today and the picture is

119
00:05:23,220 --> 00:05:26,880
very different

120
00:05:24,539 --> 00:05:30,000
you almost can't see the coast on this

121
00:05:26,880 --> 00:05:33,979
map because of all the colored dots

122
00:05:30,000 --> 00:05:37,380
yellow indicates solar either grid scale

123
00:05:33,979 --> 00:05:40,620
or what we call Dr or distributed energy

124
00:05:37,380 --> 00:05:44,160
resources behind the meter on the

125
00:05:40,620 --> 00:05:46,620
residents and small business rooftops

126
00:05:44,160 --> 00:05:50,220
green indicates wind

127
00:05:46,620 --> 00:05:53,600
blue and black and brown are coal and

128
00:05:50,220 --> 00:05:53,600
hydro as in the previous

129
00:05:53,759 --> 00:05:57,900
screen

130
00:05:55,160 --> 00:06:01,320
an interesting thing to note is biofuel

131
00:05:57,900 --> 00:06:03,360
down the bottom three units 91 megawatts

132
00:06:01,320 --> 00:06:06,780
we have three units in Northern

133
00:06:03,360 --> 00:06:08,759
Queensland generating using waste

134
00:06:06,780 --> 00:06:10,139
product from sugarcane production from

135
00:06:08,759 --> 00:06:12,360
sugar production

136
00:06:10,139 --> 00:06:15,320
it's exactly the same as it was back in

137
00:06:12,360 --> 00:06:18,240
2005.

