1
00:00:00,120 --> 00:00:02,460
The following content is
provided under a Creative

2
00:00:02,460 --> 00:00:03,880
Commons license.

3
00:00:03,880 --> 00:00:06,090
Your support will help
MIT OpenCourseWare

4
00:00:06,090 --> 00:00:10,180
continue to offer high quality
educational resources for free.

5
00:00:10,180 --> 00:00:12,720
To make a donation or to
view additional materials

6
00:00:12,720 --> 00:00:16,650
from hundreds of MIT courses,
visit MIT OpenCourseWare

7
00:00:16,650 --> 00:00:17,880
at ocw.mit.edu.

8
00:00:25,370 --> 00:00:28,140
PROFESSOR: You have to
make some design decisions.

9
00:00:28,140 --> 00:00:31,100
And then once you've made
those design decisions,

10
00:00:31,100 --> 00:00:34,250
then you can actually write your
level 2, level 3 requirements

11
00:00:34,250 --> 00:00:37,530
and so forth, it's
an iterative process.

12
00:00:37,530 --> 00:00:40,280
But when you look at
the classic v model,

13
00:00:40,280 --> 00:00:42,570
you don't really see
those iterations.

14
00:00:42,570 --> 00:00:44,220
But they actually exist.

15
00:00:44,220 --> 00:00:46,970
So what we typically do
at the system requirements

16
00:00:46,970 --> 00:00:50,780
review is we review and
then agree on the high level

17
00:00:50,780 --> 00:00:53,750
requirements, level
0, level 1, maybe

18
00:00:53,750 --> 00:00:56,780
level 2 requirement, but
not the lower level ones,

19
00:00:56,780 --> 00:00:58,610
because you can't,
because you typically

20
00:00:58,610 --> 00:01:00,398
haven't done the design yet.

21
00:01:03,990 --> 00:01:06,460
The other thing I want
to say is a lot of the--

22
00:01:06,460 --> 00:01:08,280
this is really a critical issue.

23
00:01:08,280 --> 00:01:10,630
And a lot of the problems
we talked about last time

24
00:01:10,630 --> 00:01:14,450
in the schedule, cost overruns,
really, a lot of these things

25
00:01:14,450 --> 00:01:16,730
are traceable back
to the requirements.

26
00:01:16,730 --> 00:01:19,670
Either the requirements
were over ambitious.

27
00:01:19,670 --> 00:01:23,330
You can set requirements that
were essentially unachievable.

28
00:01:23,330 --> 00:01:27,839
Or you missed requirements that
actually became clear later on.

29
00:01:27,839 --> 00:01:28,880
But they weren't written.

30
00:01:28,880 --> 00:01:31,310
So nobody paid
attention to them.

31
00:01:31,310 --> 00:01:32,350
A couple of examples.

32
00:01:32,350 --> 00:01:36,230
So I have sort of one
example of core requirements,

33
00:01:36,230 --> 00:01:39,090
and then a good example.

34
00:01:39,090 --> 00:01:44,050
So this is a mission that
you may have heard about.

35
00:01:49,100 --> 00:01:50,570
A 1998 MCO.

36
00:01:50,570 --> 00:01:52,280
So who was born in--

37
00:01:52,280 --> 00:01:53,920
I'm looking at you guys.

38
00:01:53,920 --> 00:01:55,380
Justin, right?

39
00:01:55,380 --> 00:01:58,390
Justice, when were you born?

40
00:01:58,390 --> 00:01:59,050
'94?

41
00:01:59,050 --> 00:02:03,030
OK, so you're older
than MCO by four years.

42
00:02:03,030 --> 00:02:06,580
So MCO was a very well-known
just about not quite

43
00:02:06,580 --> 00:02:11,070
20 years ago, a very well known
mission failure that happened.

44
00:02:11,070 --> 00:02:16,550
A Mars climate orbiter, it
was launched on December 1998.

45
00:02:16,550 --> 00:02:20,560
And it had multiple functions
study the Martian climate,

46
00:02:20,560 --> 00:02:22,300
weather and surface changes.

47
00:02:22,300 --> 00:02:26,380
And it was to act as a relay
satellite for the Mars polar

48
00:02:26,380 --> 00:02:29,750
Lander, which came after
it and also failed.

49
00:02:29,750 --> 00:02:32,320
Both failed.

50
00:02:32,320 --> 00:02:35,320
And unfortunately MCO burned up.

51
00:02:35,320 --> 00:02:39,310
We think contact was
lost with a spacecraft

52
00:02:39,310 --> 00:02:43,370
after it basically entered
the Martian atmosphere.

53
00:02:43,370 --> 00:02:47,170
And you know nothing-- there
was no more communication.

54
00:02:47,170 --> 00:02:49,120
So we don't know for sure.

55
00:02:49,120 --> 00:02:53,194
But their best guess is that
the spacecraft burned up

56
00:02:53,194 --> 00:02:58,580
in the approach to Mars, that
the actual altitude at which

57
00:02:58,580 --> 00:03:00,170
had entered the
Martian atmosphere

58
00:03:00,170 --> 00:03:04,420
was about 57 kilometers.

59
00:03:04,420 --> 00:03:07,190
And it was supposed
to be about 220.

60
00:03:07,190 --> 00:03:09,740
So even though the Martian
atmosphere is very thin,

61
00:03:09,740 --> 00:03:12,920
when you enter at
very high velocities,

62
00:03:12,920 --> 00:03:16,060
that's not a good thing.

63
00:03:16,060 --> 00:03:17,300
This is the very famous--

64
00:03:17,300 --> 00:03:20,170
you've heard about this--
confusion of units problem,

65
00:03:20,170 --> 00:03:25,710
like this mission that one
part of the team used SI units,

66
00:03:25,710 --> 00:03:29,890
specifically Newton's seconds,
this is momentum, right?

67
00:03:29,890 --> 00:03:34,610
Force times time, that
gives you momentum.

68
00:03:34,610 --> 00:03:38,900
So when you burn your engines
for x number of seconds

69
00:03:38,900 --> 00:03:41,780
the sum of the product of
those two is your momentum.

70
00:03:41,780 --> 00:03:43,160
And then another
part of the team

71
00:03:43,160 --> 00:03:49,150
used English units, pounds
of force times second.

72
00:03:49,150 --> 00:03:55,215
OK, and so that was
fundamentally the problem that

73
00:03:55,215 --> 00:03:56,860
caused the burn up.

74
00:03:56,860 --> 00:03:59,710
Now if you read the
accident report.

75
00:03:59,710 --> 00:04:02,620
And I have a link here
to the accident report,

76
00:04:02,620 --> 00:04:06,490
you will notice that the
requirements were actually OK.

77
00:04:06,490 --> 00:04:09,980
The requirements were
written correctly.

78
00:04:09,980 --> 00:04:12,910
So specifically, there was a
document called the software

79
00:04:12,910 --> 00:04:17,110
interface specification, SIS,
that specified everything

80
00:04:17,110 --> 00:04:20,880
should be in SI units
on the ground segment

81
00:04:20,880 --> 00:04:22,570
and on the space segment.

82
00:04:22,570 --> 00:04:27,520
But the problem is that this
was not checked and implemented.

83
00:04:27,520 --> 00:04:30,730
And so here's a quote
from the accident report.

84
00:04:30,730 --> 00:04:33,220
Items that the mission
assurance manager, which

85
00:04:33,220 --> 00:04:35,000
didn't exist for this mission.

86
00:04:35,000 --> 00:04:37,720
So the mission
Assurance Manager role

87
00:04:37,720 --> 00:04:40,760
is to make sure the requirements
are actually followed.

88
00:04:40,760 --> 00:04:43,960
There was no mission Assurance
Manager on the MCO mission.

89
00:04:43,960 --> 00:04:48,070
Included ensuring that
the AMD file, which

90
00:04:48,070 --> 00:04:52,480
is one of the files put out
by the propulsion system,

91
00:04:52,480 --> 00:04:55,120
met the requirements
of the softwares

92
00:04:55,120 --> 00:04:58,050
in the interface specification
And that did not happen.

93
00:04:58,050 --> 00:05:02,440
And so that was sort of
the root cause there.

94
00:05:02,440 --> 00:05:04,940
Here's a good example
for requirements.

95
00:05:04,940 --> 00:05:07,270
This is a little bit older.

96
00:05:07,270 --> 00:05:09,200
This is the DC-3.

97
00:05:09,200 --> 00:05:11,750
This is an old plane.

98
00:05:11,750 --> 00:05:17,790
First flight in December 1935.

99
00:05:17,790 --> 00:05:21,190
This was this was
really an airplane that

100
00:05:21,190 --> 00:05:25,420
kickstarted civil aviation
as a commercial service.

101
00:05:25,420 --> 00:05:29,290
And so first of
all, the DC-3 was

102
00:05:29,290 --> 00:05:33,820
based on an earlier
evolution of the DC-2.

103
00:05:33,820 --> 00:05:38,740
And I first saw this when
I worked in St. Louis.

104
00:05:38,740 --> 00:05:40,870
There was a little museum
at the headquarters.

105
00:05:40,870 --> 00:05:44,820
And they had like a, like
a nice plaque on the wall

106
00:05:44,820 --> 00:05:49,030
with actually a replica
of the contract.

107
00:05:49,030 --> 00:05:53,620
You know, the original letter
that left the DC-3 contract.

108
00:05:53,620 --> 00:05:56,830
And the story is that
this-- these requirements--

109
00:05:56,830 --> 00:05:57,510
these are--

110
00:05:57,510 --> 00:06:01,110
this is level 0 requirements,
what's written here,

111
00:06:01,110 --> 00:06:03,630
was hammered out
in a single quote,

112
00:06:03,630 --> 00:06:07,930
end quote marathon phone call
between Smith and Douglas.

113
00:06:07,930 --> 00:06:11,290
So Smith was, at that
time, the head of what

114
00:06:11,290 --> 00:06:13,600
was today essentially
United Airlines.

115
00:06:13,600 --> 00:06:16,630
And Douglas, who by the way, is
a graduate of this department

116
00:06:16,630 --> 00:06:21,820
here at MIT, Donald Douglas,
was the head of the Douglas

117
00:06:21,820 --> 00:06:23,260
Aircraft Company.

118
00:06:23,260 --> 00:06:27,040
And here are the requirements:
range about 1,000 miles.

119
00:06:27,040 --> 00:06:31,070
So this means that you don't
have a trans-continental range,

120
00:06:31,070 --> 00:06:31,570
right?

121
00:06:31,570 --> 00:06:33,730
If you want transcontinental
range in the US,

122
00:06:33,730 --> 00:06:35,080
you need about how many miles?

123
00:06:38,480 --> 00:06:43,080
Like Boston to San
Diego, Seattle to Miami.

124
00:06:43,080 --> 00:06:43,850
3,000?

125
00:06:43,850 --> 00:06:45,680
Yeah, reserves.

126
00:06:45,680 --> 00:06:47,420
Yeah, 3,000 miles.

127
00:06:47,420 --> 00:06:50,970
So this means that, if you go
transcontinental with the DC-3,

128
00:06:50,970 --> 00:06:53,460
you've got to refuel
twice, basically.

129
00:06:53,460 --> 00:06:55,360
But you know, that was OK.

130
00:06:55,360 --> 00:06:58,620
Today, we wouldn't-- we
probably wouldn't accept it.

131
00:06:58,620 --> 00:07:00,830
But that's OK back then.

132
00:07:00,830 --> 00:07:03,670
Cruise speed about 150 miles.

133
00:07:03,670 --> 00:07:07,130
20 to 30 passengers, depending
on the configuration.

134
00:07:07,130 --> 00:07:08,630
Twin engines.

135
00:07:08,630 --> 00:07:10,970
And then something that
we would say today.

136
00:07:10,970 --> 00:07:14,420
That's a fuzzy-- that's a not
a well written requirement,

137
00:07:14,420 --> 00:07:15,680
because it's fuzzy.

138
00:07:15,680 --> 00:07:19,440
But it said, it should be
a rugged and economical.

139
00:07:19,440 --> 00:07:21,710
And based on this sort of
high level requirements

140
00:07:21,710 --> 00:07:24,320
that were pretty
clearly defined,

141
00:07:24,320 --> 00:07:26,420
the airplane was
then designed, built,

142
00:07:26,420 --> 00:07:28,710
and very, very successful.

143
00:07:28,710 --> 00:07:34,560
And over 10,000 copies of
this airplane were built.

144
00:07:34,560 --> 00:07:38,870
So what happened since then
is the requirements explosion.

145
00:07:38,870 --> 00:07:42,080
This is a chart from
a document called

146
00:07:42,080 --> 00:07:46,160
the technical multi-disciplinary
design optimization

147
00:07:46,160 --> 00:07:48,290
white paper.

148
00:07:48,290 --> 00:07:51,500
And it's a little dated,
but the basic message here

149
00:07:51,500 --> 00:07:54,950
is pretty stunning if
you think about it.

