1
00:00:00,060 --> 00:00:02,500
The following content is
provided under a Creative

2
00:00:02,500 --> 00:00:04,010
Commons license.

3
00:00:04,010 --> 00:00:06,350
Your support will help
MIT OpenCourseWare

4
00:00:06,350 --> 00:00:10,720
continue to offer high quality
educational resources for free.

5
00:00:10,720 --> 00:00:13,340
To make a donation or
view additional materials

6
00:00:13,340 --> 00:00:17,209
from hundreds of MIT courses,
visit MIT OpenCourseWare

7
00:00:17,209 --> 00:00:17,834
at ocw.mit.edu.

8
00:00:26,110 --> 00:00:27,920
PROFESSOR: Final
presentations should

9
00:00:27,920 --> 00:00:30,650
be very exciting--
fruits of your labor

10
00:00:30,650 --> 00:00:32,540
over the entire
semester in reality.

11
00:00:32,540 --> 00:00:34,642
The fundamentals
in the first third,

12
00:00:34,642 --> 00:00:37,100
the technologies in the second,
gearing up toward the cross

13
00:00:37,100 --> 00:00:38,430
cutting themes in the third.

14
00:00:38,430 --> 00:00:41,070
I understand that we've
had an accelerated project

15
00:00:41,070 --> 00:00:42,960
schedule this semester.

16
00:00:42,960 --> 00:00:45,250
We've completed
the entire projects

17
00:00:45,250 --> 00:00:47,670
over the duration of
about a month and a half,

18
00:00:47,670 --> 00:00:50,530
so you are to be congratulated
for your hard work

19
00:00:50,530 --> 00:00:54,430
in a very intense period of
time McKenzie tiger team style.

20
00:00:54,430 --> 00:00:57,350
For that, I reward you
with doughnuts and coffee

21
00:00:57,350 --> 00:00:58,060
over there.

22
00:00:58,060 --> 00:01:00,101
I understand many of you
were up late last night,

23
00:01:00,101 --> 00:01:02,690
so you're welcome to ingest
some shortchange carbohydrates

24
00:01:02,690 --> 00:01:04,510
and some caffeine.

25
00:01:04,510 --> 00:01:07,140
If you would like to
get some, get it now.

26
00:01:07,140 --> 00:01:09,390
We're going to have another
minute of blah, blah, blah

27
00:01:09,390 --> 00:01:10,931
before we dive into
the presentations

28
00:01:10,931 --> 00:01:12,280
and the real fun begins.

29
00:01:12,280 --> 00:01:14,340
I'd like to introduce
our panelists up here

30
00:01:14,340 --> 00:01:17,230
in the front who will be
the evaluation criteria

31
00:01:17,230 --> 00:01:20,120
a la American Idol style,
except that you're, of course,

32
00:01:20,120 --> 00:01:23,750
a lot smarter and
equally well dressed.

33
00:01:23,750 --> 00:01:25,540
Starting from right
to left in front of me

34
00:01:25,540 --> 00:01:29,870
we have Dr. Jasmin Hofstetter,
who comes from IES, Spain.

35
00:01:29,870 --> 00:01:32,460
That's the Institute for
Solar Energy in Spain.

36
00:01:32,460 --> 00:01:34,955
That's where she did her
PhD with Antonio Luque.

37
00:01:34,955 --> 00:01:37,330
Those who have been studying
intermediate band solar cell

38
00:01:37,330 --> 00:01:39,520
materials may know
the name as one

39
00:01:39,520 --> 00:01:41,270
of the fathers of the field.

40
00:01:41,270 --> 00:01:44,110
She studied under Antonio
Luque's organization

41
00:01:44,110 --> 00:01:49,500
with Carlos de Canizo and is the
winner of presentation awards

42
00:01:49,500 --> 00:01:51,840
at scientific
conferences, among others,

43
00:01:51,840 --> 00:01:53,680
so Jasmin is welcome here.

44
00:01:53,680 --> 00:01:56,050
We have thought that Dr.
Mark Winkler, as well,

45
00:01:56,050 --> 00:01:59,390
a PhD in Eric Mazur's
laboratory at Harvard.

46
00:01:59,390 --> 00:02:01,270
Those who are familiar
with femtosecond laser

47
00:02:01,270 --> 00:02:04,510
characterization may be familiar
with Eric Mazur, also one

48
00:02:04,510 --> 00:02:05,970
of the fathers of the field.

49
00:02:05,970 --> 00:02:08,690
Mark started the Harvard
Journal Energy Club,

50
00:02:08,690 --> 00:02:12,250
which is Harvard's version of
the MIT Energy Club-- a lot

51
00:02:12,250 --> 00:02:15,340
smaller and a lot less
dynamic than MIT's version,

52
00:02:15,340 --> 00:02:18,060
but nevertheless to
be congratulated.

53
00:02:18,060 --> 00:02:20,690
And of course, a very
good organization.

54
00:02:20,690 --> 00:02:21,190
I kid.

55
00:02:21,190 --> 00:02:23,780
There's a little bit
of MIT Harvard rivalry.

56
00:02:23,780 --> 00:02:25,570
And of course, our
very own Joe Sullivan,

57
00:02:25,570 --> 00:02:28,200
who has been with you
the entire semester.

58
00:02:28,200 --> 00:02:30,190
For those who might
not be familiar as much

59
00:02:30,190 --> 00:02:32,920
with this research as you
are with his teaching,

60
00:02:32,920 --> 00:02:36,670
Joe is studying intermediate
band solar cell materials here

61
00:02:36,670 --> 00:02:39,350
at MIT in the Media Lab
and has been working

62
00:02:39,350 --> 00:02:41,040
for the last-- what is it now?

63
00:02:41,040 --> 00:02:41,510
JOE SULLIVAN: Three and a half.

64
00:02:41,510 --> 00:02:42,968
PROFESSOR: Three
and a half years--

65
00:02:42,968 --> 00:02:45,250
focused on intermediate
band solar cell materials,

66
00:02:45,250 --> 00:02:48,200
coming from a very broad
background in energy

67
00:02:48,200 --> 00:02:50,430
from climate science.

68
00:02:50,430 --> 00:02:54,680
So with that, I'd like to
welcome our first team down,

69
00:02:54,680 --> 00:02:56,930
and the floor is yours.

70
00:02:56,930 --> 00:03:00,400
STUDENT 1: Good morning, we are
the PV smart retrofit group,

71
00:03:00,400 --> 00:03:04,300
and our project goal was
to assess whether or not

72
00:03:04,300 --> 00:03:06,540
the there was an
electrical benefit

73
00:03:06,540 --> 00:03:10,210
or loss from retrofitting an
old home with a PV system.

74
00:03:10,210 --> 00:03:12,590
Now, in less lofty terms,
it essentially means

75
00:03:12,590 --> 00:03:14,450
from an on site
energy perspective,

76
00:03:14,450 --> 00:03:18,110
does it make sense to put
PV panels on my house?

77
00:03:18,110 --> 00:03:20,540
You'd think that that's a
kind of an obvious question.

78
00:03:20,540 --> 00:03:22,240
We'd all say, well, yeah.

79
00:03:22,240 --> 00:03:25,460
We produce energy for
free, except that you

80
00:03:25,460 --> 00:03:27,977
have to consider other
things, such as shading

81
00:03:27,977 --> 00:03:29,560
from trees that you'd
have to cut down

82
00:03:29,560 --> 00:03:31,685
or the color of the roof
that you might be changing

83
00:03:31,685 --> 00:03:33,400
by adding a black panel.

84
00:03:33,400 --> 00:03:35,900
These would both reduce the
thermal load of your house

85
00:03:35,900 --> 00:03:37,260
in normal situation.

86
00:03:37,260 --> 00:03:40,470
So by adding a PV panel and
increasing the thermal load,

87
00:03:40,470 --> 00:03:43,090
you actually add the
energy to cool your house

88
00:03:43,090 --> 00:03:43,840
during the summer.

89
00:03:46,410 --> 00:03:50,350
We considered multiple
variables in this project.

90
00:03:50,350 --> 00:03:53,010
One was location, which
has an effect on the amount

91
00:03:53,010 --> 00:03:54,870
of sunlight you're receiving.

92
00:03:54,870 --> 00:03:57,990
The PV panel's
presence and its size.

93
00:03:57,990 --> 00:04:00,060
We had a couple of
situations where

94
00:04:00,060 --> 00:04:03,650
there was no panel as
a kind of a baseline.

95
00:04:03,650 --> 00:04:07,540
And then also different sizes
to figure out if a bigger

96
00:04:07,540 --> 00:04:09,680
panel had a bigger effect.

97
00:04:09,680 --> 00:04:12,280
We looked at two different
colors, black and white.

98
00:04:12,280 --> 00:04:14,760
Those are the ends
of the spectrum,

99
00:04:14,760 --> 00:04:17,250
and they give us
endpoints to look at,

100
00:04:17,250 --> 00:04:20,130
and the color matters
because a darker

101
00:04:20,130 --> 00:04:22,690
color will absorb more heat.

102
00:04:22,690 --> 00:04:24,150
We looked at roof pitch.

103
00:04:24,150 --> 00:04:26,250
The reason for this--
well, first off, roof pitch

104
00:04:26,250 --> 00:04:28,010
is the angle of your roof.

105
00:04:28,010 --> 00:04:31,170
And we looked at this because
we assumed that our panels were

106
00:04:31,170 --> 00:04:33,570
fixed and parallel to
the roof, so this kind

107
00:04:33,570 --> 00:04:36,700
of controlled what
angle your solar panel

108
00:04:36,700 --> 00:04:38,532
was facing towards the sun.

109
00:04:38,532 --> 00:04:40,240
And finally you had
your house footprint,

110
00:04:40,240 --> 00:04:42,230
which is the area
that the house covers,

111
00:04:42,230 --> 00:04:44,146
and when you combine
that with the roof pitch,

112
00:04:44,146 --> 00:04:46,197
you get the area of the
roof, which is really

113
00:04:46,197 --> 00:04:47,280
what we're concerned with.

114
00:04:50,010 --> 00:04:52,940
We had five
scenarios, and as you

115
00:04:52,940 --> 00:04:54,840
can see from our
cute little diagrams,

116
00:04:54,840 --> 00:04:57,980
we have a black
roof, a white roof,

117
00:04:57,980 --> 00:05:01,990
a white roof with the tree, a
white roof with a solar panel,

118
00:05:01,990 --> 00:05:04,085
and a white roof with a
solar panel and a cut tree

119
00:05:04,085 --> 00:05:07,810
so the solar panel is
getting plenty of sunlight.

120
00:05:07,810 --> 00:05:09,520
In evaluating the
five scenarios,

121
00:05:09,520 --> 00:05:12,220
we had three models of
increasing complexity from left

122
00:05:12,220 --> 00:05:14,730
to right, which you
can note from the fact

123
00:05:14,730 --> 00:05:18,610
that model one only
covers three of the five,

124
00:05:18,610 --> 00:05:22,962
and these models all
had common assumptions.

125
00:05:22,962 --> 00:05:24,670
The first was that
you had a single story

126
00:05:24,670 --> 00:05:27,510
house in a suburban
locations, so you

127
00:05:27,510 --> 00:05:30,790
didn't have shading from
other buildings, for example.

128
00:05:30,790 --> 00:05:34,660
We had a common house size of
about 2000 square feet, or 186

129
00:05:34,660 --> 00:05:36,230
square meters.

130
00:05:36,230 --> 00:05:38,210
The roof pitch,
which in construction

131
00:05:38,210 --> 00:05:42,060
is set usually at 5/12,
which means 5 inches of rise

132
00:05:42,060 --> 00:05:44,790
for 12 inches of travel.

133
00:05:44,790 --> 00:05:46,490
We had an unfinished
attic space,

134
00:05:46,490 --> 00:05:48,470
which means that it's
sealed to the outside,

135
00:05:48,470 --> 00:05:52,360
but you didn't make it
livable, a five kilowatt PV

136
00:05:52,360 --> 00:05:55,870
system covering
36 square meters,

137
00:05:55,870 --> 00:06:00,960
and we chose reflectance values
of 0.08 for black and 0.35

138
00:06:00,960 --> 00:06:04,790
for white, and this, again,
reflects the effect that color

139
00:06:04,790 --> 00:06:07,800
has on heat absorption.

140
00:06:07,800 --> 00:06:09,795
With that, I will turn
you over to, Jordan.

141
00:06:15,230 --> 00:06:16,980
JORDAN: So the first
one that we looked at

142
00:06:16,980 --> 00:06:20,830
was basically using
most readily available

143
00:06:20,830 --> 00:06:22,695
and simple models there can be.

144
00:06:22,695 --> 00:06:26,170
So this is from the
Department of Energy

145
00:06:26,170 --> 00:06:28,770
to measure-- well,
to get a gauge of how

146
00:06:28,770 --> 00:06:31,460
the color of the roof
and the roof properties

147
00:06:31,460 --> 00:06:34,460
affect the thermal
loads in the house.

148
00:06:34,460 --> 00:06:36,100
This is a simple one
dimensional model

149
00:06:36,100 --> 00:06:39,760
where you've got an inside
cavity of 65 Fahrenheit,

150
00:06:39,760 --> 00:06:41,570
and you basically
input the location,

151
00:06:41,570 --> 00:06:49,010
and from that it has a lookup
table of the average insulation

152
00:06:49,010 --> 00:06:52,370
on the house, as well as the
number of heating degree days

153
00:06:52,370 --> 00:06:57,550
and cooling degree days relative
to that 65 inside temperature.

154
00:06:57,550 --> 00:06:59,300
The other parameters
are just at the roof,

155
00:06:59,300 --> 00:07:01,440
so we insert the
reflectance, and we're

156
00:07:01,440 --> 00:07:05,730
using values that represent
real tiles for an average house,

157
00:07:05,730 --> 00:07:08,760
black and white, as well as the
thermal resistance and the heat

158
00:07:08,760 --> 00:07:12,460
absorbance of the tiles--
from this model, the outputs

159
00:07:12,460 --> 00:07:16,270
and the thermal loads
in terms of the heating

160
00:07:16,270 --> 00:07:19,750
you need to put in, as well
as the cooling energy load.

161
00:07:22,760 --> 00:07:27,050
So with the thermal
model assessed,

162
00:07:27,050 --> 00:07:31,980
we can assess the
photovoltaic output.

163
00:07:31,980 --> 00:07:35,210
And for this, we're just
using a simple model PVWatts.