138
00:06:15,320 --> 00:06:20,280
this changing of the grid Generation

139
00:06:18,240 --> 00:06:22,440
profile presents us with many

140
00:06:20,280 --> 00:06:24,240
engineering challenges

141
00:06:22,440 --> 00:06:26,039
first of all the loss of what we call

142
00:06:24,240 --> 00:06:28,919
synchronous machines

143
00:06:26,039 --> 00:06:31,319
those machines with rotating mass as we

144
00:06:28,919 --> 00:06:35,039
transition to inverter-based generation

145
00:06:31,319 --> 00:06:38,880
or ibrs this makes it much harder to

146
00:06:35,039 --> 00:06:40,919
control that perfectly formed sine wave

147
00:06:38,880 --> 00:06:44,039
we have the transition from centralized

148
00:06:40,919 --> 00:06:47,460
generation to distributed generation or

149
00:06:44,039 --> 00:06:49,680
der distributed energy resources this

150
00:06:47,460 --> 00:06:52,460
increases the number of units we must

151
00:06:49,680 --> 00:06:52,460
interact with

152
00:06:52,620 --> 00:06:58,440
we have the transition from dispatchable

153
00:06:54,840 --> 00:07:01,979
energy energy created by a machine that

154
00:06:58,440 --> 00:07:05,280
we have control over to variable energy

155
00:07:01,979 --> 00:07:07,020
or variable renewable energy we tend to

156
00:07:05,280 --> 00:07:09,479
like to take everything that the weather

157
00:07:07,020 --> 00:07:13,139
can give us

158
00:07:09,479 --> 00:07:15,720
as Dr penetration grows we must plan for

159
00:07:13,139 --> 00:07:19,139
two-way flows and the grid was not

160
00:07:15,720 --> 00:07:21,780
designed to manage two-way flows

161
00:07:19,139 --> 00:07:24,599
we are planning for a two to five-fold

162
00:07:21,780 --> 00:07:26,880
increase in demand for electricity to

163
00:07:24,599 --> 00:07:29,220
support the decarbonization of other

164
00:07:26,880 --> 00:07:32,599
sectors of the economy

165
00:07:29,220 --> 00:07:38,840
and we must manage between 5 and 25

166
00:07:32,599 --> 00:07:38,840
and 20 times increase in data volumes

167
00:07:40,380 --> 00:07:46,919
narration mix right

168
00:07:43,380 --> 00:07:49,740
this table shows some fuel sources or

169
00:07:46,919 --> 00:07:52,199
some energy sources we have access to on

170
00:07:49,740 --> 00:07:54,240
the left hand column

171
00:07:52,199 --> 00:07:56,039
and it shows some characteristics that

172
00:07:54,240 --> 00:07:59,460
are generated with those energy sources

173
00:07:56,039 --> 00:08:02,520
bring to the grid across the top

174
00:07:59,460 --> 00:08:05,099
coal for example has a very slow startup

175
00:08:02,520 --> 00:08:07,199
time we call it ramp rate

176
00:08:05,099 --> 00:08:08,940
the coal is very high car has a high

177
00:08:07,199 --> 00:08:11,400
carbon intensity

178
00:08:08,940 --> 00:08:13,319
it is dispatchable we can however send a

179
00:08:11,400 --> 00:08:14,580
signal to it and say we want more or we

180
00:08:13,319 --> 00:08:17,460
want less

181
00:08:14,580 --> 00:08:20,280
it brings rotating Mass to the grid

182
00:08:17,460 --> 00:08:23,520
which gives the grid stability

183
00:08:20,280 --> 00:08:26,580
geographically it tends to be located

184
00:08:23,520 --> 00:08:29,699
far away from the load centers

185
00:08:26,580 --> 00:08:31,800
gas has a fast start time it's carbon

186
00:08:29,699 --> 00:08:35,099
intensive intensive but not as high as

187
00:08:31,800 --> 00:08:37,320
coal it's dispatchable it has rotating

188
00:08:35,099 --> 00:08:40,080
mass and it can be close to the load

189
00:08:37,320 --> 00:08:42,539
centers meaning less transmission

190
00:08:40,080 --> 00:08:45,740
Hydro fast start time low carbon

191
00:08:42,539 --> 00:08:50,540
intensity dispatchable rotating Mass

192
00:08:45,740 --> 00:08:50,540
generally a long way from load centers

193
00:08:50,580 --> 00:08:54,660
PV most start time we take everything we

194
00:08:52,920 --> 00:08:56,880
can get low carbon intensity

195
00:08:54,660 --> 00:08:59,220
dispatchable no like I said we take

196
00:08:56,880 --> 00:09:01,260
everything we can get no rotating mass

197
00:08:59,220 --> 00:09:02,040
and it can be embedded in the load

198
00:09:01,260 --> 00:09:04,800
Center

199
00:09:02,040 --> 00:09:06,839
on your rooftop or a long way from a

200
00:09:04,800 --> 00:09:09,060
load Center we have a lot of PV

201
00:09:06,839 --> 00:09:11,519
resources in Northwest and North Western

202
00:09:09,060 --> 00:09:15,660
Victoria for example

203
00:09:11,519 --> 00:09:18,779
wind low carbon intensity variable

204
00:09:15,660 --> 00:09:20,519
no rotating Mass long way from the load

205
00:09:18,779 --> 00:09:23,459
centers

206
00:09:20,519 --> 00:09:24,600
trees very fast start time low carbon

207
00:09:23,459 --> 00:09:27,120
intensity

208
00:09:24,600 --> 00:09:29,339
they're dispatchable but they

209
00:09:27,120 --> 00:09:30,959
they're a shallow storage they don't

210
00:09:29,339 --> 00:09:33,300
hold a huge amount of energy so they're

211
00:09:30,959 --> 00:09:36,720
good for injecting power to solve

212
00:09:33,300 --> 00:09:39,360
short-term discrepancy short-term losses

213
00:09:36,720 --> 00:09:42,860
they have no rotating mass and they can

214
00:09:39,360 --> 00:09:42,860
be embedded close to the load Center

215
00:09:42,899 --> 00:09:47,580
finally other engineering challenges we

216
00:09:45,120 --> 00:09:49,440
Face are cyber risks as nation states

217
00:09:47,580 --> 00:09:53,580
realize that they can attack critical

218
00:09:49,440 --> 00:09:57,060
infrastructure as part of their toolkit

219
00:09:53,580 --> 00:09:59,420
and like all of us finding Engineers is

220
00:09:57,060 --> 00:09:59,420
a challenge

221
00:09:59,700 --> 00:10:06,660
what's driving the growth of Renewables

222
00:10:03,480 --> 00:10:08,339
first of all aging and retiring Legacy

223
00:10:06,660 --> 00:10:10,440
assets

224
00:10:08,339 --> 00:10:13,080
the kit that we built 30 years ago is

225
00:10:10,440 --> 00:10:16,380
coming to the end of life

226
00:10:13,080 --> 00:10:18,420
secondly firmed or storage-backed

227
00:10:16,380 --> 00:10:22,260
Renewables are the lowest cost

228
00:10:18,420 --> 00:10:25,200
generation we can build today

229
00:10:22,260 --> 00:10:27,360
to decarbonize our economy we must scale

230
00:10:25,200 --> 00:10:30,180
electricity production by a factor of

231
00:10:27,360 --> 00:10:32,959
two to replace oil and gas that is

232
00:10:30,180 --> 00:10:36,300
currently used for space heating cooking

233
00:10:32,959 --> 00:10:38,160
low temperature industrial heat and

234
00:10:36,300 --> 00:10:40,860
transport

235
00:10:38,160 --> 00:10:43,620
and finally diversity in energy source

236
00:10:40,860 --> 00:10:46,579
and geography leads to a more resilient

237
00:10:43,620 --> 00:10:46,579
Network

238
00:10:46,920 --> 00:10:51,899
what's our record for variable renewable

239
00:10:49,620 --> 00:10:54,800
energy

240
00:10:51,899 --> 00:11:00,240
for half an hour on the 28th of October

241
00:10:54,800 --> 00:11:02,040
2022 renewable supplied 68.7 percent of

242
00:11:00,240 --> 00:11:04,680
the nem generation

243
00:11:02,040 --> 00:11:07,860
this graph shows our personal best to

244
00:11:04,680 --> 00:11:11,000
date and is available on our public

245
00:11:07,860 --> 00:11:11,000
facing website

246
00:11:12,360 --> 00:11:17,579
what's the future for variable renewable

247
00:11:15,240 --> 00:11:21,480
energy

248
00:11:17,579 --> 00:11:25,140
the solid line second from the top

249
00:11:21,480 --> 00:11:27,540
shows the maximum penetration of

250
00:11:25,140 --> 00:11:30,600
variable Renewables Against Time by

251
00:11:27,540 --> 00:11:33,959
quarter you can see slight up and down

252
00:11:30,600 --> 00:11:36,360
of the curve as it tracks the seasons of

253
00:11:33,959 --> 00:11:39,720
course we get less variable in Winter

254
00:11:36,360 --> 00:11:43,620
and we get more in the summer

255
00:11:39,720 --> 00:11:46,140
but the graph is trending slowly upwards

256
00:11:43,620 --> 00:11:48,480
if you draw a line of best fit through

257
00:11:46,140 --> 00:11:52,620
the graph and extrapolate to the point

258
00:11:48,480 --> 00:11:52,620
where it crosses the 100 mark

259
00:11:52,800 --> 00:11:58,260
you'll see it intersects at some time in

260
00:11:54,959 --> 00:12:00,720
2005 2025.

261
00:11:58,260 --> 00:12:03,300
so our goal is to have the control

262
00:12:00,720 --> 00:12:05,399
mechanisms in place to manage the Grid

263
00:12:03,300 --> 00:12:08,220
at a hundred percent

264
00:12:05,399 --> 00:12:10,800
variable renewable energy by sometime in

265
00:12:08,220 --> 00:12:12,060
2025.