150
00:07:54,950 --> 00:07:57,920
So first, heavier
than air flight,

151
00:07:57,920 --> 00:08:00,620
1903, the Wright
brothers, right?

152
00:08:00,620 --> 00:08:02,240
1G flight.

153
00:08:02,240 --> 00:08:06,530
What is the requirement
for 1G flight?

154
00:08:06,530 --> 00:08:07,940
Get off the ground.

155
00:08:07,940 --> 00:08:09,200
Get off the ground.

156
00:08:09,200 --> 00:08:12,470
And stay off the ground for
at least a few seconds, right?

157
00:08:12,470 --> 00:08:15,440
Not just ballistic.

158
00:08:15,440 --> 00:08:16,610
That's the requirement.

159
00:08:16,610 --> 00:08:17,840
Get off the ground.

160
00:08:17,840 --> 00:08:19,610
Heavier than air flight.

161
00:08:19,610 --> 00:08:20,900
Yeah, big success.

162
00:08:20,900 --> 00:08:22,440
Big milestone.

163
00:08:22,440 --> 00:08:27,380
Well, pretty clear after
not too long, it was well--

164
00:08:27,380 --> 00:08:29,054
we also want to turn.

165
00:08:29,054 --> 00:08:30,470
We don't want to
just fly straight

166
00:08:30,470 --> 00:08:36,510
on the beach and land again, so
maneuvering, gust acceleration.

167
00:08:36,510 --> 00:08:38,280
There are winds buffeting.

168
00:08:38,280 --> 00:08:39,840
You've got to
handle those gusts.

169
00:08:39,840 --> 00:08:42,090
So that that's important.

170
00:08:42,090 --> 00:08:44,810
And then what were the
first airplanes made of?

171
00:08:44,810 --> 00:08:47,050
Let's see an EPFL, guys.

172
00:08:47,050 --> 00:08:48,350
What's the first airplanes?

173
00:08:48,350 --> 00:08:50,874
What materials did they use?

174
00:08:50,874 --> 00:08:52,862
[STATIC]

175
00:08:52,862 --> 00:08:53,856
[INAUDIBLE]

176
00:08:53,856 --> 00:08:54,850
AUDIENCE: Wood.

177
00:08:54,850 --> 00:08:56,970
PROFESSOR: Wood, yeah.

178
00:08:56,970 --> 00:08:57,470
What else?

179
00:08:57,470 --> 00:08:58,178
AUDIENCE: Canvas.

180
00:08:58,178 --> 00:08:59,050
PROFESSOR: Canvas.

181
00:08:59,050 --> 00:09:01,730
AUDIENCE: [INAUDIBLE]

182
00:09:01,730 --> 00:09:02,630
PROFESSOR: Exactly.

183
00:09:02,630 --> 00:09:06,590
so then very soon after that,
maybe in the 20s and 30s,

184
00:09:06,590 --> 00:09:08,180
we started using metals, right?

185
00:09:08,180 --> 00:09:09,576
Metallic structures.

186
00:09:09,576 --> 00:09:11,450
And the big issue there,
especially if you're

187
00:09:11,450 --> 00:09:15,010
flying close to the
ocean and salt water,

188
00:09:15,010 --> 00:09:16,840
is corrosion control.

189
00:09:16,840 --> 00:09:18,740
Your airplanes can't rust.

190
00:09:18,740 --> 00:09:22,010
And so a corrosion
control became important.

191
00:09:22,010 --> 00:09:25,580
And pressurization,
what is that all about?

192
00:09:29,240 --> 00:09:31,020
Higher, faster.

193
00:09:31,020 --> 00:09:35,070
So at some point, you need to
pressurize the cabin, right?

194
00:09:35,070 --> 00:09:36,690
That's a new requirement.

195
00:09:36,690 --> 00:09:38,980
The Wright brothers
never thought about that.

196
00:09:38,980 --> 00:09:40,980
Well, they didn't-- maybe
they thought about it.

197
00:09:40,980 --> 00:09:43,900
But they certainly didn't have
to pressurize their Wright

198
00:09:43,900 --> 00:09:44,640
flyer.

199
00:09:44,640 --> 00:09:49,200
So you see you see what's
happening is we get greedy.

200
00:09:49,200 --> 00:09:50,520
You know, we were excited.

201
00:09:50,520 --> 00:09:51,580
We can fly now.

202
00:09:51,580 --> 00:09:52,930
But now we want to go higher.

203
00:09:52,930 --> 00:09:54,490
We want to stay longer.

204
00:09:54,490 --> 00:09:56,100
So we get greedy.

205
00:09:56,100 --> 00:09:59,050
And we go to more extreme
environments and so forth.

206
00:09:59,050 --> 00:10:04,230
And as we do that, it get
harvested do and more and more

207
00:10:04,230 --> 00:10:06,630
requirements start piling up.

208
00:10:06,630 --> 00:10:10,305
And you see a big step
here, World War II, right?

209
00:10:10,305 --> 00:10:12,990
Handling quality,
radar transparency.

210
00:10:12,990 --> 00:10:14,670
Radar was invented.

211
00:10:14,670 --> 00:10:16,290
So now you want to
have airplanes that

212
00:10:16,290 --> 00:10:19,880
are not visible on the radar.

213
00:10:19,880 --> 00:10:22,900
Fatigue, rough field landing.

214
00:10:22,900 --> 00:10:25,860
And then we have another big
step here in the 60s and 70s.

215
00:10:25,860 --> 00:10:30,300
This was during the Cold
War, smart weapons, nuclear,

216
00:10:30,300 --> 00:10:31,560
fly by wire, right?

217
00:10:31,560 --> 00:10:35,680
Replacing cables with
electronic flight controls.

218
00:10:35,680 --> 00:10:39,060
And then a lot of the
"ilitys" in the last 20 years,

219
00:10:39,060 --> 00:10:43,370
reducibility, affordability,
portability, et cetera.

220
00:10:43,370 --> 00:10:47,250
So if you actually look
at the requirements set

221
00:10:47,250 --> 00:10:49,200
for the new generation
of airplanes,

222
00:10:49,200 --> 00:10:52,960
whether commercial or
military, it's overwhelming.

223
00:10:52,960 --> 00:10:55,920
I mean, it's thousands and
thousands of requirements,

224
00:10:55,920 --> 00:10:58,440
because we've gotten greedy.

225
00:10:58,440 --> 00:10:59,970
And we've gotten good at it.

226
00:10:59,970 --> 00:11:01,926
So we keep adding
more requirements.

227
00:11:01,926 --> 00:11:03,300
And that's a big
issue right now.

228
00:11:03,300 --> 00:11:04,799
And we'll talk later
in the semester

229
00:11:04,799 --> 00:11:07,260
about complexity management.

230
00:11:07,260 --> 00:11:09,090
But the key message
here is requirements

231
00:11:09,090 --> 00:11:12,330
have been growing over time.

232
00:11:12,330 --> 00:11:14,460
More and more requirements
added as systems grow

233
00:11:14,460 --> 00:11:17,590
and in performance
and complexity.

234
00:11:17,590 --> 00:11:19,560
So here's some standards.

235
00:11:19,560 --> 00:11:21,270
I won't go through
those in detail.

236
00:11:21,270 --> 00:11:25,600
But people recognize, in the
system engineering community,

237
00:11:25,600 --> 00:11:27,940
how important it is,
these requirements.

238
00:11:27,940 --> 00:11:30,070
So in the system
engineering handbook,

239
00:11:30,070 --> 00:11:34,290
there's two sections, 4.2, which
is about technical requirements

240
00:11:34,290 --> 00:11:36,970
definition, and
then section 6.2,

241
00:11:36,970 --> 00:11:39,300
which is about
requirements management.

242
00:11:39,300 --> 00:11:42,180
So definition means that's
the initial definition

243
00:11:42,180 --> 00:11:43,710
of the initial requirement.

244
00:11:43,710 --> 00:11:45,510
And then management
is the process

245
00:11:45,510 --> 00:11:49,670
of updating them, adding
requirements, modifying them,

246
00:11:49,670 --> 00:11:51,530
making sure they're up to date.

247
00:11:51,530 --> 00:11:54,010
And there are a
couple of appendices.

248
00:11:54,010 --> 00:11:56,780
INCOSE, the International
Council of System Engineering

249
00:11:56,780 --> 00:11:59,570
has, in the handbook,
a whole section.

250
00:11:59,570 --> 00:12:01,970
There's even a requirement
working group, people

251
00:12:01,970 --> 00:12:05,150
who really specialize
in this, and then

252
00:12:05,150 --> 00:12:07,940
as well in the ISO
standard, you have a lot

253
00:12:07,940 --> 00:12:09,350
written about requirements.

254
00:12:09,350 --> 00:12:11,300
So this is really a big deal.

255
00:12:11,300 --> 00:12:13,520
So the last thing I want
to do here in this section

256
00:12:13,520 --> 00:12:17,180
is to sort of, at
a very high level,

257
00:12:17,180 --> 00:12:21,200
communicate to you what
requirement really are about.

258
00:12:21,200 --> 00:12:24,020
What they are about
is, like I said,

259
00:12:24,020 --> 00:12:25,880
don't set off in your
ship without-- you

260
00:12:25,880 --> 00:12:28,370
don't know what port
you're sailing to, right?

261
00:12:28,370 --> 00:12:31,340
So requirements set goals
and constrain the design

262
00:12:31,340 --> 00:12:32,960
in the objective space.

263
00:12:32,960 --> 00:12:35,150
So whatever you're
designing, you're

264
00:12:35,150 --> 00:12:38,180
always going to have two spaces
that you're dealing with.

265
00:12:38,180 --> 00:12:40,440
The design space,
on the one hand,

266
00:12:40,440 --> 00:12:44,000
that's the things you
can choose as a designer.

267
00:12:44,000 --> 00:12:45,620
These are the
knobs you can turn,

268
00:12:45,620 --> 00:12:47,150
the decisions you can make.

269
00:12:47,150 --> 00:12:48,650
And then the
objective space, which

270
00:12:48,650 --> 00:12:52,490
is the things that, essentially,
your customer cares about.

271
00:12:52,490 --> 00:12:55,800
I'll give you a
quick example here.

272
00:12:55,800 --> 00:12:59,510
So when we use the word shall,
which I'll get to in a minute.

273
00:12:59,510 --> 00:13:02,630
The English word, shall,
means this is essentially

274
00:13:02,630 --> 00:13:04,220
a constraint.

275
00:13:04,220 --> 00:13:06,310
You must accomplish this.

276
00:13:06,310 --> 00:13:11,180
And when you use the word,
should, it's more like a goal.

277
00:13:11,180 --> 00:13:12,650
It would be nice to do this.

278
00:13:12,650 --> 00:13:15,380
But it's not
absolutely mandatory.

279
00:13:15,380 --> 00:13:18,080
So shall is a hard constraint.

280
00:13:18,080 --> 00:13:22,050
Should it is
desirable as a goal.

281
00:13:22,050 --> 00:13:25,197
So let me give you this
quick example here.

282
00:13:25,197 --> 00:13:26,780
So let's say you're
designing a house.

283
00:13:26,780 --> 00:13:29,420
You're about setting
off to design a house.

284
00:13:29,420 --> 00:13:32,380
And what would be some
requirements for a house.

285
00:13:32,380 --> 00:13:34,670
So I wrote four of them here.

286
00:13:34,670 --> 00:13:38,840
And I'll map them to the design
space and the objective space.

287
00:13:38,840 --> 00:13:41,270
First requirement,
the house shall sleep

288
00:13:41,270 --> 00:13:43,580
between four and six people.

289
00:13:43,580 --> 00:13:45,740
Well, is that in
the design space

290
00:13:45,740 --> 00:13:46,857
or in the objective space?

291
00:13:49,660 --> 00:13:51,389
That's in the
objective space, right?

292
00:13:51,389 --> 00:13:52,180
Yeah, that's right.

293
00:13:52,180 --> 00:13:57,790
So here, we have an axis
called occupance, right?

294
00:13:57,790 --> 00:14:01,610
So four is the minimum and
six is sort of the maximum.

295
00:14:01,610 --> 00:14:03,730
So that's the upper bound.

296
00:14:03,730 --> 00:14:05,860
The next one, the
total build cost

297
00:14:05,860 --> 00:14:09,280
should be less than $550,000.

298
00:14:09,280 --> 00:14:10,990
Should be.

299
00:14:10,990 --> 00:14:13,180
What is that?

300
00:14:13,180 --> 00:14:15,702
Maybe an EPFL.

301
00:14:15,702 --> 00:14:17,602
AUDIENCE: It's a goal.

302
00:14:17,602 --> 00:14:18,560
PROFESSOR: It's a goal.

303
00:14:18,560 --> 00:14:19,720
Yup, that's right.

304
00:14:19,720 --> 00:14:22,540
And is it in the design
space or the objective space?

305
00:14:26,040 --> 00:14:27,617
AUDIENCE: The objective space.