164
00:07:35,210 --> 00:07:40,790
This basically takes in
the location and the angle

165
00:07:40,790 --> 00:07:42,805
of the panels, as well
as a derate factor

166
00:07:42,805 --> 00:07:46,220
from converting to AC to DC.

167
00:07:46,220 --> 00:07:50,324
It's quite a simple model,
and output from this

168
00:07:50,324 --> 00:07:51,990
is the amount of
kilowatt hours per year

169
00:07:51,990 --> 00:07:53,220
that you get from the panels.

170
00:08:18,970 --> 00:08:22,880
STUDENT 2: So I'll be talking
about a couple thermal electric

171
00:08:22,880 --> 00:08:23,380
model.

172
00:08:23,380 --> 00:08:26,720
This is the model that
we built in our group,

173
00:08:26,720 --> 00:08:28,890
and we developed
this model based

174
00:08:28,890 --> 00:08:32,600
on two sets of
individual parameters

175
00:08:32,600 --> 00:08:33,980
and individual models.

176
00:08:33,980 --> 00:08:37,360
I would say that one comprises
of the thermal model,

177
00:08:37,360 --> 00:08:39,470
and there is the electric model.

178
00:08:39,470 --> 00:08:42,559
So what we need to note
from this model here

179
00:08:42,559 --> 00:08:47,620
is it's a further
step in complexity

180
00:08:47,620 --> 00:08:52,000
when compared to the module
one that Jordan just discussed.

181
00:08:52,000 --> 00:08:55,080
And it takes into account
various input parameters

182
00:08:55,080 --> 00:08:57,880
that the model one
doesn't take into account.

183
00:08:57,880 --> 00:09:00,570
So the basic structure of
this model is as follows.

184
00:09:00,570 --> 00:09:02,960
We have a thermal model
which takes into account

185
00:09:02,960 --> 00:09:06,370
several input conditions, such
as the insulation and shading,

186
00:09:06,370 --> 00:09:09,820
and it outputs a living
space temperature,

187
00:09:09,820 --> 00:09:13,250
which is this temperature of
the living room in our house.

188
00:09:13,250 --> 00:09:16,510
And then this temperature is fed
into an electric model, which

189
00:09:16,510 --> 00:09:19,500
calculates the cost
and energy values,

190
00:09:19,500 --> 00:09:21,860
and thus we can compare an
energy production and energy

191
00:09:21,860 --> 00:09:23,310
consumption.

192
00:09:23,310 --> 00:09:25,360
Going to the thermal
model in detail,

193
00:09:25,360 --> 00:09:27,700
I've just shown a picture here.

194
00:09:27,700 --> 00:09:30,260
So it considers
basically when we

195
00:09:30,260 --> 00:09:33,930
start from the top of the house,
we use insulation and shading

196
00:09:33,930 --> 00:09:35,550
as the input parameters.

197
00:09:35,550 --> 00:09:38,050
We then calculate the
temperature of the PV panel,

198
00:09:38,050 --> 00:09:39,890
and then the
temperature of the roof

199
00:09:39,890 --> 00:09:42,510
using two different
energy balance models.

200
00:09:42,510 --> 00:09:44,600
And these two energy
balance models

201
00:09:44,600 --> 00:09:46,840
are pretty robust in the
sense that they consider

202
00:09:46,840 --> 00:09:49,190
all these physical phenomenon
which are realistic,

203
00:09:49,190 --> 00:09:51,280
such as the
convection, radiation,

204
00:09:51,280 --> 00:09:54,250
and PV electrical output.

205
00:09:54,250 --> 00:09:56,940
And based on these
energy balance equations,

206
00:09:56,940 --> 00:09:59,630
we can calculate the PV
panel temperature and then

207
00:09:59,630 --> 00:10:00,820
the roof temperature.

208
00:10:00,820 --> 00:10:02,910
And once we get these
two temperatures,

209
00:10:02,910 --> 00:10:05,640
we get the heat flux
that goes into the roof

210
00:10:05,640 --> 00:10:07,350
and that enters the attic.

211
00:10:07,350 --> 00:10:09,309
And once you get
this, we find out

212
00:10:09,309 --> 00:10:11,350
the attic temperature,
which then determines what

213
00:10:11,350 --> 00:10:12,790
is the ceiling temperature.

214
00:10:12,790 --> 00:10:15,320
And then we consider the
convection via ceiling,

215
00:10:15,320 --> 00:10:18,250
and then finally we end up with
the living space temperature.

216
00:10:18,250 --> 00:10:20,560
And then the couple this
living space temperature

217
00:10:20,560 --> 00:10:23,840
with another electric model,
which I'll discuss now.

218
00:10:23,840 --> 00:10:28,200
So the electric model is
basically based on an ideal gas

219
00:10:28,200 --> 00:10:29,150
assumption.

220
00:10:29,150 --> 00:10:31,330
So it basically--
what it does is

221
00:10:31,330 --> 00:10:34,490
it calculates the energy
needed by the AC, which

222
00:10:34,490 --> 00:10:36,470
could be the heating
or cooling, in order

223
00:10:36,470 --> 00:10:39,480
to maintain the living space
at a particular temperature.

224
00:10:39,480 --> 00:10:42,190
And we use this formula,
m dot Cp delta T,

225
00:10:42,190 --> 00:10:43,790
which is an ideal
gas formula, which

226
00:10:43,790 --> 00:10:47,190
gives out the energy
needed by the cooler.

227
00:10:47,190 --> 00:10:52,630
And then so the electrical model
uses the power consumption,

228
00:10:52,630 --> 00:10:57,210
and then we know the power
production through PV output.

229
00:10:57,210 --> 00:11:00,620
So comparing these
two, we can really

230
00:11:00,620 --> 00:11:04,520
assess whether PV installation
is favorable or not.

231
00:11:04,520 --> 00:11:07,600
And this is just
the model in making.

232
00:11:07,600 --> 00:11:09,950
What we want to signify
here is we actually

233
00:11:09,950 --> 00:11:12,030
made this model on
our own, and this

234
00:11:12,030 --> 00:11:13,620
is the MATLAB code we wrote.

235
00:11:13,620 --> 00:11:17,530
And the model not only just
predicts energy values,

236
00:11:17,530 --> 00:11:19,330
but it also can do
a lot more things,

237
00:11:19,330 --> 00:11:21,090
such as predicting temperature.

238
00:11:21,090 --> 00:11:23,440
And what I've shown
here is the PV output,

239
00:11:23,440 --> 00:11:25,780
and then the cooling
load that is required,

240
00:11:25,780 --> 00:11:28,270
so it can do a lot of
other things, as well.

241
00:11:28,270 --> 00:11:30,520
This is the
temperature of the roof

242
00:11:30,520 --> 00:11:32,960
in terms of direct sunlight
and diffuse sunlight.

243
00:11:32,960 --> 00:11:35,270
So if anyone is
interested, I'd be

244
00:11:35,270 --> 00:11:37,050
happy to discuss with them more.

245
00:11:37,050 --> 00:11:37,760
Thank you.

246
00:11:37,760 --> 00:11:39,510
And I'll now pass
it on to Heidi.

247
00:11:56,055 --> 00:11:57,680
HEIDI: So I take over
from here talking

248
00:11:57,680 --> 00:12:01,260
about the third and most
complex model that we used.

249
00:12:01,260 --> 00:12:04,930
For this model, we used
two different softwares--

250
00:12:04,930 --> 00:12:07,840
one called BEopt from
NREL, and the second,

251
00:12:07,840 --> 00:12:10,730
called EnergyPlus,
which you'll see later,

252
00:12:10,730 --> 00:12:13,080
developed by the
Department of Energy.

253
00:12:13,080 --> 00:12:15,760
So what we did in this model
was we took into account

254
00:12:15,760 --> 00:12:18,880
the 3D effects of these
thermal and electric loads

255
00:12:18,880 --> 00:12:22,220
that have been talked about
in the other two models.

256
00:12:22,220 --> 00:12:25,670
The first thing we did was to
actually model the 3D house,

257
00:12:25,670 --> 00:12:27,730
as you can see right over here.

258
00:12:27,730 --> 00:12:31,750
BEopt allows for a very nice
interface, where you can easily

259
00:12:31,750 --> 00:12:34,790
model the house and easily
and put a whole bunch of input

260
00:12:34,790 --> 00:12:36,690
parameters for the house.

261
00:12:36,690 --> 00:12:38,330
And for these input
parameters, we

262
00:12:38,330 --> 00:12:40,790
consulted with an experienced
building inspector

263
00:12:40,790 --> 00:12:44,770
for the construction inputs
and used BEopt default values

264
00:12:44,770 --> 00:12:47,840
for the rest of the inputs.

265
00:12:47,840 --> 00:12:50,110
Once this model of
the house was done,

266
00:12:50,110 --> 00:12:54,760
we had to actually
export it into EnergyPlus

267
00:12:54,760 --> 00:13:02,590
because EnergyPlus gives us
a much more detailed look

268
00:13:02,590 --> 00:13:06,550
into all these different
parameters, I guess.

269
00:13:06,550 --> 00:13:08,990
And you can actually go in
and modify different things.

270
00:13:08,990 --> 00:13:13,530
For the materials, we can
modify every single material

271
00:13:13,530 --> 00:13:16,450
property-- conductivity,
density, specific heat--

272
00:13:16,450 --> 00:13:19,050
and we actually did
that for the roof.

273
00:13:19,050 --> 00:13:20,890
And we also removed
a whole bunch

274
00:13:20,890 --> 00:13:26,430
of other miscellaneous loads
that BEopt had included.

275
00:13:26,430 --> 00:13:29,600
So this is how we modeled
the trees and the panels

276
00:13:29,600 --> 00:13:33,090
that we talked about in
our scenarios before.

277
00:13:33,090 --> 00:13:36,910
For the trees, we modelled
them as these really large 5

278
00:13:36,910 --> 00:13:41,790
by 20 meter rectangles that
act as shading for the house.

279
00:13:41,790 --> 00:13:44,220
So the trees are located on
the south side of the house,

280
00:13:44,220 --> 00:13:46,250
and we modelled them
as deciduous trees,

281
00:13:46,250 --> 00:13:49,990
so we set up a transmitting
schedule so that they

282
00:13:49,990 --> 00:13:53,130
have a higher transmittance in
the winter when the leaves have

283
00:13:53,130 --> 00:13:55,640
fallen and a low
one in the summer.

284
00:13:55,640 --> 00:13:58,820
And for the case of scenario
E in which the trees are cut,

285
00:13:58,820 --> 00:14:03,030
we modelled the trees as
being five meters tall.

286
00:14:03,030 --> 00:14:05,920
On the other hand,
for the PV panel,

287
00:14:05,920 --> 00:14:09,330
we modeled it as being
fixed on the roof.

288
00:14:09,330 --> 00:14:12,980
We actually had to completely
change the model from BEopt,

289
00:14:12,980 --> 00:14:15,230
because they modeled
it as being decoupled

290
00:14:15,230 --> 00:14:19,670
from the entire system, placed
30 meters away from the house.

291
00:14:19,670 --> 00:14:23,160
So we completely changed
that, and we read extensively

292
00:14:23,160 --> 00:14:27,970
into EnergyPlus literature and
found this particular object--

293
00:14:27,970 --> 00:14:30,290
I guess you could call
it-- called the integrated

294
00:14:30,290 --> 00:14:32,820
exterior vented cavity object.

295
00:14:32,820 --> 00:14:37,250
And what this does is it
models a surface as being,

296
00:14:37,250 --> 00:14:41,470
in our case, 0.5 meters
away from the roof,

297
00:14:41,470 --> 00:14:44,210
and it models the convection
and radiation between these two

298
00:14:44,210 --> 00:14:45,790
surfaces.

299
00:14:45,790 --> 00:14:47,780
We also considered
the solar panel

300
00:14:47,780 --> 00:14:50,470
to have a solar
absorbance of 0.92

301
00:14:50,470 --> 00:15:06,320
and thermal emissivity of 0.9.

302
00:15:06,320 --> 00:15:08,606
STUDENT 3: And so
we actually got

303
00:15:08,606 --> 00:15:10,230
a lot of results, as
you might imagine,

304
00:15:10,230 --> 00:15:12,120
from all those different
models, but just

305
00:15:12,120 --> 00:15:15,050
for the purposes of comparison
for the presentation,

306
00:15:15,050 --> 00:15:17,685
we're just going to show
you-- just summarize results.

307
00:15:17,685 --> 00:15:20,060
And basically what we're
showing you here is specifically

308
00:15:20,060 --> 00:15:22,990
for Boston, and this
figure that we're showing

309
00:15:22,990 --> 00:15:26,070
is the y-axis is the
relative energy gain.

310
00:15:26,070 --> 00:15:29,840
So in order to compare them,
we decided within each model

311
00:15:29,840 --> 00:15:32,070
compare it to a
common situation.

312
00:15:32,070 --> 00:15:35,410
So we decided to say
that, if we're in Boston,

313
00:15:35,410 --> 00:15:37,900
let's say we start with
a white roof and a tree,

314
00:15:37,900 --> 00:15:40,300
so that's scenario C there,
so that's why it's 0.

315
00:15:40,300 --> 00:15:42,500
So everything is in
comparison to that.

316
00:15:42,500 --> 00:15:45,210
And right off, we could see
that, as we might expect,

317
00:15:45,210 --> 00:15:49,050
putting a PV on makes
sense energetically.

318
00:15:49,050 --> 00:15:51,390
And specifically
scenario D, which

319
00:15:51,390 --> 00:15:53,910
is where you completely remove
the tree instead of just

320
00:15:53,910 --> 00:15:57,800
cutting it, in Boston at least,
is what makes the most sense.

321
00:15:57,800 --> 00:16:00,980
And this is for both
models-- actually,

322
00:16:00,980 --> 00:16:03,230
for all the models-- although
model one can't really

323
00:16:03,230 --> 00:16:06,690
model a tree necessarily.

324
00:16:06,690 --> 00:16:10,390
So that's why it basically
doesn't apply for that.

325
00:16:10,390 --> 00:16:13,990
In terms of comparing the
results between the models,

326
00:16:13,990 --> 00:16:17,780
model one actually does a
pretty decent job in Boston

327
00:16:17,780 --> 00:16:22,040
in getting close to model
three, which is impressive,

328
00:16:22,040 --> 00:16:25,690
because model one is
significantly more crude, much

329
00:16:25,690 --> 00:16:29,790
simpler than model three, which
required many, many inputs.