266
00:12:10,800 --> 00:12:14,820
all right

267
00:12:12,060 --> 00:12:18,060
we model the grid with a variety of

268
00:12:14,820 --> 00:12:20,100
Technologies but also with pipe

269
00:12:18,060 --> 00:12:21,720
the image on the right

270
00:12:20,100 --> 00:12:23,040
shows

271
00:12:21,720 --> 00:12:25,740
a model

272
00:12:23,040 --> 00:12:29,100
of the Wind Farm we looked at in the

273
00:12:25,740 --> 00:12:31,019
picture at the start of the talk

274
00:12:29,100 --> 00:12:34,019
the circle with the crossroad at the

275
00:12:31,019 --> 00:12:36,720
bottom is used to represent a generator

276
00:12:34,019 --> 00:12:37,680
many wind turbines are collapsed down to

277
00:12:36,720 --> 00:12:39,839
one

278
00:12:37,680 --> 00:12:42,600
point in this model

279
00:12:39,839 --> 00:12:45,180
the horizontal blue bar represents what

280
00:12:42,600 --> 00:12:46,920
we call a bus my mental picture of what

281
00:12:45,180 --> 00:12:49,019
a bus is is the lump of copper where

282
00:12:46,920 --> 00:12:52,980
things are bolted to

283
00:12:49,019 --> 00:12:54,959
the solid square is a breaker or switch

284
00:12:52,980 --> 00:12:57,560
and the overlapping circles are a

285
00:12:54,959 --> 00:12:57,560
Transformer

286
00:12:57,959 --> 00:13:02,100
we model the design of the grid with the

287
00:13:00,480 --> 00:13:04,800
help of python

288
00:13:02,100 --> 00:13:06,779
we model the flows within the grid with

289
00:13:04,800 --> 00:13:09,660
the help of python

290
00:13:06,779 --> 00:13:12,360
we modeled distributed energy resources

291
00:13:09,660 --> 00:13:14,880
or rooftop solar with the help of python

292
00:13:12,360 --> 00:13:16,680
and we move data around with the help of

293
00:13:14,880 --> 00:13:19,500
python

294
00:13:16,680 --> 00:13:21,779
why did we choose python

295
00:13:19,500 --> 00:13:23,519
the market people those interested in

296
00:13:21,779 --> 00:13:26,160
the dollars of the grid are generally

297
00:13:23,519 --> 00:13:28,500
Java or c-sharp Folk

298
00:13:26,160 --> 00:13:30,540
the grid people those of us interested

299
00:13:28,500 --> 00:13:34,079
in the physics of the grid tend to be

300
00:13:30,540 --> 00:13:36,000
python or Fortran or R people

301
00:13:34,079 --> 00:13:38,339
python is an elegant beautiful

302
00:13:36,000 --> 00:13:40,680
accessible scalable language but

303
00:13:38,339 --> 00:13:43,860
everybody in this room knows that

304
00:13:40,680 --> 00:13:46,200
python has advanced tooling formatters

305
00:13:43,860 --> 00:13:48,120
linters Pi test coverage and maths

306
00:13:46,200 --> 00:13:50,160
libraries but everybody in this room

307
00:13:48,120 --> 00:13:52,440
knows that too

308
00:13:50,160 --> 00:13:54,540
we chose python as we scale our

309
00:13:52,440 --> 00:13:57,360
engineering engineering Centric

310
00:13:54,540 --> 00:14:00,540
development efforts because of access to

311
00:13:57,360 --> 00:14:02,940
an in-house Workforce

312
00:14:00,540 --> 00:14:05,700
when a graduate power system engineer

313
00:14:02,940 --> 00:14:07,560
joins us from University they come

314
00:14:05,700 --> 00:14:09,779
trained with python

315
00:14:07,560 --> 00:14:12,420
we have a formal training program that

316
00:14:09,779 --> 00:14:14,820
lists lifts their desktop coding skills

317
00:14:12,420 --> 00:14:16,860
to what is necessary to write High

318
00:14:14,820 --> 00:14:21,740
availability software that will run

319
00:14:16,860 --> 00:14:21,740
unsuperide unsupervised in the cool room

320
00:14:22,620 --> 00:14:29,160
what else do we use Python for

321
00:14:25,800 --> 00:14:31,200
we use it to manipulate Network models

322
00:14:29,160 --> 00:14:33,120
publish apis

323
00:14:31,200 --> 00:14:34,380
we use it for personal workflow

324
00:14:33,120 --> 00:14:37,320
automation

325
00:14:34,380 --> 00:14:39,540
for personal data wrangling and for data

326
00:14:37,320 --> 00:14:41,720
science as Josh will talk about in a

327
00:14:39,540 --> 00:14:41,720
moment

328
00:14:43,980 --> 00:14:50,220
flow look like within our organization

329
00:14:47,160 --> 00:14:52,320
we receive data from many sources

330
00:14:50,220 --> 00:14:55,139
we have marketing meters that are high

331
00:14:52,320 --> 00:14:56,100
accuracy calibrated devices that are

332
00:14:55,139 --> 00:14:58,320
used

333
00:14:56,100 --> 00:15:01,019
to control the market

334
00:14:58,320 --> 00:15:04,740
we have scada meters that are of lower

335
00:15:01,019 --> 00:15:08,160
accuracy that measure volts and amps

336
00:15:04,740 --> 00:15:09,660
we have devices called pmus or phasor

337
00:15:08,160 --> 00:15:12,300
measurement units

338
00:15:09,660 --> 00:15:14,160
machines or monitors that follow the

339
00:15:12,300 --> 00:15:15,480
precise shape of that sine wave I spoke

340
00:15:14,160 --> 00:15:17,880
about

341
00:15:15,480 --> 00:15:20,100
we have apis where Market participants

342
00:15:17,880 --> 00:15:21,899
can submit data

343
00:15:20,100 --> 00:15:25,860
we receive

344
00:15:21,899 --> 00:15:27,720
der or rooftop solar information from

345
00:15:25,860 --> 00:15:29,760
third-party data providers and we

346
00:15:27,720 --> 00:15:31,920
receive weather for information from