306
00:14:27,617 --> 00:14:28,700
PROFESSOR: That's correct.

307
00:14:28,700 --> 00:14:33,050
So I put this as a dashed line,
OK, because it's a should.

308
00:14:33,050 --> 00:14:36,500
So what I've done here, just
by writing these two sentences,

309
00:14:36,500 --> 00:14:40,100
these two requirements, is I've
essentially carved out a space,

310
00:14:40,100 --> 00:14:40,600
right?

311
00:14:40,600 --> 00:14:42,740
And I can shape that box.

312
00:14:42,740 --> 00:14:44,600
Basically, what these
two requirements

313
00:14:44,600 --> 00:14:49,310
do is they put a box around
in the objective space saying

314
00:14:49,310 --> 00:14:51,770
that, whatever house
you're going to design,

315
00:14:51,770 --> 00:14:55,310
it has to fit within this box.

316
00:14:55,310 --> 00:14:59,420
Then the third one is the house
shall have at least at least

317
00:14:59,420 --> 00:15:01,000
three bedrooms.

318
00:15:01,000 --> 00:15:01,940
What is that?

319
00:15:01,940 --> 00:15:05,390
In the design space
or objective space?

320
00:15:05,390 --> 00:15:06,440
Design space, right?

321
00:15:06,440 --> 00:15:11,840
Because defining a bedroom
is a design decision, right?

322
00:15:11,840 --> 00:15:15,410
And the house should
have a fireplace.

323
00:15:15,410 --> 00:15:19,700
So you can show the lower bound
here, at least three bedrooms.

324
00:15:19,700 --> 00:15:23,540
And then the fireplace is a--

325
00:15:23,540 --> 00:15:24,680
it would be nice to have.

326
00:15:24,680 --> 00:15:27,290
But you don't absolutely
have to have it.

327
00:15:27,290 --> 00:15:29,260
So you can sort of draw this--

328
00:15:29,260 --> 00:15:31,010
it's more like a
line here, right?

329
00:15:31,010 --> 00:15:32,710
Yes, fireplace.

330
00:15:32,710 --> 00:15:34,280
But it's a dashed line.

331
00:15:34,280 --> 00:15:36,580
And then at least
three bedrooms.

332
00:15:36,580 --> 00:15:40,360
So just these four
sentences, we've now put a--

333
00:15:40,360 --> 00:15:43,760
we've defined the space that
we're going to be designing in.

334
00:15:43,760 --> 00:15:46,250
So we have constraints
in the design space.

335
00:15:46,250 --> 00:15:48,340
And we have constraints
in the objective space.

336
00:15:48,340 --> 00:15:51,240
And that's fundamentally
the role of requirement

337
00:15:51,240 --> 00:15:54,620
is to constrain, to give
us direction as to what

338
00:15:54,620 --> 00:15:57,550
we're going to design.

339
00:15:57,550 --> 00:16:02,240
OK, so let's do our first
concept question for today.

340
00:16:02,240 --> 00:16:05,000
And that's the
following question,

341
00:16:05,000 --> 00:16:07,880
do you think there's a
fundamental difference

342
00:16:07,880 --> 00:16:09,560
in the meaning
between the words,

343
00:16:09,560 --> 00:16:13,010
requirements, which we've
talked about so far,

344
00:16:13,010 --> 00:16:15,320
and then the word,
specifications,

345
00:16:15,320 --> 00:16:17,510
which I haven't mentioned yet.

346
00:16:17,510 --> 00:16:22,250
So no, you think they're
essentially the same.

347
00:16:22,250 --> 00:16:24,020
Yes, there is a difference.

348
00:16:24,020 --> 00:16:26,530
Requirements are like the
input, and specifications

349
00:16:26,530 --> 00:16:29,420
are the output of
the design process.

350
00:16:29,420 --> 00:16:31,070
Yes, you think a difference.

351
00:16:31,070 --> 00:16:35,550
Specifications include the
requirements as a subset.

352
00:16:35,550 --> 00:16:36,980
Or you're not sure.

353
00:16:36,980 --> 00:16:39,080
So think about this.

354
00:16:39,080 --> 00:16:41,600
And then submit your answers.

355
00:16:41,600 --> 00:16:44,180
And as far as these
URLs are concerned,

356
00:16:44,180 --> 00:16:46,940
for the concept questions,
I went to all lowercase.

357
00:16:46,940 --> 00:16:49,670
Hopefully, that's a
little bit easier.

358
00:16:49,670 --> 00:16:52,460
So secc3.

359
00:16:52,460 --> 00:16:54,890
tiny.cc/secc3.

360
00:16:54,890 --> 00:16:55,640
What do you think?

361
00:16:55,640 --> 00:16:58,920
Requirements and specifications.

362
00:16:58,920 --> 00:17:00,610
The same thing or not?

363
00:17:37,370 --> 00:17:38,270
Who needs more time?

364
00:17:38,270 --> 00:17:39,000
Anybody?

365
00:17:39,000 --> 00:17:39,500
OK.

366
00:17:52,880 --> 00:17:57,780
OK, so we have nobody
thinks they're the same.

367
00:17:57,780 --> 00:18:00,910
62 thirds of you think that
requirements are the input,

368
00:18:00,910 --> 00:18:02,930
specifications are the output.

369
00:18:02,930 --> 00:18:07,710
9 of you think that
requirements are a subset.

370
00:18:07,710 --> 00:18:09,540
And then 6% are not sure.

371
00:18:09,540 --> 00:18:11,370
So that's good.

372
00:18:11,370 --> 00:18:13,020
I agree with that.

373
00:18:13,020 --> 00:18:17,090
So a lot of people use
requirements and specifications

374
00:18:17,090 --> 00:18:18,060
as the same thing.

375
00:18:18,060 --> 00:18:19,130
They're really not.

376
00:18:19,130 --> 00:18:24,510
OK, so think of requirements
and you haven't actually

377
00:18:24,510 --> 00:18:25,710
started to design that.

378
00:18:25,710 --> 00:18:29,270
This is your putting
constraints in the design space,

379
00:18:29,270 --> 00:18:30,510
defining the direction.

380
00:18:30,510 --> 00:18:33,450
But we don't actually know
what the design will look like.

381
00:18:33,450 --> 00:18:37,680
You may have some requirements
may say you must use this box,

382
00:18:37,680 --> 00:18:40,050
or you must use
this sensor, right?

383
00:18:40,050 --> 00:18:43,170
And you might actually find
some of those in the CanSat.

384
00:18:43,170 --> 00:18:48,630
But basically the
requirements are the input.

385
00:18:48,630 --> 00:18:53,490
So any-- What are
your experiences?

386
00:18:53,490 --> 00:18:56,700
When have you heard those terms,
requirements or specifications

387
00:18:56,700 --> 00:19:01,140
used interchangeably
or not interchangeably?

388
00:19:01,140 --> 00:19:03,150
Has this come up as an
issue for you before?

389
00:19:08,710 --> 00:19:10,980
[? Marissa, ?]
you're worked at--

390
00:19:10,980 --> 00:19:13,940
you're worked on human
spaceflight, right?

391
00:19:13,940 --> 00:19:15,860
Space tourism vehicles.

392
00:19:15,860 --> 00:19:18,335
So what-- how did you guys
do it at Virgin Galactic?

393
00:19:20,936 --> 00:19:22,310
AUDIENCE: Well,
I think something

394
00:19:22,310 --> 00:19:24,620
that you generally saw was like
people will look at sensors.

395
00:19:24,620 --> 00:19:27,050
And they would-- there's a
specification for a sensor.

396
00:19:27,050 --> 00:19:29,600
And they would not necessarily
compare the requirements

397
00:19:29,600 --> 00:19:31,700
to the specification.

398
00:19:31,700 --> 00:19:34,700
So or they would not understand
how the two intersected.

399
00:19:34,700 --> 00:19:37,160
So I think there was a
bit of a misunderstanding

400
00:19:37,160 --> 00:19:39,760
of the difference between what
the actual requirements were,

401
00:19:39,760 --> 00:19:43,489
versus what the specifications
were designed to be.

402
00:19:43,489 --> 00:19:45,280
PROFESSOR: The sensors
you're talking about

403
00:19:45,280 --> 00:19:47,930
would be sensors, I'm assuming,
that you would purchase, right?

404
00:19:47,930 --> 00:19:49,000
As commercial--

405
00:19:49,000 --> 00:19:49,610
AUDIENCE: Yeah, absolutely.

406
00:19:49,610 --> 00:19:51,235
So I think we were
buying of the shelf.

407
00:19:51,235 --> 00:19:53,340
And so I mean, sometimes
the specifications

408
00:19:53,340 --> 00:19:54,800
were broader than what we
wanted and sometimes they

409
00:19:54,800 --> 00:19:55,650
were narrower.

410
00:19:55,650 --> 00:19:58,010
But you know, you can do
additional testing on it.

411
00:19:58,010 --> 00:20:00,343
So there was sometimes a
breakdown between understanding

412
00:20:00,343 --> 00:20:03,171
really whether the sensor met
what we are looking for or it

413
00:20:03,171 --> 00:20:03,670
didn't.

414
00:20:03,670 --> 00:20:04,801
PROFESSOR: OK.

415
00:20:04,801 --> 00:20:05,300
Good.

416
00:20:05,300 --> 00:20:08,210
What about an EPFL.

417
00:20:08,210 --> 00:20:11,101
Any experiences? [? Walker, ?]
you worked on slip rings,

418
00:20:11,101 --> 00:20:11,600
right?

419
00:20:11,600 --> 00:20:15,270
Slip rings was one of the big
specialties of your company.

420
00:20:15,270 --> 00:20:17,330
I'm assuming that
you guys had to make

421
00:20:17,330 --> 00:20:18,560
that distinction, right?

422
00:20:18,560 --> 00:20:20,990
Between requirements
and specifications.

423
00:20:20,990 --> 00:20:24,770
What are your thoughts
at EPFL on this?

424
00:20:24,770 --> 00:20:26,671
GUEST SPEAKER: Well,
yes, from outside,

425
00:20:26,671 --> 00:20:31,054
we clearly had a separation
also on an ether level.

426
00:20:31,054 --> 00:20:33,489
But I was wondering
here in the class,

427
00:20:33,489 --> 00:20:37,348
has anybody seen the difference
between specification

428
00:20:37,348 --> 00:20:38,775
requirements?

429
00:20:38,775 --> 00:20:39,316
Any comments?

430
00:20:39,316 --> 00:20:41,041
Maybe some of the
PhD students that

431
00:20:41,041 --> 00:20:43,290
have been around the block?

432
00:20:43,290 --> 00:21:18,020
AUDIENCE: [INAUDIBLE]

433
00:21:18,020 --> 00:21:20,150
PROFESSOR: Yeah, OK, great.

434
00:21:20,150 --> 00:21:23,570
So I found here, I,
you know, I wanted

435
00:21:23,570 --> 00:21:26,690
to sort of find a real
example that everybody can

436
00:21:26,690 --> 00:21:30,020
relate to very, very easily.

437
00:21:30,020 --> 00:21:32,300
So the answer is there
is a distinction.

438
00:21:32,300 --> 00:21:33,680
They are quite different.

439
00:21:33,680 --> 00:21:36,380
But they're-- got to be careful,
because you can mix them up.

440
00:21:36,380 --> 00:21:40,210
So requirements specify what
the product or system should--

441
00:21:40,210 --> 00:21:42,390
shall or should do, right?

442
00:21:42,390 --> 00:21:46,670
Functions it shall perform, how
well it should perform these.

443
00:21:46,670 --> 00:21:48,800
Also maybe the
degree of autonomy,

444
00:21:48,800 --> 00:21:50,210
how automated is the system?

445
00:21:50,210 --> 00:21:54,460
So what the operators must do,
when the system is operating.

446
00:21:54,460 --> 00:21:57,350
And also compatibility
with other devices.

447
00:21:57,350 --> 00:22:02,450
And then specifications are
about how the system is built

448
00:22:02,450 --> 00:22:04,130
and how it actually works.

449
00:22:04,130 --> 00:22:05,870
So the form, right?

450
00:22:05,870 --> 00:22:07,640
What the materials
that are used,

451
00:22:07,640 --> 00:22:11,240
the dimensions,
schematics, blueprints,

452
00:22:11,240 --> 00:22:13,650
the details of the
user interface.

453
00:22:13,650 --> 00:22:16,670
Those are things-- those
are all specifications.

454
00:22:16,670 --> 00:22:21,140
So I try to look up a very
simple consumer product

455
00:22:21,140 --> 00:22:26,300
that we all are familiar
with, a microwave oven.

456
00:22:26,300 --> 00:22:30,620
Kenmore Elite Countertop
2.2 cubic foot.

457
00:22:30,620 --> 00:22:32,720
You go to Sears website.

458
00:22:32,720 --> 00:22:36,660
And what's interesting is they
actually do a pretty good job.