330
00:16:29,790 --> 00:16:33,050
And the reason we believe the
discrepancy is between models

331
00:16:33,050 --> 00:16:36,230
two and three is that
model two-- the way

332
00:16:36,230 --> 00:16:39,710
that we basically treated the
solar insulation-- it doesn't

333
00:16:39,710 --> 00:16:42,360
treat diffuse
sunlight differently,

334
00:16:42,360 --> 00:16:48,060
whereas basically the models
PVWatts and EnergyPlus will

335
00:16:48,060 --> 00:16:50,210
take all that into
account, so we

336
00:16:50,210 --> 00:16:52,560
think that's a larger
reason for that.

337
00:16:52,560 --> 00:16:55,750
And then similar
thing in Phoenix.

338
00:16:55,750 --> 00:16:58,880
And basically we see a
slightly different case here.

339
00:16:58,880 --> 00:17:02,740
Actually scenario E, where
you just-- you install the PV,

340
00:17:02,740 --> 00:17:04,490
but you cut the tree
instead of completely

341
00:17:04,490 --> 00:17:07,880
removing it-- gives you a
slightly better increase

342
00:17:07,880 --> 00:17:11,440
in net energy gain.

343
00:17:11,440 --> 00:17:13,490
In terms of-- and
this is, again,

344
00:17:13,490 --> 00:17:15,170
just in terms of energies.

345
00:17:15,170 --> 00:17:16,619
There could be rounding errors.

346
00:17:16,619 --> 00:17:19,569
This is ignoring the fact
that, if you cut a tree down,

347
00:17:19,569 --> 00:17:23,005
the net greenhouse gas emissions
would be changed and altered,

348
00:17:23,005 --> 00:17:24,880
and this is ignoring
all those other effects.

349
00:17:24,880 --> 00:17:29,400
This is just in terms of net
energy gain of the house.

350
00:17:29,400 --> 00:17:33,310
And just to point out the
discrepancy between model one

351
00:17:33,310 --> 00:17:38,760
here, we think that model
one is actually, again--

352
00:17:38,760 --> 00:17:41,970
that's using PVWatts to
get your energy output.

353
00:17:41,970 --> 00:17:45,920
And that's basically assuming
peak solar insulation,

354
00:17:45,920 --> 00:17:55,100
whereas our models basically
use more empirical formulas

355
00:17:55,100 --> 00:17:58,560
to get the estimates
for your PV output.

356
00:17:58,560 --> 00:18:00,810
And then the last thing we
did to do the sensitivity--

357
00:18:00,810 --> 00:18:05,750
or to compare the models, was do
a sort of sensitivity analysis.

358
00:18:05,750 --> 00:18:07,380
So basically on
your y-axis, what

359
00:18:07,380 --> 00:18:10,190
you have is your percent change
in your relative energy gain--

360
00:18:10,190 --> 00:18:12,640
relative, again, to
that scenario C--

361
00:18:12,640 --> 00:18:14,770
divided by the percent
change in your parameters.

362
00:18:14,770 --> 00:18:17,050
So we just considered
four parameters.

363
00:18:17,050 --> 00:18:20,110
The x-axis is a rough
estimate of how difficult

364
00:18:20,110 --> 00:18:22,900
it would be to actually change
that parameter in your house.

365
00:18:22,900 --> 00:18:26,140
So the far left
one is the PV size.

366
00:18:26,140 --> 00:18:29,210
We just assumed if you went--
instead of a five kilowatt

367
00:18:29,210 --> 00:18:32,890
to a 5.5 kilowatt, so that's
why the price is roughly $3,000.

368
00:18:32,890 --> 00:18:37,940
We assumed about a $5 to $6
per watt installation cost

369
00:18:37,940 --> 00:18:38,660
for that.

370
00:18:38,660 --> 00:18:41,720
So that's basically
the easiest one to do,

371
00:18:41,720 --> 00:18:47,070
and you get various significant
change in your thermal energy

372
00:18:47,070 --> 00:18:49,096
gains because of that.

373
00:18:49,096 --> 00:18:50,470
And then just
because of time I'm

374
00:18:50,470 --> 00:18:54,240
going to rush through these, but
basically a lot of the models

375
00:18:54,240 --> 00:18:56,760
follow similar trends in
terms of the sensitivities.

376
00:18:56,760 --> 00:18:58,052
The last one is the roof pitch.

377
00:18:58,052 --> 00:19:00,135
You obviously wouldn't
really want to change that.

378
00:19:00,135 --> 00:19:01,389
It's very expensive to do.

379
00:19:01,389 --> 00:19:02,930
Luckily, for most
of the models, it's

380
00:19:02,930 --> 00:19:04,346
not actually that
sensitive to it,

381
00:19:04,346 --> 00:19:09,910
at least within a close amount
to where you start with.

382
00:19:09,910 --> 00:19:11,940
So in conclusion,
in all three models,

383
00:19:11,940 --> 00:19:14,879
it makes sense to
install a PV system.

384
00:19:14,879 --> 00:19:16,920
It's kind of what we
expected from the beginning,

385
00:19:16,920 --> 00:19:20,950
but it's nice to get
that sort of conclusion.

386
00:19:20,950 --> 00:19:22,750
For models one and
two, we actually

387
00:19:22,750 --> 00:19:25,247
were able to get pretty
reasonable results,

388
00:19:25,247 --> 00:19:27,080
but they're limited in
terms of what you can

389
00:19:27,080 --> 00:19:28,990
consider within those models.

390
00:19:28,990 --> 00:19:31,530
The advantage of looking
at this basically

391
00:19:31,530 --> 00:19:33,355
is that, if you have
a user that isn't

392
00:19:33,355 --> 00:19:37,340
as familiar with EnergyPlus
in model three, which

393
00:19:37,340 --> 00:19:40,170
is very sophisticated-- it
requires a lot of inputs--

394
00:19:40,170 --> 00:19:41,920
they can still get a
rough estimate, which

395
00:19:41,920 --> 00:19:46,850
is relatively close, using
these much simpler models.

396
00:19:46,850 --> 00:19:49,300
And model three--
again, we were taking

397
00:19:49,300 --> 00:19:51,830
that to be the more
realistic case,

398
00:19:51,830 --> 00:19:55,530
but you'd have to compare
it to real life data

399
00:19:55,530 --> 00:19:59,070
and do an empirical analysis
to see how close it actually

400
00:19:59,070 --> 00:20:00,800
does correlate.

401
00:20:00,800 --> 00:20:02,380
And then just,
again, to summarize

402
00:20:02,380 --> 00:20:04,800
the results we got from
model three-- in Boston,

403
00:20:04,800 --> 00:20:06,650
it makes sense to
install the PV,

404
00:20:06,650 --> 00:20:08,960
but completely remove the tree.

405
00:20:08,960 --> 00:20:11,800
And your payback period
is about 24 years.

406
00:20:11,800 --> 00:20:15,520
In Phoenix, the best scenario
is to install the PV,

407
00:20:15,520 --> 00:20:18,130
cut the tree, and
it's about 51 years.

408
00:20:18,130 --> 00:20:22,100
And interestingly, the maximum
of that relative energy gain

409
00:20:22,100 --> 00:20:24,400
was essentially
the same in both,

410
00:20:24,400 --> 00:20:26,850
even though the
scenarios were different.

411
00:20:26,850 --> 00:20:29,740
So we'd just like to
acknowledge Professor Buonassisi

412
00:20:29,740 --> 00:20:31,960
for helping assist
us and guiding

413
00:20:31,960 --> 00:20:35,020
the direction of the project,
and Bryan Urban at Fraunhofer

414
00:20:35,020 --> 00:20:38,420
and other members at Fraunhofer
for giving us guidance.

415
00:20:38,420 --> 00:20:41,151
And with that, we would like
to ask you for questions.

416
00:20:41,151 --> 00:20:44,097
[APPLAUSE]

417
00:20:49,989 --> 00:20:52,444
AUDIENCE: My
question is I grew up

418
00:20:52,444 --> 00:20:54,408
in a neighborhood that
has a lot of trees,

419
00:20:54,408 --> 00:20:56,372
and so cutting
down all the trees

420
00:20:56,372 --> 00:21:00,791
wouldn't be very practical,
but do you at all consider

421
00:21:00,791 --> 00:21:04,596
PV systems that
could handle shading

422
00:21:04,596 --> 00:21:06,218
at different times of the day?

423
00:21:06,218 --> 00:21:07,877
So somehow decoupling
different parts

424
00:21:07,877 --> 00:21:11,274
of it knowing that some of them
will be in sunlight for part

425
00:21:11,274 --> 00:21:14,990
of the day, some of them will
be shaded, and that will change?

426
00:21:14,990 --> 00:21:17,830
STUDENT 3: So you could
make the model-- especially

427
00:21:17,830 --> 00:21:21,570
in EnergyPlus, you could make
it as complex as you want.

428
00:21:21,570 --> 00:21:25,360
We just did this for simplicity,
just to put boundaries

429
00:21:25,360 --> 00:21:27,620
around what are problem is
that we were considering,

430
00:21:27,620 --> 00:21:30,460
but you could definitely-- yeah,
you could definitely add that

431
00:21:30,460 --> 00:21:32,072
to the model if you wanted to.

432
00:21:32,072 --> 00:21:33,530
STUDENT 1: Cutting
the tree was not

433
00:21:33,530 --> 00:21:37,966
worrying about the output of
the actual cell in terms of

434
00:21:37,966 --> 00:21:40,964
whether or not some shading
was going to bring down

435
00:21:40,964 --> 00:21:41,880
the rest of the cell.

436
00:21:41,880 --> 00:21:43,254
The reason we
would cut the trees

437
00:21:43,254 --> 00:21:44,632
is to increase the maximum
amount of sunlight per day

438
00:21:44,632 --> 00:21:45,570
hitting the panel.

439
00:21:45,570 --> 00:21:49,007
And it's because we were looking
mostly at endpoints, trying

440
00:21:49,007 --> 00:21:50,971
to get the spectrum ends.

441
00:21:50,971 --> 00:21:53,426
That was why we went
to such extremes.

442
00:21:53,426 --> 00:21:55,697
Cutting down selective trees
and just parts of trees

443
00:21:55,697 --> 00:21:56,863
would be kind of in between.

444
00:21:56,863 --> 00:21:58,827
It's a little bit more
difficult to assess.

445
00:22:02,755 --> 00:22:05,210
STUDENT 2: Cutting down trees
is actually [INAUDIBLE].

446
00:22:05,210 --> 00:22:08,790
It's not [INAUDIBLE] because
we found that the shading

447
00:22:08,790 --> 00:22:12,140
factor doesn't play a much
bigger role if you look

448
00:22:12,140 --> 00:22:13,340
at the relative [INAUDIBLE].

449
00:22:13,340 --> 00:22:15,400
So you would as
well have increased

450
00:22:15,400 --> 00:22:21,308
the-- there will be objectively
small loss in the energy gain,

451
00:22:21,308 --> 00:22:23,748
but that shouldn't matter much.

452
00:22:27,164 --> 00:22:29,116
AUDIENCE: Sort of a
philosophical question.

453
00:22:29,116 --> 00:22:35,940
If the payback period in Phoenix
is 51 years, is it worth it?

454
00:22:35,940 --> 00:22:42,030
That's a long time period
for-- I guess economically

455
00:22:42,030 --> 00:22:44,767
you could say that you could do
other things with that capital

456
00:22:44,767 --> 00:22:47,202
instead that would have
a shorter payback period.

457
00:22:51,534 --> 00:22:53,200
JORDAN: Well, 51
years-- then the answer

458
00:22:53,200 --> 00:22:56,483
is probably not if you're
looking at the benefit of cost

459
00:22:56,483 --> 00:22:58,130
money-wise.

460
00:22:58,130 --> 00:23:01,880
We did analysis
about the energy.

461
00:23:01,880 --> 00:23:08,270
So we found an estimate for the
embedded energy of the panel.

462
00:23:08,270 --> 00:23:12,370
This is from-- I can't
remember the source,

463
00:23:12,370 --> 00:23:14,120
but they change quite a bit.

464
00:23:14,120 --> 00:23:17,880
But this example says it's
1,500 kilowatt hours per meter

465
00:23:17,880 --> 00:23:20,420
squared, so that equivalents
to nine years payback.

466
00:23:20,420 --> 00:23:22,570
Pretty much nine
years in Boston.

467
00:23:22,570 --> 00:23:25,990
I think it turned out to
be eight years in Phoenix.

468
00:23:25,990 --> 00:23:28,510
So there is a
benefit energy-wise,

469
00:23:28,510 --> 00:23:31,160
but in this example,
perhaps not cost-wise--

470
00:23:31,160 --> 00:23:36,009
perhaps not the most
advantageous to do per dollar.

471
00:23:36,009 --> 00:23:37,800
MARK WINKLER: A related
question, actually.

472
00:23:37,800 --> 00:23:39,909
Can you back to your
two slides back maybe?

473
00:23:39,909 --> 00:23:40,450
JORDAN: Yeah.

474
00:23:40,450 --> 00:23:42,060
MARK WINKLER: Your look
at the net energy gain

475
00:23:42,060 --> 00:23:43,010
was quite similar.

476
00:23:43,010 --> 00:23:46,950
So why the large difference
in yearly savings

477
00:23:46,950 --> 00:23:47,816
and payback period.

478
00:23:47,816 --> 00:23:50,840
STUDENT 3: Just the cost of
electricity in each location.

479
00:23:50,840 --> 00:23:55,640
We estimated it as about $0.07
per kilowatt hour in Phoenix

480
00:23:55,640 --> 00:24:01,350
and $0.17 in Boston
for residential.

481
00:24:01,350 --> 00:24:04,970
STUDENT 1: Which is why you
would get a shorter payback

482
00:24:04,970 --> 00:24:07,660
period for Boston-- is because
the cost of the electricity

483
00:24:07,660 --> 00:24:09,730
that you're [INAUDIBLE].

484
00:24:09,730 --> 00:24:13,430
MARK WINKLER: So I would have
assumed that the generation

485
00:24:13,430 --> 00:24:14,910
mix is sort of similar.

486
00:24:14,910 --> 00:24:17,090
Is that regulatory,
or-- I would assume

487
00:24:17,090 --> 00:24:20,854
they're coal/gas centric
generation mixes.

488
00:24:20,854 --> 00:24:21,520
STUDENT 3: Yeah.

489
00:24:26,760 --> 00:24:28,810
You mean in terms of
how the houses are--

490
00:24:28,810 --> 00:24:29,765
MARK WINKLER: This is a
little outside the scope

491
00:24:29,765 --> 00:24:30,350
of what you guys did.