347
00:15:29,760 --> 00:15:34,079
third-party data providers

348
00:15:31,920 --> 00:15:37,079
data is ingested the combination of

349
00:15:34,079 --> 00:15:40,320
c-sharp python and Java we store our

350
00:15:37,079 --> 00:15:42,779
data in Oracle SQL server and post SQL

351
00:15:40,320 --> 00:15:44,639
and we perform analytics with c-sharp

352
00:15:42,779 --> 00:15:46,800
python and r

353
00:15:44,639 --> 00:15:50,699
we use the outputs to make operational

354
00:15:46,800 --> 00:15:54,959
decisions and we use the outputs to make

355
00:15:50,699 --> 00:15:58,260
to perform desktop analysis

356
00:15:54,959 --> 00:16:00,959
what does a visualization look like

357
00:15:58,260 --> 00:16:04,620
this graph is taken from an in-house or

358
00:16:00,959 --> 00:16:06,899
a power system engineer developed tool

359
00:16:04,620 --> 00:16:09,300
called plotting buddy

360
00:16:06,899 --> 00:16:12,060
the blue line is taken from one of our

361
00:16:09,300 --> 00:16:14,040
high-speed monitoring devices and shows

362
00:16:12,060 --> 00:16:15,180
the response of the grid to a

363
00:16:14,040 --> 00:16:18,060
disturbance

364
00:16:15,180 --> 00:16:20,699
a disturbance might be a grounding of a

365
00:16:18,060 --> 00:16:24,000
line or a lightning strike

366
00:16:20,699 --> 00:16:26,760
the green line shows the response of our

367
00:16:24,000 --> 00:16:30,420
models to that disturbance so what we're

368
00:16:26,760 --> 00:16:33,300
doing here is studying the accuracy of

369
00:16:30,420 --> 00:16:34,980
the models against reality and this

370
00:16:33,300 --> 00:16:38,240
model is looking pretty good but it

371
00:16:34,980 --> 00:16:38,240
could do with a little bit of tweaking

372
00:16:39,139 --> 00:16:44,100
this graph shows the trace coming out of

373
00:16:42,180 --> 00:16:46,320
Northwest and Victoria

374
00:16:44,100 --> 00:16:49,320
Northwestern Victoria has a very high

375
00:16:46,320 --> 00:16:51,720
penetration of grid scale solar

376
00:16:49,320 --> 00:16:55,380
grid scale solar does not bring that

377
00:16:51,720 --> 00:16:58,019
rotating Mass I spoke about so it's

378
00:16:55,380 --> 00:17:00,480
vulnerable to disturbance

379
00:16:58,019 --> 00:17:02,639
we have a python system that is watching

380
00:17:00,480 --> 00:17:04,260
the flow of data as it's ingested into

381
00:17:02,639 --> 00:17:07,500
aimo

382
00:17:04,260 --> 00:17:10,860
looking for oscillations like this

383
00:17:07,500 --> 00:17:13,559
the tech is numpy and scipy performing a

384
00:17:10,860 --> 00:17:16,679
fast for our transformation looking for

385
00:17:13,559 --> 00:17:18,660
changes to the shape of the sine wave

386
00:17:16,679 --> 00:17:21,260
foreign

387
00:17:18,660 --> 00:17:23,939
before I hand over to Josh

388
00:17:21,260 --> 00:17:28,140
aimu's first line of python was written

389
00:17:23,939 --> 00:17:30,600
in 2010 to automate power flow studies

390
00:17:28,140 --> 00:17:31,940
we started our python Journey at scale

391
00:17:30,600 --> 00:17:34,799
in August

392
00:17:31,940 --> 00:17:36,960
2018. soon after some of us attended

393
00:17:34,799 --> 00:17:39,539
pycon in Sydney

394
00:17:36,960 --> 00:17:42,120
our first task was to rewrite complex

395
00:17:39,539 --> 00:17:45,120
algorithm from Pascal

396
00:17:42,120 --> 00:17:48,120
seventy thousand lines of Pascal became

397
00:17:45,120 --> 00:17:49,919
14 000 lines of python and it took four

398
00:17:48,120 --> 00:17:52,020
of us four months

399
00:17:49,919 --> 00:17:54,600
we worked out that it cost around about

400
00:17:52,020 --> 00:17:55,919
fourteen dollars per line of Pascal to

401
00:17:54,600 --> 00:17:58,620
rewrite

402
00:17:55,919 --> 00:18:03,059
we now have 71 000 lines of production

403
00:17:58,620 --> 00:18:05,880
quality python at Amo with 52 000 lines

404
00:18:03,059 --> 00:18:08,840
of automated unit tests

405
00:18:05,880 --> 00:18:08,840
over to you Josh

406
00:18:17,039 --> 00:18:22,260
thank you very much Peter so as Peter

407
00:18:20,160 --> 00:18:24,000
touched on I work for the operational

408
00:18:22,260 --> 00:18:25,860
forecasting division

409
00:18:24,000 --> 00:18:28,020
and we'll get into what that actually

410
00:18:25,860 --> 00:18:29,940
means in the next couple of slides but

411
00:18:28,020 --> 00:18:31,559
before we do that just another brief

412
00:18:29,940 --> 00:18:34,260
overview of the name by the numbers so

413
00:18:31,559 --> 00:18:36,299
we've got the eastern states in pink

414
00:18:34,260 --> 00:18:38,760
um and interesting facts to notice there

415
00:18:36,299 --> 00:18:40,860
is that the installed utility scale

416
00:18:38,760 --> 00:18:43,860
solar is now at nearly nine gigawatts

417
00:18:40,860 --> 00:18:47,039
where's the rooftop solar on families

418
00:18:43,860 --> 00:18:49,020
homes is at nearly 17 gigawatts which is

419
00:18:47,039 --> 00:18:51,299
nearly double that and approximately

420
00:18:49,020 --> 00:18:53,640
three million homes in Australia now

421
00:18:51,299 --> 00:18:56,299
have rooftop PV systems installed which

422
00:18:53,640 --> 00:18:59,220
is some of the highest in the world

423
00:18:56,299 --> 00:19:00,600
another brief Point worth mentioning is