459
00:22:36,660 --> 00:22:40,100
It's actually pretty consistent
with what I'm saying here,

460
00:22:40,100 --> 00:22:42,210
except they don't talk
about requirements.

461
00:22:42,210 --> 00:22:44,150
They call it description.

462
00:22:44,150 --> 00:22:46,740
So what's listed on
their description.

463
00:22:46,740 --> 00:22:50,390
And this is, I'm pretty
much quoting here, verbatim.

464
00:22:50,390 --> 00:22:54,770
This microwave is large enough
to accommodate the big dishes,

465
00:22:54,770 --> 00:22:55,270
right?

466
00:22:55,270 --> 00:22:59,240
So the idea is this is sort
of family style, right?

467
00:22:59,240 --> 00:23:03,860
This is not a small microwave
just for a frozen meals.

468
00:23:03,860 --> 00:23:06,950
This is-- you should be able
to put a big casserole in it.

469
00:23:06,950 --> 00:23:10,320
And reheat meals and so forth.

470
00:23:10,320 --> 00:23:14,150
So the idea that what
are the things you're

471
00:23:14,150 --> 00:23:15,440
going to use in the microwave.

472
00:23:15,440 --> 00:23:17,000
What's the use case, right?

473
00:23:17,000 --> 00:23:18,260
CONOPS.

474
00:23:18,260 --> 00:23:19,790
CONOPS for the microwave.

475
00:23:19,790 --> 00:23:25,320
You know, family, for kids,
both parents are working,

476
00:23:25,320 --> 00:23:27,950
busy, you know,
not time to spend

477
00:23:27,950 --> 00:23:29,800
two hours preparing dinner.

478
00:23:29,800 --> 00:23:32,300
Therefore, big dishes,
a lot of people.

479
00:23:32,300 --> 00:23:33,620
It has to be quick, right?

480
00:23:33,620 --> 00:23:36,710
That's a CONOPS.

481
00:23:36,710 --> 00:23:38,720
And the requirements,
this description,

482
00:23:38,720 --> 00:23:40,130
is essentially the requirement.

483
00:23:40,130 --> 00:23:43,130
1,200 watts of power
to reheat food quickly.

484
00:23:43,130 --> 00:23:45,200
So time is important.

485
00:23:45,200 --> 00:23:48,470
And then one touch settings
for different food types,

486
00:23:48,470 --> 00:23:50,302
rice, pizza, frozen meals.

487
00:23:50,302 --> 00:23:51,260
And that, what is that?

488
00:23:51,260 --> 00:23:52,910
That's automation, right?

489
00:23:52,910 --> 00:23:55,100
Rather than having to
guess how long and at what

490
00:23:55,100 --> 00:23:59,570
power level to reheat, there's
some partial automation

491
00:23:59,570 --> 00:24:00,260
built in.

492
00:24:00,260 --> 00:24:04,320
That's essentially
requirements, very user centric.

493
00:24:04,320 --> 00:24:05,600
Do you see that?

494
00:24:05,600 --> 00:24:07,780
And then when you look
under specification,

495
00:24:07,780 --> 00:24:11,230
it says the following things:
stainless steel exterior,

496
00:24:11,230 --> 00:24:11,730
right?

497
00:24:11,730 --> 00:24:13,350
Again, you don't
want it to rust.

498
00:24:13,350 --> 00:24:15,400
The dimensions, the weight.

499
00:24:15,400 --> 00:24:17,340
There's a general
warranty one year.

500
00:24:17,340 --> 00:24:19,740
The power cord is included.

501
00:24:19,740 --> 00:24:21,000
It's very different.

502
00:24:21,000 --> 00:24:24,690
Those are things that describe
how it's made, the form of it,

503
00:24:24,690 --> 00:24:25,540
and so forth.

504
00:24:25,540 --> 00:24:30,620
So make sure you
keep those separate.

505
00:24:30,620 --> 00:24:34,870
OK, so let me talk about the
NASA requirements process.

506
00:24:34,870 --> 00:24:37,680
And then we'll talk
about challenges.

507
00:24:37,680 --> 00:24:42,160
So getting back to the
system engineering engine.

508
00:24:42,160 --> 00:24:46,825
That's sort of the heart of
the NASA system engineering

509
00:24:46,825 --> 00:24:47,325
process.

510
00:24:47,325 --> 00:24:50,320
You remember, this is the--

511
00:24:50,320 --> 00:24:53,280
this happens at every
level of decomposition.

512
00:24:53,280 --> 00:24:55,930
After stakeholder expectations.

513
00:24:55,930 --> 00:24:59,850
The second step is technical
requirements definition.

514
00:24:59,850 --> 00:25:02,520
And there's some pretty strong
language in the handbook

515
00:25:02,520 --> 00:25:04,800
that this really
has to be done well.

516
00:25:04,800 --> 00:25:07,680
So the center directors
or their designees

517
00:25:07,680 --> 00:25:10,260
shall establish and
maintain a process

518
00:25:10,260 --> 00:25:12,750
to include activities,
requirements, guidelines,

519
00:25:12,750 --> 00:25:18,730
documentation for the definition
of technical requirements.

520
00:25:18,730 --> 00:25:20,800
So I think I've
already said this,

521
00:25:20,800 --> 00:25:23,655
but I just want to make it
clear again, this is kind of--

522
00:25:23,655 --> 00:25:24,870
I like this cartoon.

523
00:25:24,870 --> 00:25:26,580
This is Moses, right?

524
00:25:26,580 --> 00:25:27,480
Up on thee--

525
00:25:27,480 --> 00:25:30,600
Moses just got the
tablets, right?

526
00:25:30,600 --> 00:25:31,770
The stone tablets.

527
00:25:31,770 --> 00:25:34,800
They-- the 10 commandments.

528
00:25:34,800 --> 00:25:37,740
And one of them is,
thou shalt not steal.

529
00:25:37,740 --> 00:25:40,290
And God says, out of
the clouds, no, they're

530
00:25:40,290 --> 00:25:42,900
requirements, not goals, right?

531
00:25:42,900 --> 00:25:49,045
So what God is saying is that
the 10 commandments are, shall,

532
00:25:49,045 --> 00:25:50,410
right?

533
00:25:50,410 --> 00:25:52,500
And not should.

534
00:25:52,500 --> 00:25:56,510
Even though, I think, some of
us don't always succeed at that.

535
00:25:56,510 --> 00:25:59,420
But that's basically the idea.

536
00:25:59,420 --> 00:26:03,000
Shall is a very hard constraint.

537
00:26:03,000 --> 00:26:05,160
And that's what you
will be judged against.

538
00:26:05,160 --> 00:26:09,330
And should is essentially
a desirable goal.

539
00:26:09,330 --> 00:26:16,920
But the degree of attainment
is somewhat flexible.

540
00:26:16,920 --> 00:26:18,340
Why are we doing this?

541
00:26:18,340 --> 00:26:22,180
Why are we spending our time
writing technical requirements?

542
00:26:22,180 --> 00:26:26,160
It's essentially to transform
those stakeholder expectations

543
00:26:26,160 --> 00:26:28,590
we talked about last
time, transform them

544
00:26:28,590 --> 00:26:31,860
into measurable
technical requirements.

545
00:26:31,860 --> 00:26:33,700
Requirements come in
different flavors.

546
00:26:33,700 --> 00:26:35,560
And I'll mention those flavors.

547
00:26:35,560 --> 00:26:40,790
And we express them in
these shall statements.

548
00:26:40,790 --> 00:26:43,080
They also provide a
basis for agreement

549
00:26:43,080 --> 00:26:45,360
among the stakeholders
and developers.

550
00:26:45,360 --> 00:26:49,230
So you will often find
requirements part of contracts,

551
00:26:49,230 --> 00:26:50,610
right?

552
00:26:50,610 --> 00:26:54,240
Now it becomes serious, you
know when the requirements set

553
00:26:54,240 --> 00:26:56,060
actually has legal implications.

554
00:26:56,060 --> 00:26:59,010
You're legally signing up
to develop a system that

555
00:26:59,010 --> 00:27:01,200
will meet these requirements.

556
00:27:01,200 --> 00:27:04,370
That's pretty serious stuff.

557
00:27:04,370 --> 00:27:06,510
By writing good
technical requirements,

558
00:27:06,510 --> 00:27:10,530
you can reduce the development
effort, because of less rework.

559
00:27:10,530 --> 00:27:13,760
So a lot of rework and
iterations and confusion

560
00:27:13,760 --> 00:27:17,510
is based on missing or poorly
written requirement, right?

561
00:27:17,510 --> 00:27:21,240
And writing those early and
before the design begins

562
00:27:21,240 --> 00:27:23,160
is helpful.

563
00:27:23,160 --> 00:27:26,640
A requirements set is also a
basis for cost and schedule

564
00:27:26,640 --> 00:27:28,120
estimates.

565
00:27:28,120 --> 00:27:30,770
And so some of these
missions that overran.

566
00:27:30,770 --> 00:27:33,300
Do you remember I showed
you this chart last time.

567
00:27:33,300 --> 00:27:37,680
We analyzed 40 missions, earth
and space science missions.

568
00:27:37,680 --> 00:27:41,285
Like 20% of them were
responsible for 80%

569
00:27:41,285 --> 00:27:42,600
of the overruns.

570
00:27:42,600 --> 00:27:45,720
A lot of that had to do
with unrealistic goals

571
00:27:45,720 --> 00:27:50,370
or expectations
and requirements.

572
00:27:50,370 --> 00:27:54,360
And then the next point here is
that the requirements provide

573
00:27:54,360 --> 00:27:56,610
the basis for verification.

574
00:27:56,610 --> 00:27:58,650
In other words, the
better, the more

575
00:27:58,650 --> 00:28:01,780
crisp, the better written
your requirements are,

576
00:28:01,780 --> 00:28:04,860
the easier it is
to test the system

577
00:28:04,860 --> 00:28:07,170
and to check whether these--

578
00:28:07,170 --> 00:28:09,540
whether the system
is in compliance

579
00:28:09,540 --> 00:28:11,040
with the requirements.

580
00:28:11,040 --> 00:28:13,800
And then, eventually,
the basis for acceptance.

581
00:28:13,800 --> 00:28:16,650
Acceptance is also
a big deal, right?

582
00:28:16,650 --> 00:28:19,470
It basically means,
I accept your design.

583
00:28:19,470 --> 00:28:21,210
I take ownership of it.

584
00:28:21,210 --> 00:28:24,120
And I say, you, as the
developer of the system,

585
00:28:24,120 --> 00:28:25,680
have done your job properly.

586
00:28:25,680 --> 00:28:29,130
And there's a legal transfer
of the assets that happens.

587
00:28:29,130 --> 00:28:34,870
So facilitates the transfer
of the product to the users.

588
00:28:34,870 --> 00:28:36,780
And then even later,
if you're going

589
00:28:36,780 --> 00:28:38,820
to do it like a
version 2 or a block

590
00:28:38,820 --> 00:28:41,120
upgrade, a next
generation product,

591
00:28:41,120 --> 00:28:44,400
it's much easier to do when
you have a clear requirements,

592
00:28:44,400 --> 00:28:47,430
because it tells you what the
original system or the earlier

593
00:28:47,430 --> 00:28:51,790
version was able
to do or not do.

594
00:28:51,790 --> 00:28:52,750
Graphically.

595
00:28:52,750 --> 00:28:56,210
So this is figure 4.0.1.

596
00:28:56,210 --> 00:28:58,940
I briefly talked
about this last time.

597
00:28:58,940 --> 00:29:02,010
We start, essentially,
here at the development.

598
00:29:02,010 --> 00:29:03,870
We have mission authority.

599
00:29:03,870 --> 00:29:05,880
We do the stakeholder
expectations.

600
00:29:05,880 --> 00:29:07,770
We talked about that last time.

601
00:29:07,770 --> 00:29:11,070
And then here, right away,
comes a high level requirement.

602
00:29:11,070 --> 00:29:15,220
So that means your level
0, level 1 requirement.

603
00:29:15,220 --> 00:29:20,990
And then as you try to get more
detailed in the requirements,

604
00:29:20,990 --> 00:29:23,590
what do you think happens?

605
00:29:23,590 --> 00:29:25,690
Ideally, you want to
write all the requirements

606
00:29:25,690 --> 00:29:26,470
upfront, right?

607
00:29:26,470 --> 00:29:27,750
That would be great.

608
00:29:27,750 --> 00:29:32,082
But you try to do
that, you hit a wall.

609
00:29:32,082 --> 00:29:34,470
Why is that?

610
00:29:34,470 --> 00:29:36,030
What's the issue?

611
00:29:36,030 --> 00:29:38,250
Why can't we just write
all the requirements

612
00:29:38,250 --> 00:29:40,330
and then be done
with it in one shot?

613
00:29:40,330 --> 00:29:40,830
Yes.

614
00:29:40,830 --> 00:29:42,170
Go ahead.

615
00:29:42,170 --> 00:29:44,100
AUDIENCE: Well, your
system changes over time.