492
00:24:30,350 --> 00:24:31,370
I was just curious
if you guys had

493
00:24:31,370 --> 00:24:33,125
any sense of why
the big difference

494
00:24:33,125 --> 00:24:36,380
in wholesale electricity
price is between Boston--

495
00:24:36,380 --> 00:24:40,890
STUDENT 3: I think part of it
is just how plentiful energy is.

496
00:24:40,890 --> 00:24:43,380
I guess Boston is at the very
end in the corner of the US.

497
00:24:43,380 --> 00:24:49,170
It's more difficult to get fuel,
oil, gas shipped over here.

498
00:24:49,170 --> 00:24:50,960
I think Arizona--
I think they're

499
00:24:50,960 --> 00:24:53,610
relatively close to a nuclear
power plant over there.

500
00:24:53,610 --> 00:24:57,290
Oil-- I think it's
just location--wise.

501
00:24:57,290 --> 00:24:59,040
JOE SULLIVAN: There's
a lot of coal there.

502
00:24:59,040 --> 00:25:01,210
They actually ship a
lot of the electricity

503
00:25:01,210 --> 00:25:04,045
to California because they
can't [INAUDIBLE] in California.

504
00:25:04,045 --> 00:25:05,170
One quick question, though.

505
00:25:05,170 --> 00:25:07,010
The relative energy gains
are the same for both.

506
00:25:07,010 --> 00:25:08,468
Do you have different
sized panels,

507
00:25:08,468 --> 00:25:10,600
or is heating that much?

508
00:25:10,600 --> 00:25:11,390
That's a big deal.

509
00:25:11,390 --> 00:25:12,530
STUDENT 3: Heating, yeah.

510
00:25:12,530 --> 00:25:12,731
JOE SULLIVAN: OK.

511
00:25:12,731 --> 00:25:13,294
So if you--

512
00:25:17,234 --> 00:25:18,234
HEIDI: Also for Phoenix.

513
00:25:20,820 --> 00:25:23,750
For Phoenix, you can see
this is the cooling over here

514
00:25:23,750 --> 00:25:27,000
and heating on the
right over there,

515
00:25:27,000 --> 00:25:29,160
and for Phoenix, you can
just look at the values

516
00:25:29,160 --> 00:25:32,410
and see that there's a lot more
cooling than heating compared

517
00:25:32,410 --> 00:25:35,700
to the Boston case, where
there's a lot more heating

518
00:25:35,700 --> 00:25:37,890
by many orders of
magnitude more.

519
00:25:37,890 --> 00:25:39,790
And so that kind
of balances it out.

520
00:25:39,790 --> 00:25:41,873
JOE SULLIVAN: And so that's
only in a cutting down

521
00:25:41,873 --> 00:25:44,590
a tree case if you were already
well shaded or not shaded

522
00:25:44,590 --> 00:25:46,080
at all?

523
00:25:46,080 --> 00:25:47,890
HEIDI: These are
actually all the curves

524
00:25:47,890 --> 00:25:49,077
for all the scenarios, and--

525
00:25:49,077 --> 00:25:49,910
JOE SULLIVAN: I see.

526
00:25:49,910 --> 00:25:51,420
OK.

527
00:25:51,420 --> 00:25:54,140
HEIDI: So it does matter
just because of location.

528
00:25:58,140 --> 00:26:01,140
So if you're not
cutting down a tree,

529
00:26:01,140 --> 00:26:07,671
then there's no
decrease in shading.

530
00:26:07,671 --> 00:26:09,920
Or is it just the panel
itself that's heating up more?

531
00:26:09,920 --> 00:26:12,440
STUDENT 3: Well, actually
if you go back to--

532
00:26:12,440 --> 00:26:14,610
JOE SULLIVAN: Sorry, I
think I missed something.

533
00:26:14,610 --> 00:26:17,210
STUDENT 3: Actually,
in Boston it actually

534
00:26:17,210 --> 00:26:20,600
makes more sense to have a
black roof than a white roof.

535
00:26:20,600 --> 00:26:24,140
So you actually want--
shading isn't necessarily

536
00:26:24,140 --> 00:26:28,176
good in Boston, just because
there's so much heating

537
00:26:28,176 --> 00:26:30,980
that you need in the winter.

538
00:26:30,980 --> 00:26:33,050
It seems to be the
dominant effect in Boston,

539
00:26:33,050 --> 00:26:34,730
and in Phoenix,
it's the opposite.

540
00:26:34,730 --> 00:26:37,709
The cooling is the
dominant effect.

541
00:26:37,709 --> 00:26:39,500
PROFESSOR: Did you
consider the possibility

542
00:26:39,500 --> 00:26:41,515
that snow also
insulates the house

543
00:26:41,515 --> 00:26:42,792
once it falls on the roof?

544
00:26:42,792 --> 00:26:43,375
STUDENT 3: No.

545
00:26:46,165 --> 00:26:47,570
We did not.

546
00:26:47,570 --> 00:26:51,020
I don't know if-- is
that built into-- I

547
00:26:51,020 --> 00:26:52,590
don't know what would happen.

548
00:26:52,590 --> 00:26:57,718
No, I don't think we did,
but that's a good point.

549
00:26:57,718 --> 00:26:59,429
AUDIENCE: Along
those lines, do you

550
00:26:59,429 --> 00:27:01,141
have any intuition
as to why in one case

551
00:27:01,141 --> 00:27:03,611
it's better to cut down a
tree other than remove it,

552
00:27:03,611 --> 00:27:06,194
and then the other is better to
remove it rather than just cut

553
00:27:06,194 --> 00:27:06,694
it down?

554
00:27:06,694 --> 00:27:08,720
Are you expecting it
to grow back and then

555
00:27:08,720 --> 00:27:10,921
have to incur more
costs because you're

556
00:27:10,921 --> 00:27:13,860
going to have to cut it down
again, or-- what's going on?

557
00:27:13,860 --> 00:27:17,260
STUDENT 3: Well, just
in terms of pure energy,

558
00:27:17,260 --> 00:27:20,530
it was very, very
slightly better

559
00:27:20,530 --> 00:27:26,480
in this case to have
the tree just cut

560
00:27:26,480 --> 00:27:31,131
just in terms of the balance
between heating and cooling

561
00:27:31,131 --> 00:27:31,630
[INAUDIBLE].

562
00:27:34,440 --> 00:27:36,580
STUDENT 1: I may be able
to help clarify that.

563
00:27:36,580 --> 00:27:39,530
The idea is in Boston
we're relatively cold most

564
00:27:39,530 --> 00:27:42,190
of the year, so the more
sunlight that hits your house

565
00:27:42,190 --> 00:27:44,390
is going to add more
heat to your house,

566
00:27:44,390 --> 00:27:47,760
and that's less energy
that you have to pay for.

567
00:27:47,760 --> 00:27:49,894
So the reason that it's
beneficial to cut down

568
00:27:49,894 --> 00:27:52,560
the tree completely in Boston is
because it allows more sunlight

569
00:27:52,560 --> 00:27:54,690
to hit your house, whereas
in Phoenix, you don't want

570
00:27:54,690 --> 00:27:55,720
the sunlight to hit your house.

571
00:27:55,720 --> 00:27:57,280
If you cut down the
tree completely,

572
00:27:57,280 --> 00:28:00,700
that's more you have to
pay for AC in the summer.

573
00:28:00,700 --> 00:28:01,253
So the--

574
00:28:01,253 --> 00:28:03,627
AUDIENCE: [INAUDIBLE] the
difference between cutting down

575
00:28:03,627 --> 00:28:06,080
the tree and removing it?

576
00:28:06,080 --> 00:28:07,280
STUDENT 1: So--

577
00:28:07,280 --> 00:28:08,470
STUDENT 3: So-- go ahead.

578
00:28:08,470 --> 00:28:09,610
STUDENT 1: Cutting
the tree is assuming

579
00:28:09,610 --> 00:28:11,901
that you're going to maintain
it at that certain level.

580
00:28:11,901 --> 00:28:14,349
So by cutting the tree,
you keep a certain amount

581
00:28:14,349 --> 00:28:16,140
of shading on the lower
part of your house,

582
00:28:16,140 --> 00:28:19,670
but you still allow sunlight
to hit your solar panel.

583
00:28:19,670 --> 00:28:21,520
Cutting down the
tree completely means

584
00:28:21,520 --> 00:28:23,220
there's no shading
on your house at all.

585
00:28:25,957 --> 00:28:28,040
AUDIENCE: I guess I have
a philosophical question.

586
00:28:28,040 --> 00:28:29,968
So I think there are
a lot of people--

587
00:28:29,968 --> 00:28:33,342
motivation for the solar panels
is not just the [INAUDIBLE],

588
00:28:33,342 --> 00:28:35,081
but rather the desire
to do something

589
00:28:35,081 --> 00:28:36,664
good for the
environment, and to lower

590
00:28:36,664 --> 00:28:38,098
carbon emissions, et cetera.

591
00:28:38,098 --> 00:28:42,995
But when you cut down trees,
that increases your carbon

592
00:28:42,995 --> 00:28:45,320
emission because you're
reducing the plant,

593
00:28:45,320 --> 00:28:47,190
that reduces your carbon output.

594
00:28:47,190 --> 00:28:52,610
So given that you won't have a
tree there for like 50 years,

595
00:28:52,610 --> 00:28:56,840
does that offset the
carbon emission gains

596
00:28:56,840 --> 00:29:00,070
that you get by--

597
00:29:00,070 --> 00:29:01,620
STUDENT 1: Just my two cents.

598
00:29:01,620 --> 00:29:04,250
If you really want
to go in depth,

599
00:29:04,250 --> 00:29:05,877
you can look at
how much carbon is

600
00:29:05,877 --> 00:29:07,710
going to be produced
by the coal power plant

601
00:29:07,710 --> 00:29:10,180
to give you the energy that
you're going to be using for 50

602
00:29:10,180 --> 00:29:12,096
years and compare that
to the amount of carbon

603
00:29:12,096 --> 00:29:15,010
that one tree was going to save
you, or you could ask yourself,

604
00:29:15,010 --> 00:29:18,070
am I planning to have a child
during those 50 years, which

605
00:29:18,070 --> 00:29:22,060
will produce so much more CO2
than that tree will take out?

606
00:29:22,060 --> 00:29:27,860
Either way, it's
relatively a small value.

607
00:29:27,860 --> 00:29:30,490
However, we do
acknowledge that there

608
00:29:30,490 --> 00:29:32,580
was a lot of
philosophical questions

609
00:29:32,580 --> 00:29:34,300
that we argued
amongst ourselves,

610
00:29:34,300 --> 00:29:37,182
but we realized we didn't have
the time to try to evaluate,

611
00:29:37,182 --> 00:29:39,595
or the materials, and scope.

612
00:29:39,595 --> 00:29:41,428
PROFESSOR: One more
question, and then we're

613
00:29:41,428 --> 00:29:42,719
going to have to switch groups.

614
00:29:42,719 --> 00:29:44,669
Jasmin?

615
00:29:44,669 --> 00:29:46,641
JASMIN HOFSTETTER: Do
you have any real data

616
00:29:46,641 --> 00:29:49,920
to compare your
model results to.

617
00:29:49,920 --> 00:29:52,250
From your results,
it seems that it

618
00:29:52,250 --> 00:29:54,410
doesn't make any
sense to install

619
00:29:54,410 --> 00:29:56,120
solar panels ins Phoenix.

620
00:29:56,120 --> 00:29:57,694
Is that right?

621
00:29:57,694 --> 00:29:59,610
That's the impression?

622
00:29:59,610 --> 00:30:01,100
STUDENT 3: Financially.

623
00:30:01,100 --> 00:30:02,970
Just purely financially, yeah.

624
00:30:07,490 --> 00:30:09,890
In terms of the PV
output, it seemed

625
00:30:09,890 --> 00:30:12,340
to be pretty close
in comparison to what

626
00:30:12,340 --> 00:30:14,610
we got from other sources.

627
00:30:14,610 --> 00:30:19,770
So it seems like the net gain
in energy is roughly right,

628
00:30:19,770 --> 00:30:22,490
but obviously people still
install panels there.

629
00:30:22,490 --> 00:30:24,860
So either, I'm guessing,
subsidies, or larger

630
00:30:24,860 --> 00:30:28,970
installations, or
something else, or just the

631
00:30:28,970 --> 00:30:31,290
desire to install it
just for installing it--

632
00:30:31,290 --> 00:30:34,393
not necessarily for financial
reason-- in Phoenix.

633
00:30:34,393 --> 00:30:37,665
JASMIN HOFSTETTER: What was the
temperature that you assumed?

634
00:30:37,665 --> 00:30:41,350
I suppose you assumed a constant
temperature in the house that

635
00:30:41,350 --> 00:30:42,220
was like the--

636
00:30:42,220 --> 00:30:44,290
STUDENT 3: Yes, that
would also change it.

637
00:30:44,290 --> 00:30:44,630
JASMIN HOFSTETTER: What
was this temperature?

638
00:30:44,630 --> 00:30:46,255
STUDENT 3: The set
points for our model

639
00:30:46,255 --> 00:30:47,805
was-- the cooling
set point was 71.

640
00:30:47,805 --> 00:30:49,052
HEIDI: 76.

641
00:30:49,052 --> 00:30:50,460
Yeah, 76.

642
00:30:50,460 --> 00:30:53,450
And then the heating
set point was 71.

643
00:30:53,450 --> 00:30:54,527
STUDENT 3: Yes.

644
00:30:54,527 --> 00:30:56,610
JASMIN HOFSTETTER: Can you
say that again, please?

645
00:30:56,610 --> 00:30:59,990
HEIDI: The cooling set point
was 76 degrees Fahrenheit,

646
00:30:59,990 --> 00:31:03,600
and the heating was 71.

647
00:31:03,600 --> 00:31:04,850
STUDENT 3: For both locations.

648
00:31:04,850 --> 00:31:06,022
HEIDI: For both locations.

649
00:31:06,022 --> 00:31:07,230
JASMIN HOFSTETTER: Thank you.

650
00:31:10,562 --> 00:31:11,990
AUDIENCE: [INAUDIBLE].

651
00:31:11,990 --> 00:31:13,300
JOE SULLIVAN: Are we out of--

652
00:31:13,300 --> 00:31:14,393
PROFESSOR: No, that's it.

653
00:31:14,393 --> 00:31:15,240
You guys are done.

654
00:31:15,240 --> 00:31:15,950
Congratulations.

655
00:31:15,950 --> 00:31:17,360
[APPLAUSE]

656
00:31:17,360 --> 00:31:19,991
[INAUDIBLE] coming up, PV grid.