424
00:18:59,220 --> 00:19:02,340
that when we build our machine learning

425
00:19:00,600 --> 00:19:05,100
models we can't simply have a single

426
00:19:02,340 --> 00:19:07,200
model for all states each region has its

427
00:19:05,100 --> 00:19:08,820
own unique physical and weather-based

428
00:19:07,200 --> 00:19:10,919
characteristics which we need to take

429
00:19:08,820 --> 00:19:13,679
account for and again just calling out

430
00:19:10,919 --> 00:19:16,140
on that red box the incredibly High

431
00:19:13,679 --> 00:19:18,539
penetration of dpv in South Australia

432
00:19:16,140 --> 00:19:21,600
where we are today reaching up to around

433
00:19:18,539 --> 00:19:24,120
93 at certain intervals which is a lot

434
00:19:21,600 --> 00:19:27,780
higher than the 68 average for the

435
00:19:24,120 --> 00:19:29,220
entire Nim that Peter mentioned earlier

436
00:19:27,780 --> 00:19:32,700
so what do we do in operational

437
00:19:29,220 --> 00:19:35,340
forecasting well we produce forecasts to

438
00:19:32,700 --> 00:19:38,039
put it bluntly but we forecast the load

439
00:19:35,340 --> 00:19:40,860
or demand and then we also forecast

440
00:19:38,039 --> 00:19:43,020
supply for variable renewable generation

441
00:19:40,860 --> 00:19:45,600
such as the utility scale solar and wind

442
00:19:43,020 --> 00:19:49,200
farms as well as all the millions of

443
00:19:45,600 --> 00:19:51,539
solar PV panels on people's roofs we do

444
00:19:49,200 --> 00:19:53,940
this for five minutes ahead up to one

445
00:19:51,539 --> 00:19:56,100
week ahead every five minutes and that

446
00:19:53,940 --> 00:19:57,780
adds up pretty quickly to us producing

447
00:19:56,100 --> 00:20:01,080
approximately three million point

448
00:19:57,780 --> 00:20:02,940
forecasts every single day

449
00:20:01,080 --> 00:20:05,100
if that doesn't keep us busy enough we

450
00:20:02,940 --> 00:20:07,620
also have a number of other roles uh we

451
00:20:05,100 --> 00:20:09,240
work very closely with the control rooms

452
00:20:07,620 --> 00:20:11,220
that Peter mentioned and providing

453
00:20:09,240 --> 00:20:12,840
situational awareness and live weather

454
00:20:11,220 --> 00:20:15,059
briefings to them

455
00:20:12,840 --> 00:20:17,100
we also help facilitate the onboarding

456
00:20:15,059 --> 00:20:19,200
of new wind and solar Farms as we

457
00:20:17,100 --> 00:20:21,120
provide forecasts for those and

458
00:20:19,200 --> 00:20:22,980
facilitate them providing their own self

459
00:20:21,120 --> 00:20:24,480
forecasts where they are able to out

460
00:20:22,980 --> 00:20:27,000
compete with us

461
00:20:24,480 --> 00:20:29,220
apart from the documentation which is

462
00:20:27,000 --> 00:20:31,140
immense and we also have a fair amount

463
00:20:29,220 --> 00:20:34,140
of international engagement with other

464
00:20:31,140 --> 00:20:36,059
independent system operators or isos as

465
00:20:34,140 --> 00:20:38,580
well as other forecasting experts and

466
00:20:36,059 --> 00:20:39,780
Commercial entities we're in quite a

467
00:20:38,580 --> 00:20:42,059
unique position in Australia

468
00:20:39,780 --> 00:20:44,600
particularly in South Australia with our

469
00:20:42,059 --> 00:20:47,400
world leading penetration of rooftop PV

470
00:20:44,600 --> 00:20:49,620
that the rest of the world looks to us

471
00:20:47,400 --> 00:20:51,299
on how we're solving these problems as a

472
00:20:49,620 --> 00:20:53,820
lead indicator on what they will have to

473
00:20:51,299 --> 00:20:57,539
deal with in the future so we really are

474
00:20:53,820 --> 00:20:59,820
working on the coal phase so to speak

475
00:20:57,539 --> 00:21:02,280
this is just a very brief summary of

476
00:20:59,820 --> 00:21:04,020
some of our forecasting processes and

477
00:21:02,280 --> 00:21:06,539
some of those data from the data flows

478
00:21:04,020 --> 00:21:08,400
that Peter touched on earlier but what

479
00:21:06,539 --> 00:21:10,919
this is really highlighting is we've

480
00:21:08,400 --> 00:21:13,559
taken a vast array of different types of

481
00:21:10,919 --> 00:21:17,160
data we train our models on historic

482
00:21:13,559 --> 00:21:18,780
data we have real-time data and then the

483
00:21:17,160 --> 00:21:21,059
weather now that we couple Our

484
00:21:18,780 --> 00:21:22,980
Generation to chaotic weather systems as

485
00:21:21,059 --> 00:21:25,620
a fuel source you can imagine that where

486
00:21:22,980 --> 00:21:27,780
the inputs are quite key to the accuracy

487
00:21:25,620 --> 00:21:29,100
Island models and we take those in from

488
00:21:27,780 --> 00:21:32,159
a number of vendors in a number of

489
00:21:29,100 --> 00:21:34,559
formats both 0.4 costs numerical gridded

490
00:21:32,159 --> 00:21:37,200
forecast satellite imagery and we also

491
00:21:34,559 --> 00:21:40,320
have a direct sample of a statistically

492
00:21:37,200 --> 00:21:42,299
significant number of rooftop PV systems

493
00:21:40,320 --> 00:21:44,460
around the country which we use to

494
00:21:42,299 --> 00:21:47,720
estimate the overall three million

495
00:21:44,460 --> 00:21:47,720
systems on people's roofs

496
00:21:47,940 --> 00:21:52,679