616
00:29:44,100 --> 00:29:45,700
And some of the
requirements clash.

617
00:29:45,700 --> 00:29:48,444
And you can't achieve all
of them simultaneously.

618
00:29:48,444 --> 00:29:50,610
PROFESSOR: Yes. so there's
two things you mentioned.

619
00:29:50,610 --> 00:29:52,187
So one is the sort of--

620
00:29:52,187 --> 00:29:52,770
and I'll talk.

621
00:29:52,770 --> 00:29:55,120
This is known as,
requirements volatility,

622
00:29:55,120 --> 00:29:57,510
like the requirements are
changing as you learn more

623
00:29:57,510 --> 00:29:58,740
about the problem.

624
00:29:58,740 --> 00:30:02,370
And then the other is you detect
conflict between requirements.

625
00:30:02,370 --> 00:30:04,550
And you have to clean those up.

626
00:30:04,550 --> 00:30:06,840
Those are two very valid issues.

627
00:30:06,840 --> 00:30:09,360
But it's not quite
what I was going for.

628
00:30:09,360 --> 00:30:12,060
AUDIENCE: There's another
one, requirements creep,

629
00:30:12,060 --> 00:30:15,540
in which your customer levies
these new requirements on you,

630
00:30:15,540 --> 00:30:16,950
once you've started the process.

631
00:30:16,950 --> 00:30:19,540
PROFESSOR: Yes, so new
requirements get added.

632
00:30:19,540 --> 00:30:22,764
Hopefully, you get
more budget, too.

633
00:30:22,764 --> 00:30:24,180
That doesn't always
happen, right?

634
00:30:24,180 --> 00:30:26,100
So that's another issue.

635
00:30:26,100 --> 00:30:27,550
But it's a little different.

636
00:30:27,550 --> 00:30:29,600
Let's see, EPFL, you guys.

637
00:30:29,600 --> 00:30:32,700
Why don't we write like all four
or five levels of requirements

638
00:30:32,700 --> 00:30:34,260
all at once?

639
00:30:34,260 --> 00:30:36,990
Why can't we do that, typically?

640
00:30:36,990 --> 00:30:38,844
Why do you think?

641
00:30:38,844 --> 00:30:42,610
AUDIENCE: Usually, the other
requirements aren't clear.

642
00:30:42,610 --> 00:30:44,560
They don't exist yet.

643
00:30:44,560 --> 00:30:47,490
Later on in the process,
they become clear.

644
00:30:47,490 --> 00:30:50,250
And you start to realize
the details, which at first,

645
00:30:50,250 --> 00:30:51,259
you don't.

646
00:30:51,259 --> 00:30:53,550
PROFESSOR: That's-- I think
that's pretty close to what

647
00:30:53,550 --> 00:30:54,370
I'm looking for.

648
00:30:54,370 --> 00:30:58,980
So the issue is you do your
level 0, level 1 requirements.

649
00:30:58,980 --> 00:31:01,860
And then sort of as
you get to level 2,

650
00:31:01,860 --> 00:31:04,710
you can't really write
that level 2 requirement

651
00:31:04,710 --> 00:31:08,040
until you've made some
key design decisions.

652
00:31:08,040 --> 00:31:10,890
Are we going to use electric
propulsion or are we going

653
00:31:10,890 --> 00:31:12,870
to use chemical propulsion?

654
00:31:12,870 --> 00:31:16,290
You know, that's a huge
decision in space system design.

655
00:31:16,290 --> 00:31:19,770
And unless you've
made that decision,

656
00:31:19,770 --> 00:31:22,560
you can't really write
lower level requirements,

657
00:31:22,560 --> 00:31:25,860
because the fundamental
working principles

658
00:31:25,860 --> 00:31:29,580
of chemical and electrical
propulsion are quite different.

659
00:31:29,580 --> 00:31:33,390
So that's the key issue is that
you can only do the high level

660
00:31:33,390 --> 00:31:38,310
requirements in a
solution neutral space.

661
00:31:38,310 --> 00:31:40,320
And then you hit the
wall, because you've

662
00:31:40,320 --> 00:31:45,750
got to make some key concept
technology selection decision.

663
00:31:45,750 --> 00:31:47,640
And that's shown
here by this red box

664
00:31:47,640 --> 00:31:50,670
called functional and
logical composition.

665
00:31:50,670 --> 00:31:54,140
I don't like that
nomenclature a lot.

666
00:31:54,140 --> 00:31:56,490
This should really say,
system architecture

667
00:31:56,490 --> 00:31:58,440
or concept selection,
which we'll get

668
00:31:58,440 --> 00:32:01,110
into in the next couple weeks.

669
00:32:01,110 --> 00:32:04,200
Once you've chosen a high
level architecture and concept,

670
00:32:04,200 --> 00:32:08,222
then you say, OK, we're going
to go for ion propulsion.

671
00:32:08,222 --> 00:32:09,930
Well, then you can
write the requirements

672
00:32:09,930 --> 00:32:13,620
for the ion propulsion
system, which

673
00:32:13,620 --> 00:32:18,210
are we through the find
in this yellow box here.

674
00:32:18,210 --> 00:32:20,340
So this is your design
and product structure,

675
00:32:20,340 --> 00:32:23,910
derived an allocated
requirements at lower levels.

676
00:32:23,910 --> 00:32:25,430
You see that?

677
00:32:25,430 --> 00:32:28,140
That's the fundament--
all the things you said,

678
00:32:28,140 --> 00:32:30,390
requirements creep
requirements conflict.

679
00:32:33,420 --> 00:32:35,170
All those things are true.

680
00:32:35,170 --> 00:32:36,750
But the fundamental
reason why we

681
00:32:36,750 --> 00:32:38,700
can't write all the
requirements upfront,

682
00:32:38,700 --> 00:32:42,240
because, at some point, the
lower level requirements depend

683
00:32:42,240 --> 00:32:44,880
on design decisions made.

684
00:32:44,880 --> 00:32:47,910
That's the fundamental issue.

685
00:32:47,910 --> 00:32:51,480
OK, any questions
about that point?

686
00:32:51,480 --> 00:32:52,947
Yes?

687
00:32:52,947 --> 00:32:54,780
AUDIENCE: It's not
totally about that point.

688
00:32:54,780 --> 00:32:55,654
But it's still like--

689
00:32:55,654 --> 00:32:57,600
So requirements
versus specification.

690
00:32:57,600 --> 00:33:00,630
So if you're like
buying a component

691
00:33:00,630 --> 00:33:02,460
from a vendor, some
piece of hardware,

692
00:33:02,460 --> 00:33:04,876
and you want identify like
the specific locations

693
00:33:04,876 --> 00:33:06,250
where it should
attach to things,

694
00:33:06,250 --> 00:33:09,042
it's more of like a form
as opposed to really--

695
00:33:09,042 --> 00:33:10,500
so does that fall
into requirements

696
00:33:10,500 --> 00:33:11,550
or is that like a specification?

697
00:33:11,550 --> 00:33:12,040
PROFESSOR: Great point.

698
00:33:12,040 --> 00:33:13,950
So there are-- what
you're describing

699
00:33:13,950 --> 00:33:16,230
is a peculiar type
of requirement

700
00:33:16,230 --> 00:33:18,240
that we call interface
requirements.

701
00:33:18,240 --> 00:33:21,930
And it's a perfect
segue to this next--

702
00:33:21,930 --> 00:33:23,310
you know, if I
could only include

703
00:33:23,310 --> 00:33:26,650
like I don't five
slides in this lecture,

704
00:33:26,650 --> 00:33:28,920
this would be one of them.

705
00:33:28,920 --> 00:33:31,450
This basically is what
are the different flavors

706
00:33:31,450 --> 00:33:33,400
of requirements or
types of requirements.

707
00:33:33,400 --> 00:33:35,920
And there are six of them here.

708
00:33:35,920 --> 00:33:38,050
So first of all, the
functional requirements.

709
00:33:38,050 --> 00:33:40,050
I think we've sort of
talked about those, right?

710
00:33:40,050 --> 00:33:42,870
Define the functions that
need to be done to accomplish

711
00:33:42,870 --> 00:33:44,630
the mission objectives.

712
00:33:44,630 --> 00:33:48,490
There's some examples here
around thrust vector control.

713
00:33:48,490 --> 00:33:50,310
So basically, the
idea here is that you

714
00:33:50,310 --> 00:33:51,570
have a thrusting system.

715
00:33:51,570 --> 00:33:56,420
And it has or you can
actually direct the thrust.

716
00:33:56,420 --> 00:33:59,130
And in this case, you
should control the thrust.

717
00:33:59,130 --> 00:34:03,420
You shall provide thrust control
around pitch and your axes.

718
00:34:03,420 --> 00:34:06,360
So this statement is a high
level functional statement.

719
00:34:06,360 --> 00:34:11,429
And it's written in the actor
verb, object form, right?

720
00:34:11,429 --> 00:34:14,080
So that's sort of the
classic, you know,

721
00:34:14,080 --> 00:34:16,409
the classic requirement and
the functional requirement.

722
00:34:16,409 --> 00:34:19,010
And then we have
performance requirements.

723
00:34:19,010 --> 00:34:20,960
The performance requirements
are, in a sense,

724
00:34:20,960 --> 00:34:24,719
qualifiers on the
functional requirements.

725
00:34:24,719 --> 00:34:26,460
So a performance
requirement will

726
00:34:26,460 --> 00:34:31,770
specify how well the
function should be performed,

727
00:34:31,770 --> 00:34:34,679
how fast should it
fly, how much thrust.

728
00:34:34,679 --> 00:34:38,230
You know, this is where you
have to actually put numbers in.

729
00:34:38,230 --> 00:34:41,780
So in this case, this
thrust vector controller

730
00:34:41,780 --> 00:34:47,040
shall gimbal the engine
9 degrees, right?

731
00:34:47,040 --> 00:34:50,280
That's the deflection
angle, at least 9 degrees,

732
00:34:50,280 --> 00:34:52,229
plus or minus 1 degree, degree.

733
00:34:52,229 --> 00:34:53,250
0.1 degree.

734
00:34:53,250 --> 00:34:56,820
That's the performance
requirement.

735
00:34:56,820 --> 00:34:59,170
Then we have constraints.

736
00:34:59,170 --> 00:35:02,000
OK, yeah?

737
00:35:02,000 --> 00:35:04,000
STUDENT: Isn't the
performance department saying

738
00:35:04,000 --> 00:35:05,940
that the engine has
to gimbal 9 degrees,

739
00:35:05,940 --> 00:35:08,550
isn't that kind of pushing
toward the specification.

740
00:35:08,550 --> 00:35:09,400
Or is that?

741
00:35:09,400 --> 00:35:11,420
I mean, I guess, is can
also be a requirement.

742
00:35:11,420 --> 00:35:12,480
But isn't that?

743
00:35:12,480 --> 00:35:15,570
PROFESSOR: Right, in
this case, we have a set,

744
00:35:15,570 --> 00:35:20,610
and we're not specifying how
those 9 degrees are achieved,

745
00:35:20,610 --> 00:35:27,090
you know, whether it's through
a gimbal or the actual mechanism

746
00:35:27,090 --> 00:35:29,280
or how it's not described here.

747
00:35:29,280 --> 00:35:31,690
All that's described is
the angular [? range, ?]

748
00:35:31,690 --> 00:35:33,110
essentially, that this.

749
00:35:33,110 --> 00:35:36,760
And that is a
performance requirement.

750
00:35:36,760 --> 00:35:38,820
But you'll see, there
will be some examples

751
00:35:38,820 --> 00:35:42,630
of things that are putting
limits on the form, which looks

752
00:35:42,630 --> 00:35:44,640
more like a specification.

753
00:35:44,640 --> 00:35:46,440
So then we have constraints.

754
00:35:46,440 --> 00:35:52,362
Constraints are things on
like weight, mass, power,

755
00:35:52,362 --> 00:35:53,070
things like that.

756
00:35:53,070 --> 00:35:56,070
So constraints are requirements
that cannot be traded off with

757
00:35:56,070 --> 00:35:58,450
respect that cost,
schedule, or performance.

758
00:35:58,450 --> 00:36:03,300
So for example, the prospector
controller unit module

759
00:36:03,300 --> 00:36:06,210
shall weigh less
than 120 pounds.

760
00:36:06,210 --> 00:36:08,860
So that's a-- that's not,
if you think about it,

761
00:36:08,860 --> 00:36:10,770
that is not functional, right?

762
00:36:10,770 --> 00:36:16,190
The weight of the
thrust vector controller

763
00:36:16,190 --> 00:36:20,450
shall not be more than 100--
that is not performance.

764
00:36:20,450 --> 00:36:23,120
It's not functional requirement,
but it's a constraint

765
00:36:23,120 --> 00:36:24,860
on the form, essentially.

766
00:36:24,860 --> 00:36:28,570
So this looks more like what
we would call a specification

767
00:36:28,570 --> 00:36:30,230
constraint.