657
00:31:19,991 --> 00:31:22,491
What happens when you install
loads, and oodles, and oodles,

658
00:31:22,491 --> 00:31:24,190
and oodles of solar
onto the grid?

659
00:31:24,190 --> 00:31:25,440
We're going to hear all about.

660
00:31:25,440 --> 00:31:27,431
And take it away.

661
00:31:27,431 --> 00:31:28,555
Knock it out of park, guys.

662
00:31:28,555 --> 00:31:30,055
IBRAHIM: So as
[? Tony ?] mentioned,

663
00:31:30,055 --> 00:31:31,600
we're the PV grid project.

664
00:31:31,600 --> 00:31:32,935
I'm Ibrahim.

665
00:31:32,935 --> 00:31:33,560
MARY: I'm Mary.

666
00:31:33,560 --> 00:31:34,462
RITA: I'm Rita.

667
00:31:34,462 --> 00:31:35,364
ASHLEY: I'm Ashley.

668
00:31:35,364 --> 00:31:36,139
JARED: I'm Jared.

669
00:31:36,139 --> 00:31:37,680
IBRAHIM: All right,
so I'm just going

670
00:31:37,680 --> 00:31:42,480
to start with the motivation
behind our project.

671
00:31:42,480 --> 00:31:46,830
So as we discussed in
class, PV installations

672
00:31:46,830 --> 00:31:49,330
have witnessed very
significant growth rates

673
00:31:49,330 --> 00:31:50,890
over the last few years.

674
00:31:50,890 --> 00:31:55,440
Last year alone PV installation
growth rates were around 17%.

675
00:31:55,440 --> 00:32:00,150
Around 18 gigawatts
globally were installed.

676
00:32:00,150 --> 00:32:04,240
As the cost of PV
approaches grid parity,

677
00:32:04,240 --> 00:32:06,870
more investors and
consumers are going

678
00:32:06,870 --> 00:32:11,070
to want to adopt PV systems.

679
00:32:11,070 --> 00:32:15,830
However, one lingering
or major obstacle

680
00:32:15,830 --> 00:32:20,620
preventing the further or
high penetration levels

681
00:32:20,620 --> 00:32:25,410
of PV systems is intermittency.

682
00:32:25,410 --> 00:32:27,530
So as we discussed
in class, there's

683
00:32:27,530 --> 00:32:30,700
variability in terms
of the solar resource,

684
00:32:30,700 --> 00:32:35,160
both on a long-term scale and
a short-term scale seconds

685
00:32:35,160 --> 00:32:37,060
to minutes.

686
00:32:37,060 --> 00:32:39,270
So on a long-term
scale, we're talking

687
00:32:39,270 --> 00:32:42,350
about the position of the
sun relative to the Earth

688
00:32:42,350 --> 00:32:43,480
and so on.

689
00:32:43,480 --> 00:32:47,420
So in that respect,
that's predictable

690
00:32:47,420 --> 00:32:49,010
and can be planned for.

691
00:32:49,010 --> 00:32:52,125
When we define or talk
about intermittency,

692
00:32:52,125 --> 00:32:55,980
it's the short-term
unpredictable effects

693
00:32:55,980 --> 00:33:03,060
that change the power
output significantly.

694
00:33:03,060 --> 00:33:08,060
So what we have here is
the fractional change

695
00:33:08,060 --> 00:33:11,970
in power output over
the course of one day.

696
00:33:11,970 --> 00:33:15,520
So as you can see, between
the two consecutive seconds,

697
00:33:15,520 --> 00:33:18,000
the power output
can almost double,

698
00:33:18,000 --> 00:33:21,870
and it can at other
times drop by half.

699
00:33:21,870 --> 00:33:24,730
So from a system
operator perspective,

700
00:33:24,730 --> 00:33:26,880
that's obviously
a major challenge

701
00:33:26,880 --> 00:33:31,940
because demand should
match supply at all times.

702
00:33:31,940 --> 00:33:36,110
So again, these effects, or
these intermittency issues,

703
00:33:36,110 --> 00:33:40,330
arise due to regional weather
patterns that can be predicted

704
00:33:40,330 --> 00:33:44,290
and also due to local
weather patterns that

705
00:33:44,290 --> 00:33:46,420
are less predictable.

706
00:33:46,420 --> 00:33:50,190
So in our project, what
we tried to address is,

707
00:33:50,190 --> 00:33:53,460
can the weather report be
used to predict the power

708
00:33:53,460 --> 00:33:57,920
output from an ensemble
of smaller distributive PV

709
00:33:57,920 --> 00:33:58,420
systems?

710
00:33:58,420 --> 00:34:04,240
That is, can we average out
these local less predictable

711
00:34:04,240 --> 00:34:07,680
intermittency effects?

712
00:34:07,680 --> 00:34:11,040
I'll give it to Mary to
discuss our approach.

713
00:34:24,810 --> 00:34:27,210
MARY: So our goal
of this project

714
00:34:27,210 --> 00:34:30,080
was to design a model
that could quantitatively

715
00:34:30,080 --> 00:34:34,620
analyze a PV grid and determine
its robustness in terms

716
00:34:34,620 --> 00:34:36,080
of variability.

717
00:34:36,080 --> 00:34:39,370
And our main components were
meantime between failure--

718
00:34:39,370 --> 00:34:43,540
which is the average
time between two system

719
00:34:43,540 --> 00:34:46,820
failures, which Rita
will define and discuss

720
00:34:46,820 --> 00:34:49,300
later-- number of
systems in the grid,

721
00:34:49,300 --> 00:34:51,530
and geographic
dispersion, which we

722
00:34:51,530 --> 00:34:53,670
measured through
geometric mean distance.

723
00:34:56,199 --> 00:35:00,420
Our data set was from
the Oahu airport,

724
00:35:00,420 --> 00:35:04,070
which is part of the National
Renewable Energy Laboratory.

725
00:35:04,070 --> 00:35:08,220
There are 17 systems all
within about a kilometer

726
00:35:08,220 --> 00:35:11,160
of each other, so it's a very
small, very dense system,

727
00:35:11,160 --> 00:35:17,200
but there was second interval
data for a year, which we used.

728
00:35:17,200 --> 00:35:19,370
So there's a fair amount
of data to give us

729
00:35:19,370 --> 00:35:21,600
an estimate of how
intermittency varies

730
00:35:21,600 --> 00:35:25,200
over the course of a system
and the number of systems

731
00:35:25,200 --> 00:35:25,800
and density.

732
00:35:37,590 --> 00:35:39,360
RITA: So our first
step was to define

733
00:35:39,360 --> 00:35:41,530
what was a PV system failure.

734
00:35:41,530 --> 00:35:44,150
In order to do so, we
accessed the CAISO website--

735
00:35:44,150 --> 00:35:46,950
that is, the California
Independent System Operator--

736
00:35:46,950 --> 00:35:52,080
and we took that data from
one week of the actual demand

737
00:35:52,080 --> 00:35:54,580
and hour hand demand forecast.

738
00:35:54,580 --> 00:35:56,750
They give this value
for every hour,

739
00:35:56,750 --> 00:35:59,050
so we took the value for
every hour of the week,

740
00:35:59,050 --> 00:36:01,460
and then we plotted
in this graph

741
00:36:01,460 --> 00:36:03,825
that we have a line for
each day of the week,

742
00:36:03,825 --> 00:36:07,650
and can see that both the
magnitudes and the shapes

743
00:36:07,650 --> 00:36:10,470
throughout the week
are almost the same.

744
00:36:10,470 --> 00:36:13,650
We can also see that the
values are almost all positive.

745
00:36:13,650 --> 00:36:16,720
This means that they
usually underestimate.

746
00:36:16,720 --> 00:36:18,250
They usually think
that the demand

747
00:36:18,250 --> 00:36:22,910
is going to be under
what it really happens.

748
00:36:22,910 --> 00:36:28,500
And so what we defined was
that, if this estimation is

749
00:36:28,500 --> 00:36:31,200
OK for CAISO, if
they can manage that

750
00:36:31,200 --> 00:36:34,040
the grid with this
variation, then they

751
00:36:34,040 --> 00:36:38,680
could also manage the grid with
this variation in a PV output.

752
00:36:38,680 --> 00:36:40,760
And so we looked at 5:00 PM.

753
00:36:40,760 --> 00:36:44,320
That is the hour that
we have the biggest

754
00:36:44,320 --> 00:36:47,800
variation between the two,
and we averaged the value,

755
00:36:47,800 --> 00:36:50,200
and we got to 6%.

756
00:36:50,200 --> 00:36:54,340
So this means that, if our
intermittency is above 6%,

757
00:36:54,340 --> 00:36:56,900
we are going to have
a PV system failure.

758
00:36:56,900 --> 00:37:01,720
If the variation is below
6%, then the intermittency

759
00:37:01,720 --> 00:37:04,072
is not going to be a failure.

760
00:37:04,072 --> 00:37:06,280
Then we could define the
mean time between failures--

761
00:37:06,280 --> 00:37:09,450
that is, the mean time between
two intermittencies higher

762
00:37:09,450 --> 00:37:09,950
than 6%.

763
00:37:21,880 --> 00:37:24,300
JARED: OK, now that we
have some context of what

764
00:37:24,300 --> 00:37:27,950
the problem is, and we have an
idea of what variability is,

765
00:37:27,950 --> 00:37:30,700
and we have a data
set to work with,

766
00:37:30,700 --> 00:37:33,690
I'm going to talk about how
we actually solve the problem.

767
00:37:33,690 --> 00:37:38,930
We use coding in MATLAB to
handle this huge data set.

768
00:37:38,930 --> 00:37:43,190
NREL had 17 systems out
there for every second

769
00:37:43,190 --> 00:37:44,880
of an entire year.

770
00:37:44,880 --> 00:37:48,930
And so we took all
the files from NREL

771
00:37:48,930 --> 00:37:51,240
and put them all
into one huge matrix.

772
00:37:51,240 --> 00:37:52,740
You can imagine it
was-- it ended up

773
00:37:52,740 --> 00:37:56,130
being about 23 fields
by several million,

774
00:37:56,130 --> 00:37:58,400
and it's about 677 megabytes.

775
00:37:58,400 --> 00:38:02,050
So actually handling the
data was an issue in itself.

776
00:38:02,050 --> 00:38:05,490
I don't recommend it
with an old computer.

777
00:38:05,490 --> 00:38:07,400
And we also the GPS
coordinates for each

778
00:38:07,400 --> 00:38:10,110
of those locations
and the variability

779
00:38:10,110 --> 00:38:12,530
from the California
ISO, so with that data

780
00:38:12,530 --> 00:38:16,740
we could begin to build
our code to figure out

781
00:38:16,740 --> 00:38:20,750
a quantitative description
of mean time between failure

782
00:38:20,750 --> 00:38:25,520
and our idea of density.

783
00:38:25,520 --> 00:38:28,060
So once everything was
loaded into one big matrix

784
00:38:28,060 --> 00:38:29,660
that we could work
with, we moved on

785
00:38:29,660 --> 00:38:31,570
to use the GPS coordinates.

786
00:38:31,570 --> 00:38:34,760
And of those 17
systems, we found

787
00:38:34,760 --> 00:38:38,260
every single combination
of 17 choose 2,

788
00:38:38,260 --> 00:38:41,780
17 choose 3-- every possible
way that you could connect

789
00:38:41,780 --> 00:38:44,070
these systems-- and
came up with something

790
00:38:44,070 --> 00:38:47,300
like 60,000 different
ways of connecting these,

791
00:38:47,300 --> 00:38:49,610
and then for each
possible connection,

792
00:38:49,610 --> 00:38:51,820
we had a function
that would calculate

793
00:38:51,820 --> 00:38:54,530
this geometric
mean distance that

794
00:38:54,530 --> 00:38:57,600
would give you an
idea of the density

795
00:38:57,600 --> 00:39:00,050
of that particular connection.

796
00:39:00,050 --> 00:39:03,710
And so to compare our
mean time between failure

797
00:39:03,710 --> 00:39:06,950
for these systems while
holding the density constant,

798
00:39:06,950 --> 00:39:10,440
we then searched through
those possible combinations

799
00:39:10,440 --> 00:39:13,350
and found this magic
number that kind of

800
00:39:13,350 --> 00:39:16,090
existed for each of those
possible combinations of two,

801
00:39:16,090 --> 00:39:18,030
of three, or four,
all the way through.

802
00:39:18,030 --> 00:39:20,710
And it kind of lined up
for geometric mean distance

803
00:39:20,710 --> 00:39:22,100
of 400 meters.

804
00:39:22,100 --> 00:39:25,660
So using that set,
we could then go on

805
00:39:25,660 --> 00:39:28,340
and see how increasing
the number of systems

806
00:39:28,340 --> 00:39:30,460
helped the mean time
between failure.

807
00:39:30,460 --> 00:39:34,421
And then for a given set,
we ended up using eight.

808
00:39:34,421 --> 00:39:39,160
17 choose 8 gave us like 24,000
possible ways to connect them.

809
00:39:39,160 --> 00:39:44,770
We searched through and
found varying densities

810
00:39:44,770 --> 00:39:46,980
for one set number.

811
00:39:46,980 --> 00:39:50,020
Then finally we
wrote a function that

812
00:39:50,020 --> 00:39:52,040
calculated the mean time
between failure that

813
00:39:52,040 --> 00:39:55,815
went through our data
from NREL and said--

814
00:39:55,815 --> 00:39:57,690
looked at the fractional
difference and said,

815
00:39:57,690 --> 00:40:00,820
OK, each time it's about
6%, that's a failure,

816
00:40:00,820 --> 00:40:03,270
and then measured that distance,
took the average of that,

817
00:40:03,270 --> 00:40:05,135
and that was our mean
time between failure.

818
00:40:05,135 --> 00:40:06,510
And then finally,
once we had all

819
00:40:06,510 --> 00:40:09,190
that together, we
crunched all the numbers,

820
00:40:09,190 --> 00:40:12,830
took a long time on my computer.

821
00:40:12,830 --> 00:40:15,990
We were able to plot it together
and get some very nice trends.

822
00:40:15,990 --> 00:40:18,410
One of the great things,
I think, about our code

823
00:40:18,410 --> 00:40:22,840
is that it was only 525
lines, and if you've ever

824
00:40:22,840 --> 00:40:24,720
built a programmer, a
big application, that's

825
00:40:24,720 --> 00:40:25,780
really small.

826
00:40:25,780 --> 00:40:28,610
It's very easy just to go in
and see exactly what's going on.

827
00:40:28,610 --> 00:40:29,890
So it's very flexible.