this slide touches on something else

497
00:21:50,460 --> 00:21:54,539
which is worth drawing attention to when

498
00:21:52,679 --> 00:21:56,700
we publish forecasts to the market we

499
00:21:54,539 --> 00:22:00,659
publish a single deterministic forecast

500
00:21:56,700 --> 00:22:02,159
as well as a Poe 90 and a poe10 but

501
00:22:00,659 --> 00:22:03,659
behind the scenes is actually a lot more

502
00:22:02,159 --> 00:22:05,880
going on

503
00:22:03,659 --> 00:22:08,760
um we have a large number of machine

504
00:22:05,880 --> 00:22:10,340
learning models calibrated and tuned

505
00:22:08,760 --> 00:22:12,600
slightly differently or running

506
00:22:10,340 --> 00:22:14,220
simultaneously and we then have other

507
00:22:12,600 --> 00:22:16,320
models which Ensemble those in a

508
00:22:14,220 --> 00:22:18,539
consensus approach to get the most

509
00:22:16,320 --> 00:22:20,520
accurate forecast possible this

510
00:22:18,539 --> 00:22:21,900
technique has been used with great

511
00:22:20,520 --> 00:22:24,780
success for many years in the weather

512
00:22:21,900 --> 00:22:27,299
industry and is now adopted in other

513
00:22:24,780 --> 00:22:29,419
Industries like ours it also has the

514
00:22:27,299 --> 00:22:32,220
added benefits of producing

515
00:22:29,419 --> 00:22:34,020
risk analysis and probabilistic

516
00:22:32,220 --> 00:22:36,960
forecasting as well which we use for our

517
00:22:34,020 --> 00:22:39,480
risk assessments and to assess scenarios

518
00:22:36,960 --> 00:22:41,640
in the control room we also produce

519
00:22:39,480 --> 00:22:43,020
other custom tooling like those to

520
00:22:41,640 --> 00:22:45,000
forecast higher end and temperature

521
00:22:43,020 --> 00:22:47,340
cutouts of wind farms again for

522
00:22:45,000 --> 00:22:48,659
situational awareness and you may

523
00:22:47,340 --> 00:22:50,340
recognize some of the open source

524
00:22:48,659 --> 00:22:53,700
plotting libraries behind you in some of

525
00:22:50,340 --> 00:22:55,320
those charts and dashboards

526
00:22:53,700 --> 00:22:57,179
so there's some important trends that

527
00:22:55,320 --> 00:23:00,120
are affecting us one is the growth in

528
00:22:57,179 --> 00:23:02,240
VRE and dpv which Peter mentioned but

529
00:23:00,120 --> 00:23:04,740
there's also the change in shape of the

530
00:23:02,240 --> 00:23:06,360
Generation profile over the day which

531
00:23:04,740 --> 00:23:08,760
we'll touch on shortly

532
00:23:06,360 --> 00:23:11,280
but probably the most important is the

533
00:23:08,760 --> 00:23:13,200
increased uncertainty we face as we

534
00:23:11,280 --> 00:23:16,039
couple our fuel source to chaotic

535
00:23:13,200 --> 00:23:16,039
weather systems

536
00:23:16,440 --> 00:23:21,900
slide briefly highlights the growth in

537
00:23:19,440 --> 00:23:25,260
utility scale wind and solar assets over

538
00:23:21,900 --> 00:23:27,000
the last or since 2015 shall I say and

539
00:23:25,260 --> 00:23:29,520
from this picture the key takeaway is

540
00:23:27,000 --> 00:23:31,860
that from 2015 to 17 it was

541
00:23:29,520 --> 00:23:34,559
predominantly large-scale wind farms and

542
00:23:31,860 --> 00:23:36,840
not much change but from 2018 onwards

543
00:23:34,559 --> 00:23:39,419
there was really an explosive growth of

544
00:23:36,840 --> 00:23:41,700
new utility scale Soda farms and that

545
00:23:39,419 --> 00:23:44,640
continued to outpace wind and at the end

546
00:23:41,700 --> 00:23:46,200
of 2022 there are now more utility scale

547
00:23:44,640 --> 00:23:48,900
solar farms in Australia than they are

548
00:23:46,200 --> 00:23:51,480
wind farms and a total of 150 utility

549
00:23:48,900 --> 00:23:54,000
scale or greater than 150 utility scale

550
00:23:51,480 --> 00:23:57,500
assets out in the field today and we

551
00:23:54,000 --> 00:23:57,500
only foresee that to keep on growing

552
00:23:57,600 --> 00:24:03,900
so how does this look during a typical

553
00:24:00,240 --> 00:24:06,360
day so this chart shows the generation

554
00:24:03,900 --> 00:24:08,340
stack in South Australia again it

555
00:24:06,360 --> 00:24:09,260
provides a good example and it's close

556
00:24:08,340 --> 00:24:11,820
to home

557
00:24:09,260 --> 00:24:14,039
so this is looking at a 24-hour period

558
00:24:11,820 --> 00:24:16,080
from Midnight to midnight and you can

559
00:24:14,039 --> 00:24:18,539
see there that overnight obviously the

560
00:24:16,080 --> 00:24:20,940
sun is not shining so the majority of

561
00:24:18,539 --> 00:24:23,039
demand is met from conventional plants

562
00:24:20,940 --> 00:24:24,480
such as gas in South Australia as well

563
00:24:23,039 --> 00:24:27,000
as imports from other states like

564
00:24:24,480 --> 00:24:28,919
Victoria and a component of wind which

565
00:24:27,000 --> 00:24:31,860
does fortunately below during the night

566
00:24:28,919 --> 00:24:34,620
time then as the sun rises you can see

567
00:24:31,860 --> 00:24:37,260
that solar really takes a large chunk

568
00:24:34,620 --> 00:24:39,000
out of that and displaces a lot of the

569
00:24:37,260 --> 00:24:42,299
conventional fuel

570
00:24:39,000 --> 00:24:44,179
however as Peter mentioned solar does

571