768
00:36:30,230 --> 00:36:32,390
The fourth one is what
[? Marissa ?] brought up.

769
00:36:32,390 --> 00:36:34,580
This is an interface
requirement.

770
00:36:34,580 --> 00:36:38,510
OK, so in this case, our
thrust vector controller

771
00:36:38,510 --> 00:36:42,560
shall interface with
the J-2x The J-2x is

772
00:36:42,560 --> 00:36:44,400
a very famous engine.

773
00:36:44,400 --> 00:36:48,120
And J-2x is a kind of--

774
00:36:48,120 --> 00:36:50,060
this is sort of,
this was written

775
00:36:50,060 --> 00:36:53,150
during the constellation days
of the constellation program

776
00:36:53,150 --> 00:36:57,500
at NASA is the idea of a new
generation of the J-2 engine.

777
00:36:57,500 --> 00:37:00,500
So the idea is that
whatever you do,

778
00:37:00,500 --> 00:37:03,590
however you design your
thrust vector controller,

779
00:37:03,590 --> 00:37:07,670
it must be able to interface
with the J-2x engine,

780
00:37:07,670 --> 00:37:10,490
according to
conditions specified

781
00:37:10,490 --> 00:37:13,890
in this interface
control document.

782
00:37:13,890 --> 00:37:18,230
So this is called an
interface requirement.

783
00:37:18,230 --> 00:37:22,050
Then the fifth category are
environmental requirements.

784
00:37:22,050 --> 00:37:27,120
So the TVC shall use [? Biber ?]
acoustic shocks and loads

785
00:37:27,120 --> 00:37:30,020
according, again, to some
environmental document.

786
00:37:30,020 --> 00:37:32,300
And by the way, for
the CanSat competition,

787
00:37:32,300 --> 00:37:33,300
we have that, right?

788
00:37:33,300 --> 00:37:34,640
There's two documents.

789
00:37:34,640 --> 00:37:36,080
There's the mission guide.

790
00:37:36,080 --> 00:37:38,660
And then there's this
environmental testing guide.

791
00:37:38,660 --> 00:37:43,930
So what this essentially says
is under what conditions shall

792
00:37:43,930 --> 00:37:46,860
the performance,
and functional--

793
00:37:46,860 --> 00:37:48,950
the functions and
performance that

794
00:37:48,950 --> 00:37:50,960
are specified in
the first two type

795
00:37:50,960 --> 00:37:54,350
of requirements, under
what conditions shall that

796
00:37:54,350 --> 00:37:56,300
be performed, right?

797
00:37:56,300 --> 00:37:58,340
And if anything
you think about it,

798
00:37:58,340 --> 00:38:01,640
designing whether it's
a sensor or anything

799
00:38:01,640 --> 00:38:07,880
to operate between 0 degrees
Celsius and 30 degrees Celsius

800
00:38:07,880 --> 00:38:13,400
or between minus 60 and plus
80, is a huge difference, right?

801
00:38:13,400 --> 00:38:15,350
You can write that down.

802
00:38:15,350 --> 00:38:18,170
But what that actually
means to open up

803
00:38:18,170 --> 00:38:21,200
the range of
environmental conditions,

804
00:38:21,200 --> 00:38:23,330
it has a big, big
impact on the design.

805
00:38:23,330 --> 00:38:25,970
And so this is a big
deal in practice.

806
00:38:25,970 --> 00:38:29,797
And then, at least in
Space Systems design,

807
00:38:29,797 --> 00:38:31,880
you know a lot of these
environmental requirements

808
00:38:31,880 --> 00:38:35,930
are of course, driven by the
space environment or the launch

809
00:38:35,930 --> 00:38:36,980
environment.

810
00:38:36,980 --> 00:38:39,530
If you're designing
airplanes, whether you're

811
00:38:39,530 --> 00:38:42,140
flying in a small mountainous
country like Switzerland

812
00:38:42,140 --> 00:38:45,320
or on the ocean off of
an aircraft carrier,

813
00:38:45,320 --> 00:38:47,539
those are very
different environments.

814
00:38:47,539 --> 00:38:48,830
And they're going to influence.

815
00:38:48,830 --> 00:38:51,050
You have to specify the
operating requirement.

816
00:38:51,050 --> 00:38:53,480
The bigger the
envelope that you make

817
00:38:53,480 --> 00:38:56,780
for the environmental
requirements, the more complex

818
00:38:56,780 --> 00:38:59,120
the system will be.

819
00:38:59,120 --> 00:39:04,505
Also medical-- anybody working
on medical devices here?

820
00:39:04,505 --> 00:39:05,340
EPFL.

821
00:39:05,340 --> 00:39:05,960
Anybody?

822
00:39:05,960 --> 00:39:07,940
Medical devices?

823
00:39:07,940 --> 00:39:11,030
So in medical devices,
it's the same thing.

824
00:39:11,030 --> 00:39:12,680
Is this medical
device going to be

825
00:39:12,680 --> 00:39:15,100
used in a hospital
setting, where

826
00:39:15,100 --> 00:39:16,910
everything is kind of clean.

827
00:39:16,910 --> 00:39:19,190
You have clean power.

828
00:39:19,190 --> 00:39:21,980
The nurses, everybody
is very well-trained.

829
00:39:21,980 --> 00:39:22,800
That's one thing.

830
00:39:22,800 --> 00:39:26,510
Or is his medical device
going to be used in the field,

831
00:39:26,510 --> 00:39:27,580
you know?

832
00:39:27,580 --> 00:39:31,670
In Africa, in India, in a rural
area, where people are not,

833
00:39:31,670 --> 00:39:35,090
maybe not trained, medically
and professionally and so forth.

834
00:39:35,090 --> 00:39:35,910
Very different.

835
00:39:35,910 --> 00:39:37,890
The function may be the same.

836
00:39:37,890 --> 00:39:40,130
But the environmental
conditions are very different.

837
00:39:40,130 --> 00:39:43,130
You have to really
specify those.

838
00:39:43,130 --> 00:39:46,460
And then we have-- the last
category is kind of the other.

839
00:39:46,460 --> 00:39:49,220
It's sort of a catchall.

840
00:39:49,220 --> 00:39:51,210
But it can be very important.

841
00:39:51,210 --> 00:39:54,980
So this includes a lot of the
human factors, reliability

842
00:39:54,980 --> 00:39:56,390
and safety requirements.

843
00:39:56,390 --> 00:39:59,900
Those are-- and you know,
they're listed here as other.

844
00:39:59,900 --> 00:40:01,340
And I think it's
really important

845
00:40:01,340 --> 00:40:04,820
to say that just because
they're listed here

846
00:40:04,820 --> 00:40:06,410
doesn't mean they're
less important.

847
00:40:06,410 --> 00:40:09,740
These are often
neglected, unfortunately.

848
00:40:09,740 --> 00:40:12,360
And it really can hurt
you in the long term.

849
00:40:12,360 --> 00:40:16,070
So pay attention to those
human factors, reliability,

850
00:40:16,070 --> 00:40:17,540
and safety requirements.

851
00:40:17,540 --> 00:40:22,250
So six types of requirements.

852
00:40:22,250 --> 00:40:25,220
So let's talk about what
makes good or acceptable

853
00:40:25,220 --> 00:40:26,810
requirements.

854
00:40:26,810 --> 00:40:30,500
And there's a distinction here
between a single requirement

855
00:40:30,500 --> 00:40:33,050
statement and then
sets of a requirements.

856
00:40:33,050 --> 00:40:35,480
So first of all,
requirements should

857
00:40:35,480 --> 00:40:37,670
be written in natural language.

858
00:40:37,670 --> 00:40:41,480
They should be
complete sentences.

859
00:40:41,480 --> 00:40:44,600
And each of the
requirement statements

860
00:40:44,600 --> 00:40:52,010
should be clear and consistent,
meaning that it's not a novel

861
00:40:52,010 --> 00:40:54,260
or it's not a poem.

862
00:40:54,260 --> 00:40:55,100
But it's clear.

863
00:40:55,100 --> 00:40:56,240
It's understandable.

864
00:40:56,240 --> 00:40:57,890
It's well-written.

865
00:40:57,890 --> 00:40:58,730
It's correct.

866
00:40:58,730 --> 00:40:59,930
There's no errors in it.

867
00:40:59,930 --> 00:41:00,980
It's feasible.

868
00:41:00,980 --> 00:41:02,780
Now that's a really tricky one.

869
00:41:02,780 --> 00:41:06,170
That's the first thing
here that's really tricky,

870
00:41:06,170 --> 00:41:07,490
when you think about it.

871
00:41:07,490 --> 00:41:09,140
Feasible means, what?

872
00:41:09,140 --> 00:41:12,680
It means this requirement
can be satisfied

873
00:41:12,680 --> 00:41:17,120
within the laws of physics and
state of the art technologies

874
00:41:17,120 --> 00:41:18,680
and other project constraints.

875
00:41:18,680 --> 00:41:22,022
So why is that a tricky one?

876
00:41:22,022 --> 00:41:23,960
Go ahead.

877
00:41:23,960 --> 00:41:26,270
AUDIENCE: I was going to ask
a question regarding that,

878
00:41:26,270 --> 00:41:29,120
in terms of how do you
deal with a program

879
00:41:29,120 --> 00:41:31,360
where you're working
on things or you're

880
00:41:31,360 --> 00:41:34,160
trying to actually maybe
define the state of the art

881
00:41:34,160 --> 00:41:37,220
or figure out what is feasible.

882
00:41:37,220 --> 00:41:38,060
PROFESSOR: Right.

883
00:41:38,060 --> 00:41:41,720
so if you're basically
designing a product or a project

884
00:41:41,720 --> 00:41:46,010
or a system that's a repeat
of what's already been done,

885
00:41:46,010 --> 00:41:49,160
then you can have pretty
good confidence, right,

886
00:41:49,160 --> 00:41:51,230
that those poles are feasible.

887
00:41:51,230 --> 00:41:55,121
But what if you're
doing, let's say the--

888
00:41:55,121 --> 00:41:57,110
I think I mentioned
this last time.

889
00:41:57,110 --> 00:41:58,880
You're going to Europa.

890
00:41:58,880 --> 00:42:00,980
And you're going to
drill through the ice.

891
00:42:00,980 --> 00:42:03,980
And you're going to explore the
ocean under the ice in Europa.

892
00:42:03,980 --> 00:42:05,720
It's never been done before.

893
00:42:05,720 --> 00:42:07,370
You know, can we
actually do this?

894
00:42:07,370 --> 00:42:09,920
So this is the tricky thing.

895
00:42:09,920 --> 00:42:13,400
How ambitious, how ambitious
can the requirements

896
00:42:13,400 --> 00:42:16,040
be and you still
claim feasibility?

897
00:42:16,040 --> 00:42:18,380
And that's also one
of the big reasons

898
00:42:18,380 --> 00:42:21,210
why programs get in trouble
is when they're actually

899
00:42:21,210 --> 00:42:26,870
defining requirements that are
not really feasible within--

900
00:42:26,870 --> 00:42:29,060
they're definitely way
beyond the state of the art.

901
00:42:29,060 --> 00:42:32,900
And within the time frame
and budget allocated,

902
00:42:32,900 --> 00:42:34,460
you can't get there.

903
00:42:34,460 --> 00:42:37,430
And many, many programs
that run into this problem

904
00:42:37,430 --> 00:42:39,650
are the ones where
the technologies

905
00:42:39,650 --> 00:42:42,570
that you're going to use
are not really ready yet.

906
00:42:42,570 --> 00:42:45,260
You know, there-- we'll
talk later about technology

907
00:42:45,260 --> 00:42:47,020
readiness levels scale.

908
00:42:47,020 --> 00:42:49,730
You're not really
ready yet to do this,

909
00:42:49,730 --> 00:42:51,830
but you're going to try anyway.

910
00:42:51,830 --> 00:42:54,530
So that's a tricky
one, feasibility.

911
00:42:54,530 --> 00:42:57,920
Flexibility is, you
know, don't over specify

912
00:42:57,920 --> 00:42:59,010
how things should be done.

913
00:42:59,010 --> 00:43:01,850
So don't say how it
should be satisfied.

914
00:43:01,850 --> 00:43:02,990
Without ambiguity.

915
00:43:02,990 --> 00:43:06,320
That means if 10 people
read this requirement,

916
00:43:06,320 --> 00:43:10,390
they should have the same or
very similar interpretation.

917
00:43:10,390 --> 00:43:13,350
Singular statement, one
actor [? verb ?] object.

918
00:43:13,350 --> 00:43:16,580
And then the last point
here is verifiability.

919
00:43:16,580 --> 00:43:19,730
How are you going to
check whether or not

920
00:43:19,730 --> 00:43:22,794
this requirement will,
in fact, satisfy you.

921
00:43:22,794 --> 00:43:24,710
AUDIENCE: With regard
to that, the feasibility

922
00:43:24,710 --> 00:43:26,570
again for some of
these programs I

923
00:43:26,570 --> 00:43:32,220
had large, [? overrun ?]
schedule, overruns.