828
00:40:29,890 --> 00:40:31,600
We could hand it off
to another company,

829
00:40:31,600 --> 00:40:33,900
to another research group,
and they go in and adapt

830
00:40:33,900 --> 00:40:35,800
it to just about any data set.

831
00:40:35,800 --> 00:40:39,880
If you are able to get data in
California, or from Germany,

832
00:40:39,880 --> 00:40:42,950
or from somewhere else, and
bring it into our format,

833
00:40:42,950 --> 00:40:46,240
it's very easy just to plug
it in and run the data.

834
00:40:46,240 --> 00:40:49,930
Very, very minimal
changes within our code.

835
00:40:49,930 --> 00:40:52,280
And then you could
also build on our code

836
00:40:52,280 --> 00:40:53,680
to look at other problems.

837
00:40:53,680 --> 00:40:57,420
So we have the change in the--
we have the variability data

838
00:40:57,420 --> 00:40:58,794
as a function of time.

839
00:40:58,794 --> 00:41:00,960
We also have the solar
output as a function of time.

840
00:41:00,960 --> 00:41:03,580
So you could conceivably
go in and figure out

841
00:41:03,580 --> 00:41:05,460
how your meantime
between failure changes

842
00:41:05,460 --> 00:41:10,522
based on the time of day and
change your critical percentage

843
00:41:10,522 --> 00:41:11,980
based on the time
of day, and there

844
00:41:11,980 --> 00:41:14,640
are several other problems that
you could go, and launch off

845
00:41:14,640 --> 00:41:17,660
of our code, and continue on.

846
00:41:17,660 --> 00:41:20,060
And if you're interested
at all, I actually

847
00:41:20,060 --> 00:41:22,660
put the code of my public space.

848
00:41:22,660 --> 00:41:23,989
There's the link there.

849
00:41:23,989 --> 00:41:24,530
Check it out.

850
00:41:24,530 --> 00:41:26,490
It's pretty cool.

851
00:41:26,490 --> 00:41:29,410
And then Ashley is going
to talk about our results.

852
00:41:37,300 --> 00:41:43,240
ASHLEY: Cool, so the first
thing that we did in order

853
00:41:43,240 --> 00:41:47,550
to try to see the trends in
these huge fields of data

854
00:41:47,550 --> 00:41:49,010
was just to plot the data.

855
00:41:49,010 --> 00:41:50,494
It was actually a
much bigger task

856
00:41:50,494 --> 00:41:51,910
than I thought it
was going to be.

857
00:41:54,590 --> 00:41:57,190
The plot on the left is for
one day's worth of data,

858
00:41:57,190 --> 00:42:01,930
and the plot on the right
is one week's worth of data.

859
00:42:01,930 --> 00:42:04,880
The y-axis is power density
in watts per square meter,

860
00:42:04,880 --> 00:42:08,280
and the x-axis is
the time in seconds.

861
00:42:08,280 --> 00:42:11,370
The blue is all 17 of
our systems together,

862
00:42:11,370 --> 00:42:13,820
and the red is just
for one system.

863
00:42:13,820 --> 00:42:17,766
So as you would assume,
the power output for all 17

864
00:42:17,766 --> 00:42:19,140
together is clearly
a lot greater

865
00:42:19,140 --> 00:42:21,250
than the output from
just one system,

866
00:42:21,250 --> 00:42:24,300
but this give us a sense of
being able to see fluctuations

867
00:42:24,300 --> 00:42:26,770
within one day, and
also were able to see

868
00:42:26,770 --> 00:42:29,770
when the sun rose, and peaked,
and also fell each day.

869
00:42:33,370 --> 00:42:36,680
And in order to quantify all
those different fluctuations,

870
00:42:36,680 --> 00:42:41,680
we did the fractional change
in power density versus time,

871
00:42:41,680 --> 00:42:43,900
once again, for one day,
and then for one week.

872
00:42:43,900 --> 00:42:46,030
And red is the one system.

873
00:42:46,030 --> 00:42:48,680
Blue is all 17 systems
together, and we can already

874
00:42:48,680 --> 00:42:53,090
see just from plotting the
data that having all 17

875
00:42:53,090 --> 00:42:55,330
systems together does
start to average out

876
00:42:55,330 --> 00:42:58,480
the fluctuations of
individual systems

877
00:42:58,480 --> 00:42:59,833
by a significant amount.

878
00:43:04,270 --> 00:43:07,310
So then Rita earlier
mentioned that we use 6%

879
00:43:07,310 --> 00:43:09,870
as our cut off for failure.

880
00:43:09,870 --> 00:43:12,920
We actually went ahead and
did 6%, 12%, and 18% just

881
00:43:12,920 --> 00:43:16,110
to see how sensitive our
analysis was to that threshold

882
00:43:16,110 --> 00:43:17,640
value.

883
00:43:17,640 --> 00:43:19,504
So here we have
plotted on the left

884
00:43:19,504 --> 00:43:21,920
the meantime between failure
versus the number of systems,

885
00:43:21,920 --> 00:43:23,790
and on the right,
meantime between failure

886
00:43:23,790 --> 00:43:26,690
versus the geometric
mean distance.

887
00:43:26,690 --> 00:43:29,611
I also calculated
these values for using

888
00:43:29,611 --> 00:43:31,610
a week's worth of data,
a month's worth of data,

889
00:43:31,610 --> 00:43:33,330
and a year's worth of data.

890
00:43:33,330 --> 00:43:35,900
So the week would give
you more fluctuations,

891
00:43:35,900 --> 00:43:38,820
but the year would give you the
more long-term overall system

892
00:43:38,820 --> 00:43:39,849
behavior.

893
00:43:39,849 --> 00:43:42,390
Relationship between the mean
time between failure and number

894
00:43:42,390 --> 00:43:45,430
of systems is quadratic, and
we found a linear relationship

895
00:43:45,430 --> 00:43:49,200
between the mean time
between failure and the GMD.

896
00:43:49,200 --> 00:43:52,720
So this is four 6% cut off.

897
00:43:52,720 --> 00:43:58,910
This is for 12% cut off,
and this is for 18% cut off.

898
00:43:58,910 --> 00:44:01,930
And the mean time
between failure

899
00:44:01,930 --> 00:44:04,590
increases dramatically
as you go from the 6% cut

900
00:44:04,590 --> 00:44:07,224
off to the 18% cut off.

901
00:44:07,224 --> 00:44:08,890
So a lot of this makes
sense, but it was

902
00:44:08,890 --> 00:44:10,098
really cool to quantify that.

903
00:44:24,920 --> 00:44:28,580
RITA: So after
applying those graphs

904
00:44:28,580 --> 00:44:32,400
we could take our conclusions
and answer our question.

905
00:44:32,400 --> 00:44:36,840
And so the first thing that we
noticed, but we were expecting,

906
00:44:36,840 --> 00:44:38,530
is that a big data
sample should be

907
00:44:38,530 --> 00:44:42,520
used if conclusions are going
to be used as a design tool.

908
00:44:42,520 --> 00:44:46,260
As Ashley said, we used for
a week, a month, and a year.

909
00:44:46,260 --> 00:44:51,380
And so we know that the
bigger the data set,

910
00:44:51,380 --> 00:44:53,480
it's going to be--
it's not going

911
00:44:53,480 --> 00:44:56,000
to be influenced by
abnormal things that

912
00:44:56,000 --> 00:44:58,600
can happen in a given day.

913
00:44:58,600 --> 00:45:00,920
And we also saw that
there is a linear relation

914
00:45:00,920 --> 00:45:04,470
between mean time
between failure and GMD.

915
00:45:04,470 --> 00:45:08,200
When GMD increases-- that
is, when density decreases--

916
00:45:08,200 --> 00:45:10,880
we are going to have an increase
in mean time between failure.

917
00:45:10,880 --> 00:45:12,950
This was also what
we were expecting

918
00:45:12,950 --> 00:45:15,480
because the local
effects will not affect

919
00:45:15,480 --> 00:45:17,820
systems that are further apart.

920
00:45:17,820 --> 00:45:20,340
We also saw that there
was a quadratic relation

921
00:45:20,340 --> 00:45:23,640
between mean time between
failure and number of systems.

922
00:45:23,640 --> 00:45:25,050
Number of systems increased.

923
00:45:25,050 --> 00:45:27,710
Mean time between
failures also increased.

924
00:45:27,710 --> 00:45:31,330
This was also according
to what we expected

925
00:45:31,330 --> 00:45:36,010
because we know that the
percentage and the total output

926
00:45:36,010 --> 00:45:39,290
is going to be lower.

927
00:45:39,290 --> 00:45:43,980
We also saw that the mean time
between failure is very low,

928
00:45:43,980 --> 00:45:46,570
even when we can see
the 17 systems together,

929
00:45:46,570 --> 00:45:50,540
we have about 900
seconds between failures.

930
00:45:50,540 --> 00:45:52,820
This means that
some backup systems

931
00:45:52,820 --> 00:45:55,610
should be used in order
to take over the load

932
00:45:55,610 --> 00:45:57,270
when we have a failure.

933
00:45:57,270 --> 00:45:59,030
And so now we're
running conditions

934
00:45:59,030 --> 00:46:00,630
to ask our first question.

935
00:46:00,630 --> 00:46:02,720
And so we conclude
that localized

936
00:46:02,720 --> 00:46:07,010
predictable intermittency
do average out

937
00:46:07,010 --> 00:46:09,860
and that this effect decreases
as the number of systems

938
00:46:09,860 --> 00:46:12,370
and the GMD increase.

939
00:46:12,370 --> 00:46:15,580
The data that we used
was for 17 systems,

940
00:46:15,580 --> 00:46:19,130
and the biggest distance
between them was one kilometer.

941
00:46:19,130 --> 00:46:21,490
So we believe that
it's important to run

942
00:46:21,490 --> 00:46:28,090
our code for a bigger set of
data, because only in this way

943
00:46:28,090 --> 00:46:31,660
we can confirm our
conclusions and guidelines

944
00:46:31,660 --> 00:46:34,920
for the design of PV
systems can be defined.

945
00:46:34,920 --> 00:46:37,538
Thank you, and we'll be happy
to answer your questions.

946
00:46:37,538 --> 00:46:40,526
[APPLAUSE]

947
00:46:44,385 --> 00:46:45,760
JOE SULLIVAN: So
a couple things.

948
00:46:45,760 --> 00:46:47,676
First of all, you ended
at exactly 15 minutes.

949
00:46:47,676 --> 00:46:49,956
I find that remarkable.

950
00:46:49,956 --> 00:46:51,080
Additionally, just-- sorry.

951
00:46:51,080 --> 00:46:54,030
Can you repeat what exactly
a failure mode is defined as?

952
00:46:54,030 --> 00:46:56,270
Are you looking at 6%
intermittency varying

953
00:46:56,270 --> 00:46:57,250
from second to second?

954
00:46:57,250 --> 00:46:59,624
So if you look at the output
from one second to the next,

955
00:46:59,624 --> 00:47:00,880
does that change by over 6%?

956
00:47:00,880 --> 00:47:01,419
RITA: Mm-hm.

957
00:47:01,419 --> 00:47:03,460
JOE SULLIVAN: It wasn't
average out over an hour.

958
00:47:03,460 --> 00:47:04,270
RITA: No, no.

959
00:47:04,270 --> 00:47:05,270
It was second by second.

960
00:47:05,270 --> 00:47:08,490
JOE SULLIVAN: You got
the 6% from Cal ISO.

961
00:47:08,490 --> 00:47:15,486
RITA: We said that if there--
in a given hour, we measured--

962
00:47:15,486 --> 00:47:17,158
let me just--

963
00:47:17,158 --> 00:47:19,350
JARED: They only had an
hour of data [INAUDIBLE].

964
00:47:19,350 --> 00:47:21,220
RITA: Yeah, they only
gave hourly data.

965
00:47:21,220 --> 00:47:25,750
So the difference
between the actual demand

966
00:47:25,750 --> 00:47:28,140
and the hour-ahead
demand forecast.

967
00:47:28,140 --> 00:47:32,812
So this is what they are
expecting, but the difference

968
00:47:32,812 --> 00:47:35,270
between what they are expecting
and what the grid is really

969
00:47:35,270 --> 00:47:36,630
asking them.

970
00:47:36,630 --> 00:47:41,050
So if they can manage this
difference on a second base,

971
00:47:41,050 --> 00:47:43,772
they can also manage this
difference on the PV grid.

972
00:47:43,772 --> 00:47:45,480
JOE SULLIVAN: So you
took the worst case.

973
00:47:45,480 --> 00:47:46,850
Is that how you got 6%?

974
00:47:46,850 --> 00:47:49,040
RITA: Yeah, we took the
average of the worst case.

975
00:47:49,040 --> 00:47:50,430
It's the 5:00 PM.

976
00:47:50,430 --> 00:47:52,740
The 5:00 PM is always
the worst hour.

977
00:47:52,740 --> 00:47:54,690
It's always when
they have that peak.

978
00:47:54,690 --> 00:47:57,195
And in fact, all of the
base-- almost all of the base

979
00:47:57,195 --> 00:47:58,780
were around 6%.

980
00:47:58,780 --> 00:48:04,397
Our peak was like 6.8%, and
we averaged, and it was 6%.

981
00:48:04,397 --> 00:48:05,855
MARK WINKLER: So
that's essentially

982
00:48:05,855 --> 00:48:08,350
their peaking capacity?

983
00:48:08,350 --> 00:48:09,470
RITA: Yeah.

984
00:48:09,470 --> 00:48:12,180
ASHLEY: Also, so I
actually wrote down

985
00:48:12,180 --> 00:48:16,370
the numbers for
6% function, 18%--

986
00:48:16,370 --> 00:48:17,880
like our mean time
between failure.

987
00:48:17,880 --> 00:48:24,280
For 6%, we had up to 15
minutes between failure.

988
00:48:24,280 --> 00:48:30,010
So it's a pretty low amount
of time between failures.

989
00:48:30,010 --> 00:48:33,180
And if you allow 12%
as your intermittency,

990
00:48:33,180 --> 00:48:35,290
you can get up to
about half a day.

991
00:48:35,290 --> 00:48:39,730
And then for 18% as your cut
off, you get about nine days.

992
00:48:39,730 --> 00:48:42,675
So it is still
very intermittent,

993
00:48:42,675 --> 00:48:45,420
and you would pretty often have
to have backup systems if you

994
00:48:45,420 --> 00:48:47,460
had the small of a system.