00:24:42,299 --> 00:24:46,860
not have you know grid stability

572
00:24:44,179 --> 00:24:49,740
provisions and hence we often have to

573
00:24:46,860 --> 00:24:51,299
direct on conventional gas plant to

574
00:24:49,740 --> 00:24:53,100
maintain system Security in South

575
00:24:51,299 --> 00:24:55,620
Australia so that's where you can see it

576
00:24:53,100 --> 00:24:57,960
dipping below the the zero line and then

577
00:24:55,620 --> 00:25:00,720
we'll be exporting access to Victoria as

578
00:24:57,960 --> 00:25:02,220
well and then challenging periods come

579
00:25:00,720 --> 00:25:04,200
when this when the Sun starts to ramp

580
00:25:02,220 --> 00:25:06,600
down the conventional plant then needs

581
00:25:04,200 --> 00:25:10,679
to be ramp able to ramp up quickly again

582
00:25:06,600 --> 00:25:12,960
to meet demand for the evening Peak and

583
00:25:10,679 --> 00:25:15,419
you can imagine in such a dynamic system

584
00:25:12,960 --> 00:25:16,980
the accuracy of our forecasts to ensure

585
00:25:15,419 --> 00:25:20,360
that the system is able to meet this

586
00:25:16,980 --> 00:25:20,360
change is pretty crucial

587
00:25:20,400 --> 00:25:25,980
this little GIF or GIF if you're that

588
00:25:22,679 --> 00:25:28,500
way inclined it shows the evolution in

589
00:25:25,980 --> 00:25:32,100
the penetration of renewable energy in

590
00:25:28,500 --> 00:25:33,779
South Australia so 2008 you can see

591
00:25:32,100 --> 00:25:36,179
really low penetration on the vertical

592
00:25:33,779 --> 00:25:37,140
axis or on about 12 percent and it's all

593
00:25:36,179 --> 00:25:40,140
wind

594
00:25:37,140 --> 00:25:42,299
skip forward only 10 years to 2018 and

595
00:25:40,140 --> 00:25:44,940
you can see that wind has grown

596
00:25:42,299 --> 00:25:47,700
massively backed up with a fair chunk of

597
00:25:44,940 --> 00:25:49,980
rooftop PVE and it's you know happily

598
00:25:47,700 --> 00:25:51,779
hitting 75 penetration in South

599
00:25:49,980 --> 00:25:52,980
Australia but it also ramps down pretty

600
00:25:51,779 --> 00:25:55,860
low as well

601
00:25:52,980 --> 00:25:59,340
and then fast forward to 21 and you've

602
00:25:55,860 --> 00:26:01,679
got your utility scale solar massively

603
00:25:59,340 --> 00:26:04,460
large quantities of rooftop EV and wind

604
00:26:01,679 --> 00:26:06,539
and that's peaking up at around 93

605
00:26:04,460 --> 00:26:08,580
instantaneous penetration in South

606
00:26:06,539 --> 00:26:11,220
Australia however you can see with those

607
00:26:08,580 --> 00:26:12,900
stiff Cliffs and troughs that can change

608
00:26:11,220 --> 00:26:15,000
very very quickly and we need to be able

609
00:26:12,900 --> 00:26:16,740
to see that coming in order to configure

610
00:26:15,000 --> 00:26:20,659
the grid to be able to support that

611
00:26:16,740 --> 00:26:20,659
reliably and keep the lights on

612
00:26:20,700 --> 00:26:26,100
this is what we like to call our ramp

613
00:26:23,159 --> 00:26:29,220
bubbles or bubble tea but this just

614
00:26:26,100 --> 00:26:31,860
shows off again the rate of change of

615
00:26:29,220 --> 00:26:34,260
that mix so you've again got the

616
00:26:31,860 --> 00:26:36,240
penetration on your vertical axis and

617
00:26:34,260 --> 00:26:39,360
then your change in megawatts over 30

618
00:26:36,240 --> 00:26:42,240
minutes on your horizontal axis so back

619
00:26:39,360 --> 00:26:43,260
in 2008 a nice stable system not much is

620
00:26:42,240 --> 00:26:46,200
changing

621
00:26:43,260 --> 00:26:48,480
skip forward 2018 and you've got much

622
00:26:46,200 --> 00:26:50,700
higher penetrations and you can see some

623
00:26:48,480 --> 00:26:53,100
quite large ramps you know pushing up to

624
00:26:50,700 --> 00:26:54,600
700 megawatts in half an hour which is a

625
00:26:53,100 --> 00:26:55,320
significant change in a short period of

626
00:26:54,600 --> 00:26:57,600
time

627
00:26:55,320 --> 00:27:01,020
and then to 2021 you can see we're

628
00:26:57,600 --> 00:27:02,580
closing up at near 93 94 our ramps are

629
00:27:01,020 --> 00:27:04,919
getting bigger and there's a lot more

630
00:27:02,580 --> 00:27:07,620
solar contributing to the ramps as

631
00:27:04,919 --> 00:27:09,419
transient cloud cover really starts to

632
00:27:07,620 --> 00:27:11,880
impact the grid

633
00:27:09,419 --> 00:27:14,159
so again just highlighting the dynamic

634
00:27:11,880 --> 00:27:18,600
nature of what we're working with

635
00:27:14,159 --> 00:27:21,000
the slot uh hopes to illustrate that in

636
00:27:18,600 --> 00:27:22,679
a slightly different manager manner you

637
00:27:21,000 --> 00:27:25,140
can see back in the day we had a large

638
00:27:22,679 --> 00:27:27,960
sphere of control and low uncertainty

639
00:27:25,140 --> 00:27:30,240
today our uncertainty is growing and our

640
00:27:27,960 --> 00:27:31,620
sphere of control is diminishing and we

641
00:27:30,240 --> 00:27:34,080
don't want to get to the undesirable

642
00:27:31,620 --> 00:27:36,840
future state where we

643
00:27:34,080 --> 00:27:39,179
keep the lights off I'm going to skim

644
00:27:36,840 --> 00:27:40,620
through these slides because we've

645
00:27:39,179 --> 00:27:43,679