924
00:43:32,220 --> 00:43:34,370
I mean, would it be
almost better in a sense

925
00:43:34,370 --> 00:43:40,270
to have some of the requirements
be more shalls, like--

926
00:43:40,270 --> 00:43:41,810
or goal.

927
00:43:41,810 --> 00:43:45,410
In other words, you know, if
you can do this, go for it.

928
00:43:45,410 --> 00:43:47,690
But if you're going
to run over 10 years,

929
00:43:47,690 --> 00:43:50,580
maybe back off a
couple of percent.

930
00:43:50,580 --> 00:43:51,580
PROFESSOR: That's right.

931
00:43:51,580 --> 00:43:53,496
And what-- we'll talk
about this a little bit.

932
00:43:53,496 --> 00:43:56,750
But if you-- all of them
are shall statements.

933
00:43:56,750 --> 00:43:58,520
They're all hard constraints.

934
00:43:58,520 --> 00:44:01,100
The objective space
that I showed you

935
00:44:01,100 --> 00:44:04,730
may, in fact, have
0 feasible space.

936
00:44:04,730 --> 00:44:08,060
So really knowing where, what
is a hard constraint, what

937
00:44:08,060 --> 00:44:11,880
is a hard requirement,
and what is flexible,

938
00:44:11,880 --> 00:44:13,790
that's very tricky.

939
00:44:13,790 --> 00:44:17,570
And that's why it's important
to actually not just accept

940
00:44:17,570 --> 00:44:18,290
requirements.

941
00:44:18,290 --> 00:44:21,026
You know, if you're going to run
a project, every requirement,

942
00:44:21,026 --> 00:44:22,400
you want to really
understand it.

943
00:44:22,400 --> 00:44:27,260
And if you think this
requirement is infeasible,

944
00:44:27,260 --> 00:44:28,550
you have to negotiate.

945
00:44:28,550 --> 00:44:30,130
That's where the
upfront negotiation

946
00:44:30,130 --> 00:44:31,650
becomes really important.

947
00:44:31,650 --> 00:44:35,251
Did we lose our EPFL?

948
00:44:35,251 --> 00:44:35,750
OK.

949
00:44:35,750 --> 00:44:36,020
GUEST SPEAKER: Hello?

950
00:44:36,020 --> 00:44:37,186
PROFESSOR: Go ahead, please.

951
00:44:37,186 --> 00:44:40,002
GUEST SPEAKER: [INAUDIBLE]

952
00:44:40,002 --> 00:44:41,210
PROFESSOR: Please, go for it.

953
00:44:44,280 --> 00:44:49,882
AUDIENCE: So in 1990, we got
the requirements from NASA

954
00:44:49,882 --> 00:44:52,512
to build a space
[? bioreactor. ?]

955
00:44:52,512 --> 00:44:54,400
And all the students
will love this.

956
00:44:54,400 --> 00:44:57,694
It was the smallest brewery
that ever flew in space.

957
00:44:57,694 --> 00:45:02,910
And the initial requirements was
that all fluid containers had

958
00:45:02,910 --> 00:45:05,570
to be solid with double walls.

959
00:45:05,570 --> 00:45:10,855
And this meant that we had
another solution with bladders.

960
00:45:10,855 --> 00:45:11,793
[INAUDIBLE] bladders.

961
00:45:11,793 --> 00:45:15,550
So we had a real fight up
front in the proposal phase

962
00:45:15,550 --> 00:45:19,885
to demonstrate the
compliancy to the requirement

963
00:45:19,885 --> 00:45:23,436
of going the cell culture,
yeast cell culture,

964
00:45:23,436 --> 00:45:27,930
and still being
non-compliant, actually,

965
00:45:27,930 --> 00:45:33,271
to this requirement of hard
double-walled fluid containers.

966
00:45:33,271 --> 00:45:37,199
And so this is exactly
[? the point, finding these ?]

967
00:45:37,199 --> 00:45:42,271
[INAUDIBLE] since the
last 15 to 20 years.

968
00:45:42,271 --> 00:45:44,210
And this was a
breakthrough in technology.

969
00:45:44,210 --> 00:45:46,653
And there, you will always
have these challenges

970
00:45:46,653 --> 00:45:49,479
with the agencies,
with your customers.

971
00:45:49,479 --> 00:45:51,632
They are [INAUDIBLE]
set in their ways.

972
00:45:51,632 --> 00:45:55,880
And you have to demonstrate
the feasibility to them

973
00:45:55,880 --> 00:45:58,330
to make them change the
requirement that allows you

974
00:45:58,330 --> 00:46:00,074
to change the specification.

975
00:46:00,074 --> 00:46:01,990
PROFESSOR: No, I think
that's a great example.

976
00:46:01,990 --> 00:46:04,730
So in the end, you were
successful to argue

977
00:46:04,730 --> 00:46:07,886
for the bladder the
bladder solution.

978
00:46:07,886 --> 00:46:10,035
GUEST SPEAKER: Well,
the bladder was just how

979
00:46:10,035 --> 00:46:14,550
to keep the fresh medium and the
used medium, meaning the beer.

980
00:46:14,550 --> 00:46:19,270
The technology was
to put in [INAUDIBLE]

981
00:46:19,270 --> 00:46:23,160
And actually, it needed
to have flexibility

982
00:46:23,160 --> 00:46:26,825
on the pressure
in the reservoir.

983
00:46:26,825 --> 00:46:29,775
That's why the argument to
put [INAUDIBLE] in space

984
00:46:29,775 --> 00:46:32,740
finally won the day,
and allowed to relax

985
00:46:32,740 --> 00:46:33,850
the other requirements.

986
00:46:33,850 --> 00:46:36,200
PROFESSOR: Yeah,
thank you, thank you.

987
00:46:36,200 --> 00:46:37,880
A great example there.

988
00:46:37,880 --> 00:46:39,440
Any other comments?

989
00:46:39,440 --> 00:46:42,730
AUDIENCE: [COUGHING]

990
00:46:42,730 --> 00:46:45,690
PROFESSOR: OK, so so all
this is a pretty long list.

991
00:46:45,690 --> 00:46:48,600
And this applies to
single requirements.

992
00:46:48,600 --> 00:46:51,480
And then there's a set of--
this is really important.

993
00:46:51,480 --> 00:46:53,370
Then there's a set
of characteristics

994
00:46:53,370 --> 00:46:56,760
that we want to see when you
look at sets of requirements,

995
00:46:56,760 --> 00:46:58,440
groups that require.

996
00:46:58,440 --> 00:47:00,220
Absence of redundancy.

997
00:47:00,220 --> 00:47:04,220
This means that each requirement
is specified only once, right?

998
00:47:04,220 --> 00:47:06,870
You don't want to
have redundancy.

999
00:47:06,870 --> 00:47:09,240
Redundancy can be
good in system design,

1000
00:47:09,240 --> 00:47:11,850
but not in the requirements.

1001
00:47:11,850 --> 00:47:14,220
Consistency, using terms.

1002
00:47:14,220 --> 00:47:16,410
Completeness, this
basically means

1003
00:47:16,410 --> 00:47:18,480
not missing key requirements.

1004
00:47:18,480 --> 00:47:21,240
And then this idea of
absence of conflict.

1005
00:47:21,240 --> 00:47:23,670
And this is also similar
to the feasibility.

1006
00:47:23,670 --> 00:47:25,650
This is a tricky one.

1007
00:47:25,650 --> 00:47:28,150
Requirements can be in
tension with each other,

1008
00:47:28,150 --> 00:47:30,810
particularly their should goals.

1009
00:47:30,810 --> 00:47:33,960
But they shouldn't be in direct
conflict with each other,

1010
00:47:33,960 --> 00:47:37,410
like you know you shall
use aluminum for this unit.

1011
00:47:37,410 --> 00:47:40,020
And then another
requirements says, no metals

1012
00:47:40,020 --> 00:47:42,660
are allowed to be
used in this unit.

1013
00:47:42,660 --> 00:47:44,420
You can't.

1014
00:47:44,420 --> 00:47:46,260
That conflict is not solvable.

1015
00:47:46,260 --> 00:47:48,120
That's a direct contradiction.

1016
00:47:48,120 --> 00:47:51,240
That's different from having
competing requirements

1017
00:47:51,240 --> 00:47:54,010
or requirements in tension.

1018
00:47:54,010 --> 00:47:56,610
So let's do a quick exercise.

1019
00:47:56,610 --> 00:47:59,230
And then we'll actually
take a break, as well.

1020
00:47:59,230 --> 00:48:03,870
So this is a turn to
your partner exercise.

1021
00:48:03,870 --> 00:48:05,490
And then we'll have a break.

1022
00:48:05,490 --> 00:48:08,670
And we'll restart in
like about seven minutes.

1023
00:48:08,670 --> 00:48:10,200
So here's what I'd
like you to do.

1024
00:48:10,200 --> 00:48:12,240
I have three systems here.

1025
00:48:12,240 --> 00:48:15,130
They're very different
in scale and complexity.

1026
00:48:15,130 --> 00:48:17,280
So A is sticky tape.

1027
00:48:17,280 --> 00:48:19,710
And I'll tell you
this quick story here.

1028
00:48:19,710 --> 00:48:23,440
I grew up in Switzerland
surrounded by farms, you know.

1029
00:48:23,440 --> 00:48:25,650
And I spent all my
time at the farms

1030
00:48:25,650 --> 00:48:28,260
and you know, big
tables, big families.

1031
00:48:28,260 --> 00:48:30,850
And this is-- they
have these things here.

1032
00:48:30,850 --> 00:48:34,380
This is basically-- these
are flies sticking on tape.

1033
00:48:34,380 --> 00:48:37,570
This is to keep the flies
off the dinner table.

1034
00:48:37,570 --> 00:48:41,880
It's called Mr. sticky
tape for trapping flies.

1035
00:48:41,880 --> 00:48:43,560
Really kind of gross.

1036
00:48:43,560 --> 00:48:45,890
But I remember it.

1037
00:48:45,890 --> 00:48:48,670
B is I just a couple
of weeks ago had

1038
00:48:48,670 --> 00:48:50,820
a test drive and the new I3.

1039
00:48:50,820 --> 00:48:55,110
This is the BMW small
electric city car.

1040
00:48:55,110 --> 00:48:56,850
Very cool.

1041
00:48:56,850 --> 00:49:01,590
And then C is something that at
EPFL, you guys know very well.

1042
00:49:01,590 --> 00:49:03,570
This is the Rolex center.

1043
00:49:03,570 --> 00:49:05,190
It looks kind of
like Swiss cheese,

1044
00:49:05,190 --> 00:49:06,690
when you look from the top.

1045
00:49:06,690 --> 00:49:10,630
This is the equivalent
of W-20 here at MIT.

1046
00:49:10,630 --> 00:49:11,970
This is the student center.

1047
00:49:11,970 --> 00:49:14,260
So there's a library there.

1048
00:49:14,260 --> 00:49:16,930
There's a cafeteria
there and so forth.

1049
00:49:16,930 --> 00:49:21,010
So what I'd like you to do
is turn to your partner,

1050
00:49:21,010 --> 00:49:25,140
pick one of those
three, and come up

1051
00:49:25,140 --> 00:49:29,280
with one single statement,
one good requirement

1052
00:49:29,280 --> 00:49:33,030
that you think was possibly
used in the development

1053
00:49:33,030 --> 00:49:36,300
of that solution,
whichever one you pick, OK?

1054
00:49:36,300 --> 00:49:38,460
So pick one of those three.

1055
00:49:38,460 --> 00:49:41,790
And then jointly discuss
and write a requirement

1056
00:49:41,790 --> 00:49:45,510
that you think led
to this design, OK?

1057
00:49:45,510 --> 00:49:47,550
So take about five minutes.

1058
00:49:47,550 --> 00:49:48,420
Take a break.

1059
00:49:48,420 --> 00:49:51,900
And then we'll sample
what you came up with.

1060
00:49:51,900 --> 00:49:55,520
So let's hear from EPFL first.

1061
00:49:55,520 --> 00:49:59,200
did anybody do A?

1062
00:49:59,200 --> 00:50:00,570
AUDIENCE: We did A.

1063
00:50:00,570 --> 00:50:02,010
PROFESSOR: OK, go for it.

1064
00:50:02,010 --> 00:50:04,200
Speak up.

1065
00:50:04,200 --> 00:50:06,660
AUDIENCE: We had four
requirements for.

1066
00:50:06,660 --> 00:50:08,430
PROFESSOR: OK, go for it.

1067
00:50:08,430 --> 00:50:09,470
AUDIENCE: All right.

1068
00:50:09,470 --> 00:50:15,570
So The tape shows
[INAUDIBLE] of the tape

1069
00:50:15,570 --> 00:50:18,600
should be less than 4 grand.

1070
00:50:18,600 --> 00:50:26,554
The tape should be able
to catch up to 60 insects.