995
00:48:47,460 --> 00:48:52,350
So if you had a much wider
spread system and a lot more

996
00:48:52,350 --> 00:48:54,730
systems in your grid,
then you could definitely

997
00:48:54,730 --> 00:48:58,126
significantly increase the
mean time between failure.

998
00:48:58,126 --> 00:48:59,606
Yeah, Joe?

999
00:48:59,606 --> 00:49:01,480
JOE SULLIVAN: So you
have this awesome graph.

1000
00:49:01,480 --> 00:49:04,090
So if you go back to
the time between failure

1001
00:49:04,090 --> 00:49:06,850
number of systems.

1002
00:49:06,850 --> 00:49:09,730
The interesting takeaway
is how large of an area

1003
00:49:09,730 --> 00:49:12,290
do you have average over, right?

1004
00:49:12,290 --> 00:49:14,580
So 300 seconds on
a grid perspective

1005
00:49:14,580 --> 00:49:17,720
is unacceptable for
widespread PV developed point.

1006
00:49:17,720 --> 00:49:19,770
We need to be on
the order of years.

1007
00:49:19,770 --> 00:49:22,209
And so do you have an idea
of what that distance is?

1008
00:49:22,209 --> 00:49:23,792
ASHLEY: If we just
extrapolate it out?

1009
00:49:23,792 --> 00:49:25,958
JOE SULLIVAN: If you
extrapolate-- this is obviously

1010
00:49:25,958 --> 00:49:29,537
like we're taking the very, very
edge of that function and then

1011
00:49:29,537 --> 00:49:30,620
extrapolating [INAUDIBLE].

1012
00:49:30,620 --> 00:49:31,690
ASHLEY: So looking
at the numbers--

1013
00:49:31,690 --> 00:49:33,450
JOE SULLIVAN: But it looks like
it's going up exponentially,

1014
00:49:33,450 --> 00:49:35,900
or do you have an idea
of what that trend is?

1015
00:49:35,900 --> 00:49:37,830
ASHLEY: For 6% for
the one year, it

1016
00:49:37,830 --> 00:49:39,705
was almost exactly x squared.

1017
00:49:39,705 --> 00:49:44,142
It was like x squared plus
50, or 100, or whatever

1018
00:49:44,142 --> 00:49:47,535
that would be.

1019
00:49:47,535 --> 00:49:50,400
So if you want, you
could say, mean time

1020
00:49:50,400 --> 00:49:54,750
how many seconds are in a
year equals number squared.

1021
00:49:54,750 --> 00:49:57,350
So the square root of however
many number of seconds

1022
00:49:57,350 --> 00:49:59,070
there are in a
year would give you

1023
00:49:59,070 --> 00:50:02,666
your number of systems required
for a year between failures.

1024
00:50:02,666 --> 00:50:05,040
IBRAHIM: But this is for a
given geometric mean distance,

1025
00:50:05,040 --> 00:50:06,950
so you have two factors.

1026
00:50:06,950 --> 00:50:09,330
If you sort of
spread them out more,

1027
00:50:09,330 --> 00:50:12,580
probably going to
require less systems.

1028
00:50:12,580 --> 00:50:14,902
JARED: And if you
looked that map,

1029
00:50:14,902 --> 00:50:17,530
that's all at the end of a
runway at the Honolulu airport.

1030
00:50:17,530 --> 00:50:20,570
So if you have a huge
field in Arizona,

1031
00:50:20,570 --> 00:50:24,940
thousands of systems, your
mean time between failure

1032
00:50:24,940 --> 00:50:26,260
is going to be a lot better.

1033
00:50:26,260 --> 00:50:28,718
MARK WINKLER: So I'm actually
really surprised that there's

1034
00:50:28,718 --> 00:50:30,630
such a huge effect
from adding systems,

1035
00:50:30,630 --> 00:50:34,130
just because it seems as
though the relevant length

1036
00:50:34,130 --> 00:50:36,650
scales for weather
should be very large.

1037
00:50:36,650 --> 00:50:39,177
Do you guys say
anything about that?

1038
00:50:39,177 --> 00:50:41,510
JARED: I think the idea was
that, for long-term weather,

1039
00:50:41,510 --> 00:50:42,360
you can predict that.

1040
00:50:42,360 --> 00:50:43,776
So if you know
there's going to be

1041
00:50:43,776 --> 00:50:46,250
a storm front coming through,
you can add natural gas.

1042
00:50:46,250 --> 00:50:47,594
You can add coal to the system.

1043
00:50:47,594 --> 00:50:48,094
Back up--

1044
00:50:48,094 --> 00:50:50,052
ASHLEY: And that would
cover the entire system.

1045
00:50:50,052 --> 00:50:52,650
JARED: Our kind of variability
we're talking about

1046
00:50:52,650 --> 00:50:55,879
is say, if one cloud goes
over, or a flock of birds,

1047
00:50:55,879 --> 00:50:56,420
or something.

1048
00:50:56,420 --> 00:50:58,630
So we were thinking
that would be

1049
00:50:58,630 --> 00:51:00,760
on a few seconds
for a single module

1050
00:51:00,760 --> 00:51:04,490
for a cloud just go over
shade it for a short distance.

1051
00:51:04,490 --> 00:51:06,630
So if you add
thousands of modules,

1052
00:51:06,630 --> 00:51:09,744
the other modules wouldn't
be shaded while that one is.

1053
00:51:09,744 --> 00:51:11,160
MARK WINKLER: But
these fields-- I

1054
00:51:11,160 --> 00:51:16,380
mean, 100 meters on the
scale of cloud cover,

1055
00:51:16,380 --> 00:51:19,300
this still seems like a
somewhat small length scale.

1056
00:51:19,300 --> 00:51:20,940
Let me rephrase the question.

1057
00:51:20,940 --> 00:51:22,870
Do you think that the
graph on the right

1058
00:51:22,870 --> 00:51:24,970
would be a smooth
function of distance,

1059
00:51:24,970 --> 00:51:26,428
or do you think
there's some length

1060
00:51:26,428 --> 00:51:28,870
scale at which the
behavior on that plot

1061
00:51:28,870 --> 00:51:30,130
changes significantly?

1062
00:51:33,136 --> 00:51:35,260
JARED: That would be
interesting if we could find--

1063
00:51:35,260 --> 00:51:37,230
ASHLEY: The assumption
is definitely

1064
00:51:37,230 --> 00:51:41,842
evenly dispersed in an area.

1065
00:51:41,842 --> 00:51:44,300
JARED: That would be something
that, if we had another data

1066
00:51:44,300 --> 00:51:46,250
set that had wider
distances, it would

1067
00:51:46,250 --> 00:51:48,262
be very easy to plug it in.

1068
00:51:48,262 --> 00:51:49,720
I think our code's
really flexible.

1069
00:51:49,720 --> 00:51:51,860
It would show us
that relationship.

1070
00:51:51,860 --> 00:51:53,780
MARK WINKLER: What do
you guys think, though?

1071
00:51:53,780 --> 00:51:55,372
JARED: It's a good question.

1072
00:51:55,372 --> 00:52:00,380
ASHLEY: I wouldn't be surprised
if it was linear still.

1073
00:52:00,380 --> 00:52:02,400
I guess another
complexity we could do

1074
00:52:02,400 --> 00:52:06,040
would be you would
have-- right now we just

1075
00:52:06,040 --> 00:52:09,180
have one big field of
systems, but if you

1076
00:52:09,180 --> 00:52:13,620
had one set of systems that
was spaced x distances apart,

1077
00:52:13,620 --> 00:52:16,380
and then you had some
number of kilometers

1078
00:52:16,380 --> 00:52:19,160
away from another
one space-- I'm not

1079
00:52:19,160 --> 00:52:21,130
sure how exactly
would model that,

1080
00:52:21,130 --> 00:52:22,784
but I think that
at that point I'm

1081
00:52:22,784 --> 00:52:24,450
not sure what the
curve would look like,

1082
00:52:24,450 --> 00:52:30,912
but a continuing linear
trend seems reasonable to me.

1083
00:52:30,912 --> 00:52:33,798
IBRAHIM: So I guess another
thing to keep in mind

1084
00:52:33,798 --> 00:52:36,680
is we did not take into
account transmission costs,

1085
00:52:36,680 --> 00:52:41,280
so I guess you'd have to weigh
the cost of failure versus, I

1086
00:52:41,280 --> 00:52:44,610
guess, the added incurred
cost for transmission lines

1087
00:52:44,610 --> 00:52:47,940
and so on, so there's
sort of an optimum point

1088
00:52:47,940 --> 00:52:50,090
where you want to
space them and have

1089
00:52:50,090 --> 00:52:53,580
a certain number of
systems where I guess,

1090
00:52:53,580 --> 00:52:57,580
after a certain point, your
returns diminish and are not

1091
00:52:57,580 --> 00:53:00,840
equal to, I guess,
the cost of failure.

1092
00:53:00,840 --> 00:53:02,660
So that's something
where, I guess,

1093
00:53:02,660 --> 00:53:04,892
future people can
come in and expand on.

1094
00:53:09,820 --> 00:53:13,860
AUDIENCE: So all this
is data from Hawaii,

1095
00:53:13,860 --> 00:53:20,370
which has a very notable
climate and weather.

1096
00:53:20,370 --> 00:53:22,800
I've never been there, but--

1097
00:53:22,800 --> 00:53:25,230
[LAUGHTER]

1098
00:53:25,230 --> 00:53:29,750
Do you think that this is
really-- your code is flexible,

1099
00:53:29,750 --> 00:53:33,680
so I understand that, but do
you think the conclusions are

1100
00:53:33,680 --> 00:53:38,810
really extensible to
other parts of the world

1101
00:53:38,810 --> 00:53:42,884
with different weather
patterns or climate?

1102
00:53:42,884 --> 00:53:47,250
RITA: That's why we think
that the future mark is really

1103
00:53:47,250 --> 00:53:51,570
to do it for a different place
and for a bigger set of date

1104
00:53:51,570 --> 00:53:55,914
because we really want to be
sure that the conclusions are

1105
00:53:55,914 --> 00:54:03,420
going to be applicable, because
we had that same question.

1106
00:54:03,420 --> 00:54:06,085
We were talking just
about a small place.

1107
00:54:06,085 --> 00:54:09,550
We said that it's one kilometer
apart for the distance

1108
00:54:09,550 --> 00:54:10,810
that we have.

1109
00:54:10,810 --> 00:54:14,516
So we also want to run
for a bigger set of data

1110
00:54:14,516 --> 00:54:19,849
and for another place just to
be sure that our conclusions are

1111
00:54:19,849 --> 00:54:22,320
applicable everywhere.

1112
00:54:22,320 --> 00:54:25,120
JARED: And I would say the
relationship would probably

1113
00:54:25,120 --> 00:54:27,870
hold because if you have-- say,
if your regional weather is

1114
00:54:27,870 --> 00:54:30,220
very different, that
wouldn't show up

1115
00:54:30,220 --> 00:54:33,586
in the fractional second to
second difference that we had.

1116
00:54:33,586 --> 00:54:37,580
And so the timescale
that we measured it on I

1117
00:54:37,580 --> 00:54:42,140
think would be, say, small
clouds or intermittent events

1118
00:54:42,140 --> 00:54:45,850
that would occur over a wide
range of different climates.

1119
00:54:45,850 --> 00:54:49,120
The general regional
weather is predictable,

1120
00:54:49,120 --> 00:54:53,820
and it isn't investigated
in our study at all.

1121
00:54:53,820 --> 00:54:57,182
So I would say I think
the relation would hold.

1122
00:54:57,182 --> 00:54:58,640
ASHLEY: I think
that the big change

1123
00:54:58,640 --> 00:55:00,990
between different regions
would just be the total output

1124
00:55:00,990 --> 00:55:03,670
power that you can
get, but I think--

1125
00:55:03,670 --> 00:55:07,990
I wouldn't be surprised if the
fluctuation is still the same

1126
00:55:07,990 --> 00:55:09,360
or is similar.

1127
00:55:09,360 --> 00:55:11,864
And I think certainly that, as
you increase a number systems

1128
00:55:11,864 --> 00:55:14,280
and as you decrease the density
with which they're packed,

1129
00:55:14,280 --> 00:55:16,320
you should be able to
have a more robust grid.

1130
00:55:16,320 --> 00:55:19,804
I would be very surprised
if that weren't the case.

1131
00:55:19,804 --> 00:55:23,710
AUDIENCE: Do you see shading
for planes at the airport?

1132
00:55:27,030 --> 00:55:29,140
ASHLEY: There's no way
for us to determine

1133
00:55:29,140 --> 00:55:30,690
what causes the shading.

1134
00:55:30,690 --> 00:55:33,188
The raw data we
have is just output.

1135
00:55:33,188 --> 00:55:34,854
JOE SULLIVAN: Can you
see how they move?

1136
00:55:34,854 --> 00:55:35,752
[LAUGHTER]

1137
00:55:35,752 --> 00:55:37,335
ASHLEY: It's like
there is this line--

1138
00:55:37,335 --> 00:55:38,010
IBRAHIM: We actually
did that for one plot.

1139
00:55:38,010 --> 00:55:40,400
You could see the cloud
moving around the plot.

1140
00:55:40,400 --> 00:55:40,910
JOE SULLIVAN: That's cool.

1141
00:55:40,910 --> 00:55:42,680
IBRAHIM: And you see the
power output for [INAUDIBLE].

1142
00:55:42,680 --> 00:55:44,740
ASHLEY: Yeah, it was
on the order of-- we

1143
00:55:44,740 --> 00:55:48,540
had like two billion
data points, I think,

1144
00:55:48,540 --> 00:55:51,990
which was overwhelming.

1145
00:55:51,990 --> 00:55:53,940
But yeah, it was really cool.

1146
00:55:53,940 --> 00:55:54,905
Any more questions?

1147
00:55:54,905 --> 00:55:55,405
Yeah?

1148
00:55:58,550 --> 00:56:00,976
AUDIENCE: Can you describe
a little more what these PV

1149
00:56:00,976 --> 00:56:02,425
system failures entail?

1150
00:56:02,425 --> 00:56:04,234
And what happens,
and how long does it

1151
00:56:04,234 --> 00:56:05,806
take to get them
back up and running?

1152
00:56:05,806 --> 00:56:07,738
What has to be done to do that?

1153
00:56:07,738 --> 00:56:08,221
ASHLEY: You wanna get that one?

1154
00:56:08,221 --> 00:56:08,721
JARED: Sure.

1155
00:56:08,721 --> 00:56:11,445
So basically there
is a certain capacity

1156
00:56:11,445 --> 00:56:13,490
that the grid would have.