covered this already again just showing

646
00:27:40,620 --> 00:27:45,480
stability and this is showing you know a

647
00:27:43,679 --> 00:27:48,000
typical forecast but what can actually

648
00:27:45,480 --> 00:27:50,340
happen due to a change in cloud cover so

649
00:27:48,000 --> 00:27:52,080
the dotted line in light pink is our

650
00:27:50,340 --> 00:27:53,120
forecast the dark purple is what

651
00:27:52,080 --> 00:27:55,799
actually happened

652
00:27:53,120 --> 00:27:56,940
so you can see we don't always get

653
00:27:55,799 --> 00:27:58,740
things right

654
00:27:56,940 --> 00:28:00,840
and if that's not fun enough Mother

655
00:27:58,740 --> 00:28:03,360
Nature likes to get involved and throw

656
00:28:00,840 --> 00:28:05,220
things at us as well like bushfires dust

657
00:28:03,360 --> 00:28:06,380
storms tornadoes and convective

658
00:28:05,220 --> 00:28:08,400
downbursts

659
00:28:06,380 --> 00:28:09,179
the latter of which are very good at

660
00:28:08,400 --> 00:28:11,400
bringing down transmission

661
00:28:09,179 --> 00:28:14,000
infrastructure in South Australia as we

662
00:28:11,400 --> 00:28:16,140
can see in the lower right picture and

663
00:28:14,000 --> 00:28:19,620
the New South Wales and Victorian

664
00:28:16,140 --> 00:28:21,900
bushfires from 1920 had significant

665
00:28:19,620 --> 00:28:23,640
impact on solar generation as well as

666
00:28:21,900 --> 00:28:25,980
obviously Health implications in

667
00:28:23,640 --> 00:28:28,140
people's lives

668
00:28:25,980 --> 00:28:30,900
it's a way to you from here

669
00:28:28,140 --> 00:28:33,360
uh firstly I guess where we are today

670
00:28:30,900 --> 00:28:35,220
we've developed a number of situational

671
00:28:33,360 --> 00:28:36,900
awareness tools

672
00:28:35,220 --> 00:28:39,120
um that image is just showing that we

673
00:28:36,900 --> 00:28:41,100
pipe real-time data from all the bureau

674
00:28:39,120 --> 00:28:44,220
weather stations around the country that

675
00:28:41,100 --> 00:28:45,900
updates every minute and we pipe that

676
00:28:44,220 --> 00:28:47,340
into the control room so we can see cool

677
00:28:45,900 --> 00:28:50,039
changes of weather change is happening

678
00:28:47,340 --> 00:28:51,840
in real time and reconfigure the grid as

679
00:28:50,039 --> 00:28:53,100
well as other bespoke tools which you

680
00:28:51,840 --> 00:28:54,299
probably don't have time to go into too

681
00:28:53,100 --> 00:28:56,520
much today

682
00:28:54,299 --> 00:28:58,860
I will touch on some of the other models

683
00:28:56,520 --> 00:29:01,559
similar to Peter's team we've

684
00:28:58,860 --> 00:29:03,539
redeveloped some of our awifs and asifs

685
00:29:01,559 --> 00:29:06,179
models which are our Australian wind

686
00:29:03,539 --> 00:29:08,580
energy forecasting and Australian solar

687
00:29:06,179 --> 00:29:10,740
energy forecasting models and by

688
00:29:08,580 --> 00:29:13,080
redeveloping these in-house using open

689
00:29:10,740 --> 00:29:15,600
source software over the last year or so

690
00:29:13,080 --> 00:29:18,240
we've been able to realize at 11 and 14

691
00:29:15,600 --> 00:29:19,919
increase in accuracy which ultimately

692
00:29:18,240 --> 00:29:23,220
results in lower powerballs for

693
00:29:19,919 --> 00:29:25,679
everybody so we think that's pretty good

694
00:29:23,220 --> 00:29:28,980
so why python in addition to the reasons

695
00:29:25,679 --> 00:29:32,039
that Peta mentioned for our team being a

696
00:29:28,980 --> 00:29:33,539
data science oriented the

697
00:29:32,039 --> 00:29:35,940
um

698
00:29:33,539 --> 00:29:38,000
the vast majority of ml development

699
00:29:35,940 --> 00:29:40,500
happens in Python NR

700
00:29:38,000 --> 00:29:44,039
this obviously gives us the ability to

701
00:29:40,500 --> 00:29:47,039
dive under the hood and helps us deliver

702
00:29:44,039 --> 00:29:49,460
on our goal of 100 Renewables I'm on my

703
00:29:47,039 --> 00:29:49,460
last slide

704
00:29:50,100 --> 00:29:54,360
um so yeah our data science team and I

705
00:29:52,799 --> 00:29:56,220
realize this is a safe space so I can

706
00:29:54,360 --> 00:29:58,200
say this we've typically been in our

707
00:29:56,220 --> 00:30:00,120
shop historically and we still like oh

708
00:29:58,200 --> 00:30:01,440
for data ranking however as we're

709
00:30:00,120 --> 00:30:04,140
producing more and more production

710
00:30:01,440 --> 00:30:06,600
forecasts we gravitating more towards

711
00:30:04,140 --> 00:30:08,460
python primarily for

712
00:30:06,600 --> 00:30:10,980
um I guess the reasons Peter mentioned

713
00:30:08,460 --> 00:30:13,320
but in addition the supporting tooling

714
00:30:10,980 --> 00:30:16,440
that it allows us to bring our products

715
00:30:13,320 --> 00:30:20,779
to Market so to speak internally faster

716
00:30:16,440 --> 00:30:20,779
with less support than doing it in our

717
00:30:20,940 --> 00:30:24,679
um so yeah that that's the end

718
00:30:26,399 --> 00:30:30,059
schedule no that's all good that's it

719
00:30:28,320 --> 00:30:31,870
we're done could we have a uh thank you

720
00:30:30,059 --> 00:30:38,950
for that energizing tool

721
00:30:31,870 --> 00:30:38,950
[Applause]