1157
00:56:13,490 --> 00:56:19,612
Say, you can compensate for a
6% drop in this case, or a 20%

1158
00:56:19,612 --> 00:56:20,820
drop, or something like that.

1159
00:56:20,820 --> 00:56:23,470
So if your system is
completely powered by PV,

1160
00:56:23,470 --> 00:56:26,700
which is not realistic,
and you have, say,

1161
00:56:26,700 --> 00:56:28,960
a 20% drop and
nothing to compensate

1162
00:56:28,960 --> 00:56:31,400
that, you have a blackout.

1163
00:56:31,400 --> 00:56:34,470
And so we investigated
18%, for instance.

1164
00:56:34,470 --> 00:56:37,020
So that would be,
say, if your grid

1165
00:56:37,020 --> 00:56:39,290
is a certain percentage
of PV and then has

1166
00:56:39,290 --> 00:56:41,140
natural gas, or
coal, or something

1167
00:56:41,140 --> 00:56:43,840
that you can bring
online quickly

1168
00:56:43,840 --> 00:56:46,310
to compensate a drop in PV.

1169
00:56:46,310 --> 00:56:50,780
That would be an idea
of what a failure is--

1170
00:56:50,780 --> 00:56:54,332
if you aren't able to
compensate that fluctuation

1171
00:56:54,332 --> 00:56:58,390
AUDIENCE: And how
long [INAUDIBLE]?

1172
00:56:58,390 --> 00:57:01,390
JARED: How long
would a failure last?

1173
00:57:01,390 --> 00:57:03,672
It depends.

1174
00:57:03,672 --> 00:57:05,110
If you can't meet the demand--

1175
00:57:05,110 --> 00:57:07,070
AUDIENCE: [INAUDIBLE].

1176
00:57:07,070 --> 00:57:09,340
JARED: For a PV system,
I think the problem is

1177
00:57:09,340 --> 00:57:12,170
the PV system would come back up
right after the cloud was over,

1178
00:57:12,170 --> 00:57:14,800
but if you can't
meet power demand,

1179
00:57:14,800 --> 00:57:17,340
you've got all kind
of protection systems

1180
00:57:17,340 --> 00:57:20,360
that would trip off,
and it would be mess.

1181
00:57:20,360 --> 00:57:24,361
So I don't think would
come back very quickly.

1182
00:57:24,361 --> 00:57:25,985
ASHLEY: That's a good
question, though.

1183
00:57:25,985 --> 00:57:27,720
JARED: That's a good question.

1184
00:57:27,720 --> 00:57:30,515
PROFESSOR: It's relevant because
you can envision back up power

1185
00:57:30,515 --> 00:57:33,390
that could kick in really
quick, but exhaust itself

1186
00:57:33,390 --> 00:57:37,040
within the period of
the delta t necessary.

1187
00:57:37,040 --> 00:57:39,750
AUDIENCE: For your
definition of intermittency,

1188
00:57:39,750 --> 00:57:43,170
did you look at the absolute
value or just the drop?

1189
00:57:43,170 --> 00:57:47,806
Because the grid can't deal with
excess power as well, and so

1190
00:57:47,806 --> 00:57:50,674
I was just wondering if
you had insight on that.

1191
00:57:50,674 --> 00:57:53,815
Like if you dumped 60%
more power in the demand,

1192
00:57:53,815 --> 00:57:54,940
there's no way for you to--

1193
00:57:54,940 --> 00:57:56,980
JARED: We did the
absolute value.

1194
00:57:56,980 --> 00:57:59,810
So 6% more, 6% less.

1195
00:58:03,650 --> 00:58:06,070
JASMIN HOFSTETTER: So I'm
going to ask you for real data.

1196
00:58:06,070 --> 00:58:08,760
So do you know where
more or less data points

1197
00:58:08,760 --> 00:58:14,050
would lie for, let's say,
PV systems on houses that

1198
00:58:14,050 --> 00:58:17,730
are like-- with a
typical distance

1199
00:58:17,730 --> 00:58:19,634
in some kind of neighborhood.

1200
00:58:22,580 --> 00:58:24,350
JARED: I think that
would just be you

1201
00:58:24,350 --> 00:58:26,710
would adjust your
geometric mean distance

1202
00:58:26,710 --> 00:58:30,690
to whatever the distance
from the houses are.

1203
00:58:30,690 --> 00:58:33,510
I don't think our data has to
be a solar farm, for instance.

1204
00:58:33,510 --> 00:58:36,000
I think it could be houses in
a neighborhood, for instance.

1205
00:58:36,000 --> 00:58:39,970
So if they're perfectly
connected to the grid,

1206
00:58:39,970 --> 00:58:42,314
I think that our code
would account for that.

1207
00:58:42,314 --> 00:58:43,730
ASHLEY: This was
for eight, right?

1208
00:58:43,730 --> 00:58:45,250
JARED: Uh-huh.

1209
00:58:45,250 --> 00:58:47,820
ASHLEY: The right-hand graphic
held the number of systems

1210
00:58:47,820 --> 00:58:48,393
at eight.

1211
00:58:48,393 --> 00:58:51,370
And so if you had eight
houses spread apart

1212
00:58:51,370 --> 00:58:56,260
by an average of 150
meters, then you would--

1213
00:58:56,260 --> 00:59:01,620
and if you considered a year's
worth of data-- is it like 250?

1214
00:59:01,620 --> 00:59:03,480
I just can't see it.

1215
00:59:03,480 --> 00:59:08,660
So you'd have meantime between
failure of 250 seconds, which

1216
00:59:08,660 --> 00:59:12,426
is four minutes?

1217
00:59:12,426 --> 00:59:13,530
Doing math under pressure.

1218
00:59:13,530 --> 00:59:15,890
JARED: Right, but if you
have a grid to back that up,

1219
00:59:15,890 --> 00:59:18,697
it's not big of a deal.

1220
00:59:18,697 --> 00:59:21,030
AUDIENCE: I'm confused about
the plot on the right here.

1221
00:59:21,030 --> 00:59:24,000
What it's suggesting is
that one week you picked

1222
00:59:24,000 --> 00:59:26,972
was significantly below the
year average [INAUDIBLE],

1223
00:59:26,972 --> 00:59:29,096
and you could have equally
picked another week that

1224
00:59:29,096 --> 00:59:29,450
was significantly above.

1225
00:59:29,450 --> 00:59:30,150
JARED: Right.

1226
00:59:30,150 --> 00:59:33,197
This was, I think,
just to give the trend.

1227
00:59:33,197 --> 00:59:35,530
The relationship between the
day, and a week, and a year

1228
00:59:35,530 --> 00:59:37,295
is just the day
that we-- I'm sorry.

1229
00:59:37,295 --> 00:59:39,753
A week, and a month, and a year
is just the week we picked,

1230
00:59:39,753 --> 00:59:40,640
the month we picked.

1231
00:59:40,640 --> 00:59:42,490
I think you see on
some of the other plots

1232
00:59:42,490 --> 00:59:44,360
that the week and the
month actually shift.

1233
00:59:44,360 --> 00:59:47,065
It's just the year was kind
of the average of those.

1234
00:59:50,040 --> 00:59:53,690
MARK WINKLER: I would
assume that areas,

1235
00:59:53,690 --> 00:59:58,080
or specifically countries, that
made large investments in solar

1236
00:59:58,080 --> 01:00:01,880
would have studied this
question in a detailed fashion.

1237
01:00:01,880 --> 01:00:04,950
Do you know if, for example,
Germany or Spain have looked

1238
01:00:04,950 --> 01:00:07,814
at this problem when it's spread
across hundreds of kilometers.

1239
01:00:07,814 --> 01:00:09,630
ASHLEY: Ibrahim, do
you know that one?

1240
01:00:09,630 --> 01:00:10,546
I think you might be--

1241
01:00:10,546 --> 01:00:12,600
IBRAHIM: I was actually
very-- we didn't find

1242
01:00:12,600 --> 01:00:14,350
a lot of literature actually.

1243
01:00:14,350 --> 01:00:16,820
For wind, there was a
lot of data out there,

1244
01:00:16,820 --> 01:00:20,920
I guess, because the high
penetration levels with PV.

1245
01:00:20,920 --> 01:00:24,760
There were very few studies.

1246
01:00:24,760 --> 01:00:27,110
Most of them actually
were addressing the US.

1247
01:00:27,110 --> 01:00:29,810
I didn't find any, actually,
on Germany or Spain.

1248
01:00:29,810 --> 01:00:31,690
Probably maybe they're
in Spanish or German,

1249
01:00:31,690 --> 01:00:32,670
so I don't know.

1250
01:00:36,100 --> 01:00:37,600
JOE SULLIVAN: So
what I find really

1251
01:00:37,600 --> 01:00:39,370
startling is that,
for a given system,

1252
01:00:39,370 --> 01:00:41,840
the time between failure
of the 6% intermittency

1253
01:00:41,840 --> 01:00:43,660
is on the order of a minute.

1254
01:00:43,660 --> 01:00:46,900
Do you have any the idea-- is
that vastly different for wind

1255
01:00:46,900 --> 01:00:48,300
and what that number is?

1256
01:00:48,300 --> 01:00:49,927
And this is outside
of your-- I'm

1257
01:00:49,927 --> 01:00:51,885
just wondering if in your
literature searching.

1258
01:00:51,885 --> 01:00:52,675
JARED: You probably
know the most about it.

1259
01:00:52,675 --> 01:00:53,500
ASHLEY: You would know from it.

1260
01:00:53,500 --> 01:00:54,360
JOE SULLIVAN: It seems like
you have this big rotor.

1261
01:00:54,360 --> 01:00:56,850
There's some momentum, and
that to slow that thing down

1262
01:00:56,850 --> 01:01:00,109
requires more time, but
I don't-- as opposed

1263
01:01:00,109 --> 01:01:00,650
to electrons.

1264
01:01:00,650 --> 01:01:03,317
JARED: I would say
wind would definitely

1265
01:01:03,317 --> 01:01:05,400
have a much longer time
scale than solar, I think.

1266
01:01:05,400 --> 01:01:08,197
There's a lot of momentum there.

1267
01:01:08,197 --> 01:01:12,109
RITA: But when wind stops,
the times that you have

1268
01:01:12,109 --> 01:01:14,046
intermittency is going
to be much bigger.

1269
01:01:14,046 --> 01:01:15,420
And there'd be
backup systems you

1270
01:01:15,420 --> 01:01:19,100
need to have to take over
for a long period of time.

1271
01:01:19,100 --> 01:01:20,595
JARED: And maybe
in high winds you

1272
01:01:20,595 --> 01:01:23,500
would have more of an issue,
because if the turbine is

1273
01:01:23,500 --> 01:01:25,500
spinning too fast, you
actually have to stop it.

1274
01:01:25,500 --> 01:01:28,960
So maybe there you'd run
into issues of variability

1275
01:01:28,960 --> 01:01:30,600
on the order of minutes.

1276
01:01:37,320 --> 01:01:39,048
AUDIENCE: So I
think-- and this is

1277
01:01:39,048 --> 01:01:42,350
kind of going back to location
data set-- comes from Hawaii,

1278
01:01:42,350 --> 01:01:46,251
which I would imagine has
mostly direct sunlight.

1279
01:01:46,251 --> 01:01:49,017
For locations such as
Boston, would the data

1280
01:01:49,017 --> 01:01:52,490
set change for,
say, diffuse light

1281
01:01:52,490 --> 01:01:57,902
and would that generally bring
in panels closer together

1282
01:01:57,902 --> 01:02:00,977
or require more panels at
the same geometric distance

1283
01:02:00,977 --> 01:02:03,970
to get the same results?

1284
01:02:03,970 --> 01:02:08,450
ASHLEY: Well, the raw data
that we have doesn't separate

1285
01:02:08,450 --> 01:02:12,440
direct and diffuse, so I think
that the first thing would

1286
01:02:12,440 --> 01:02:16,240
be we'd want to
look at a data set

1287
01:02:16,240 --> 01:02:19,400
and from whatever
other location you

1288
01:02:19,400 --> 01:02:24,100
wanted to know about and look at
how diffuse and direct differs.

1289
01:02:24,100 --> 01:02:28,470
I don't think we have a
sense here of that effect.

1290
01:02:28,470 --> 01:02:31,170
Does that answer your
question to some degree?

1291
01:02:31,170 --> 01:02:32,086
AUDIENCE: Some degree.

1292
01:02:32,086 --> 01:02:34,044
I don't know if someone
else wants to add more.

1293
01:02:34,044 --> 01:02:35,930
JOE SULLIVAN: [INAUDIBLE]
after you respond.

1294
01:02:35,930 --> 01:02:37,410
JARED: I think it's
interesting-- I was just

1295
01:02:37,410 --> 01:02:37,970
thinking about this.

1296
01:02:37,970 --> 01:02:40,053
Something that might be
interesting to investigate

1297
01:02:40,053 --> 01:02:42,080
is concentrated solar.

1298
01:02:42,080 --> 01:02:44,650
If it's easier to
shade, it would

1299
01:02:44,650 --> 01:02:46,770
look like a denser system.

1300
01:02:46,770 --> 01:02:49,310
So maybe that would be-- maybe
a concentrated solar farm

1301
01:02:49,310 --> 01:02:51,920
might be a bad idea if you
have lots of little clouds.

1302
01:02:51,920 --> 01:02:54,890
So that's something I
think that you could

1303
01:02:54,890 --> 01:02:56,354
expand into from this project.

1304
01:02:56,354 --> 01:02:58,770
IBRAHIM: And another thing, I
guess, to add to your point,

1305
01:02:58,770 --> 01:03:01,350
if you look at, I guess,
solar thermal systems

1306
01:03:01,350 --> 01:03:04,310
probably because of the diffuse
sunlight, the intermittency I

1307
01:03:04,310 --> 01:03:06,620
would expect is going
to be probably less.

1308
01:03:06,620 --> 01:03:08,870
You're going to have less,
or the mean time to failure

1309
01:03:08,870 --> 01:03:12,080
is going to be longer,
so you could maybe

1310
01:03:12,080 --> 01:03:15,700
add a solar thermal system, sort
of balance the power output,

1311
01:03:15,700 --> 01:03:18,150
and decrease your
intermittency even further.

1312
01:03:22,390 --> 01:03:23,816
JOE SULLIVAN: Any
last questions?

1313
01:03:23,816 --> 01:03:24,668
No?

1314
01:03:24,668 --> 01:03:26,890
All right, let's
thank our group.

1315
01:03:26,890 --> 01:03:30,240
[APPLAUSE]