1071
00:50:26,554 --> 00:50:29,270
And the sticky tape should
not be toxic for humans.

1072
00:50:29,270 --> 00:50:31,500
PROFESSOR: OK, I like,
yeah, the toxicity.

1073
00:50:31,500 --> 00:50:32,250
That's good.

1074
00:50:32,250 --> 00:50:33,570
Excellent.

1075
00:50:33,570 --> 00:50:34,320
Very good.

1076
00:50:34,320 --> 00:50:36,120
ow who did A here at MIT.

1077
00:50:36,120 --> 00:50:36,810
Anybody?

1078
00:50:36,810 --> 00:50:37,990
A?

1079
00:50:37,990 --> 00:50:38,490
Go for it.

1080
00:50:42,400 --> 00:50:46,830
AUDIENCE: We said
if greater than 10%

1081
00:50:46,830 --> 00:50:50,070
of the surface area of the
fly contacts the paper,

1082
00:50:50,070 --> 00:50:53,630
it shall not be able
to release itself.

1083
00:50:53,630 --> 00:50:55,000
PROFESSOR: Ah, I see.

1084
00:50:55,000 --> 00:50:57,750
So this is a
trapping requirement.

1085
00:50:57,750 --> 00:50:59,459
AUDIENCE: Yeah, how
effective should it--

1086
00:50:59,459 --> 00:51:01,958
PROFESSOR: There are different,
very different flies, right?

1087
00:51:01,958 --> 00:51:03,630
There like these
little day flies.

1088
00:51:03,630 --> 00:51:05,400
And then there's
huge horseflies.

1089
00:51:05,400 --> 00:51:08,710
So which type of fly
did this apply to.

1090
00:51:08,710 --> 00:51:11,190
AUDIENCE: We said 10% of
the surface area of the fly.

1091
00:51:11,190 --> 00:51:12,840
PROFESSOR: Oh, any type of fly.

1092
00:51:12,840 --> 00:51:14,835
Any different fly species.

1093
00:51:14,835 --> 00:51:15,460
AUDIENCE: Yeah.

1094
00:51:15,460 --> 00:51:16,043
PROFESSOR: OK.

1095
00:51:16,594 --> 00:51:17,260
All right, cool.

1096
00:51:17,260 --> 00:51:18,516
So go ahead.

1097
00:51:18,516 --> 00:51:19,890
Well, you know
the point of this.

1098
00:51:19,890 --> 00:51:21,570
You get the point
of this is, right?

1099
00:51:21,570 --> 00:51:26,290
A is like the sort of
you think trivial system.

1100
00:51:26,290 --> 00:51:28,170
But once you really
start thinking about it,

1101
00:51:28,170 --> 00:51:32,700
it's pretty tricky, right?

1102
00:51:32,700 --> 00:51:36,300
Who else wants to, A-- something
that hasn't been mentioned yet

1103
00:51:36,300 --> 00:51:38,860
on A. Sam, go for it.

1104
00:51:38,860 --> 00:51:40,550
Make sure you push the button.

1105
00:51:43,310 --> 00:51:45,990
AUDIENCE: Yeah,
for the sticky trap

1106
00:51:45,990 --> 00:51:50,470
we said that the product
shall fit on a store shelf

1107
00:51:50,470 --> 00:51:52,060
when packaged.

1108
00:51:52,060 --> 00:51:54,180
PROFESSOR: OK, so that's
kind of packaging,

1109
00:51:54,180 --> 00:51:57,240
logistics, distribution
requirement.

1110
00:51:57,240 --> 00:51:58,990
Very good.

1111
00:51:58,990 --> 00:52:00,715
Mike?

1112
00:52:00,715 --> 00:52:02,590
AUDIENCE: They kind of
already touched on it,

1113
00:52:02,590 --> 00:52:05,570
but the sticky tape shall
not be toxic and allow for--

1114
00:52:05,570 --> 00:52:08,716
allow for removal of human
skin without bodily harm.

1115
00:52:08,716 --> 00:52:09,590
PROFESSOR: OK, great.

1116
00:52:09,590 --> 00:52:12,492
So that's in the same--
you know, human factors

1117
00:52:12,492 --> 00:52:13,450
requirement, basically.

1118
00:52:13,450 --> 00:52:13,950
Yeah.

1119
00:52:13,950 --> 00:52:15,210
Very good.

1120
00:52:15,210 --> 00:52:17,715
Anything else on A at EPFL?

1121
00:52:17,715 --> 00:52:29,100
AUDIENCE: [INAUDIBLE]

1122
00:52:29,100 --> 00:52:31,290
PROFESSOR: OK, so
installation requirement.

1123
00:52:31,290 --> 00:52:32,200
OK.

1124
00:52:32,200 --> 00:52:36,090
Nobody mentioned the ones
I remember as a child are

1125
00:52:36,090 --> 00:52:36,930
bigger than this.

1126
00:52:36,930 --> 00:52:38,880
They were really long.

1127
00:52:38,880 --> 00:52:40,350
So capacity, right?

1128
00:52:40,350 --> 00:52:43,470
You could sort of have a
length or capacity requirement,

1129
00:52:43,470 --> 00:52:44,500
as well.

1130
00:52:44,500 --> 00:52:45,090
Good.

1131
00:52:45,090 --> 00:52:45,590
All right.

1132
00:52:45,590 --> 00:52:51,480
I think we want to move on from
the sticky flies to BMW i3.

1133
00:52:51,480 --> 00:52:55,170
So we had a little discussion
here during the break.

1134
00:52:55,170 --> 00:52:56,100
[? Lucy, ?] go for it.

1135
00:52:58,950 --> 00:53:02,700
AUDIENCE: So the car shell
meet environmental regulations

1136
00:53:02,700 --> 00:53:08,270
through mechanical or software
tweaking with a proposal.

1137
00:53:08,270 --> 00:53:10,410
PROFESSOR: Could you
guys hear this at EPFL?

1138
00:53:10,410 --> 00:53:12,010
You've been following
the news, right?

1139
00:53:12,010 --> 00:53:12,990
With Volkswagon.

1140
00:53:14,390 --> 00:53:16,180
So what happened?

1141
00:53:16,180 --> 00:53:20,190
Why did Volks-- the CEO
just step down, right?

1142
00:53:20,190 --> 00:53:21,720
It's kind of a big deal.

1143
00:53:21,720 --> 00:53:23,205
Trust in the company has eroded.

1144
00:53:26,400 --> 00:53:29,100
What happened there?

1145
00:53:29,100 --> 00:53:33,130
AUDIENCE: So the requirements
that were set at the start.

1146
00:53:33,130 --> 00:53:36,030
So the car shell meet
environmental regulations

1147
00:53:36,030 --> 00:53:37,930
on emissions.

1148
00:53:37,930 --> 00:53:40,290
This requirement was
probably not met.

1149
00:53:40,290 --> 00:53:43,140
But the realization that it
wasn't met was probably too

1150
00:53:43,140 --> 00:53:46,770
late in the design process,
such that it cannot be changed.

1151
00:53:46,770 --> 00:53:51,540
And so the easy or cheap way to
meet environmental regulations,

1152
00:53:51,540 --> 00:53:55,272
in that case, was to tweak
the software, I guess.

1153
00:53:55,272 --> 00:53:55,980
PROFESSOR: Right.

1154
00:53:55,980 --> 00:53:58,660
So by the way this is
kind of interesting.

1155
00:53:58,660 --> 00:54:02,110
But how do you measure
environmental compliance

1156
00:54:02,110 --> 00:54:03,525
of emissions for cars?

1157
00:54:03,525 --> 00:54:06,152
Do you know how that's
done in practice?

1158
00:54:06,152 --> 00:54:06,860
Do you guys know?

1159
00:54:09,460 --> 00:54:13,000
AUDIENCE: So it's having
[? the car ?] run on a--

1160
00:54:13,000 --> 00:54:17,350
having the engine run, but
the car is not rolling.

1161
00:54:17,350 --> 00:54:23,200
And there's, behind the
[FRENCH] the exhaust.

1162
00:54:23,200 --> 00:54:26,050
Behind the exhaust, you
measure whatever is emitted.

1163
00:54:26,050 --> 00:54:28,570
PROFESSOR: And it's like
another rolling carpet, right?

1164
00:54:28,570 --> 00:54:32,390
It's a dynamo, basically.

1165
00:54:32,390 --> 00:54:35,980
And the way they do it is they
have so-called drive cycles,

1166
00:54:35,980 --> 00:54:38,890
like the [? FT6 ?] drive cycle.

1167
00:54:38,890 --> 00:54:41,030
You know, the highway,
the city cycle,

1168
00:54:41,030 --> 00:54:42,592
which was a lot of on and off.

1169
00:54:42,592 --> 00:54:44,050
And those drive
cycles are actually

1170
00:54:44,050 --> 00:54:47,590
different in Europe and
the US and other countries.

1171
00:54:47,590 --> 00:54:50,380
Every country has
different drive cycles.

1172
00:54:50,380 --> 00:54:55,360
So that's-- and what is a
drive cycle, essentially?

1173
00:54:55,360 --> 00:54:57,859
What word would we, as
systems engineers, what word

1174
00:54:57,859 --> 00:54:58,650
would we put on it?

1175
00:54:58,650 --> 00:54:59,500
AUDIENCE: CONOPS.

1176
00:54:59,500 --> 00:55:01,810
PROFESSOR: A drive cycle
is just an, over time,

1177
00:55:01,810 --> 00:55:04,862
accelerations velocity
of the car, right?

1178
00:55:04,862 --> 00:55:05,820
STUDENT: It's a CONOPS.

1179
00:55:05,820 --> 00:55:09,370
PROFESSOR: It's a CONOPS
A drive cycle is a CONOPS.

1180
00:55:09,370 --> 00:55:11,980
And so for diesel
engines, this issue that

1181
00:55:11,980 --> 00:55:14,620
happened with
Volkswagen, it's only

1182
00:55:14,620 --> 00:55:17,170
an issue with the
diesel, the TDI engines,

1183
00:55:17,170 --> 00:55:19,060
right, not the gasoline engine.

1184
00:55:19,060 --> 00:55:23,530
So the issue is that the
CONOPS, the drive cycle,

1185
00:55:23,530 --> 00:55:27,640
in the US with TDI engines,
they couldn't meet it.

1186
00:55:27,640 --> 00:55:30,952
So they came up with
this trick, right?

1187
00:55:30,952 --> 00:55:32,410
Whereas in Europe,
they didn't have

1188
00:55:32,410 --> 00:55:35,590
to do that, because the European
drive cycle, there's a lot more

1189
00:55:35,590 --> 00:55:38,710
diesel engines use
in European cars.

1190
00:55:38,710 --> 00:55:41,440
They didn't have to do that
in Europe, because the CONOPS,

1191
00:55:41,440 --> 00:55:43,960
the drive cycle it's
used for checking

1192
00:55:43,960 --> 00:55:47,260
the environmental compliance
is different in Europe.

1193
00:55:47,260 --> 00:55:48,730
So great.

1194
00:55:48,730 --> 00:55:51,520
Anybody else on the I-3.

1195
00:55:51,520 --> 00:55:54,730
Now the I3 is interesting,
because that's actually

1196
00:55:54,730 --> 00:55:56,740
an electric car, right?

1197
00:55:56,740 --> 00:56:00,640
And so it doesn't have an
engine, a diesel engine

1198
00:56:00,640 --> 00:56:05,280
or a combustion engine, except
if you get the range extender.

1199
00:56:05,280 --> 00:56:06,880
So it's all
electric, but you can

1200
00:56:06,880 --> 00:56:10,630
actually, as an option get
the range extender, which does

1201
00:56:10,630 --> 00:56:11,870
use fuel.

1202
00:56:11,870 --> 00:56:19,140
OK, EPFL, the i3, any
other requirements there?

1203
00:56:19,140 --> 00:56:37,144
AUDIENCE: So we have
four [INAUDIBLE]

1204
00:56:37,144 --> 00:56:37,810
PROFESSOR: Yeah.

1205
00:56:37,810 --> 00:56:53,560
AUDIENCE: As a third one, the
car shall carry [INAUDIBLE]

1206
00:56:53,560 --> 00:56:55,450
PROFESSOR: OK, so recharging.

1207
00:56:55,450 --> 00:56:57,200
The third requirement
is interesting.

1208
00:56:57,200 --> 00:56:59,380
You said of average
build, right?

1209
00:56:59,380 --> 00:57:01,630
That's the word you use.

1210
00:57:01,630 --> 00:57:02,290
AUDIENCE: Yes.

1211
00:57:02,290 --> 00:57:04,000
PROFESSOR: Now that's great.

1212
00:57:04,000 --> 00:57:06,370
That's a human
factors requirement.

1213
00:57:06,370 --> 00:57:10,120
Now of average build is
a little fuzzy, right?

1214
00:57:10,120 --> 00:57:14,110
How would you make
that more verifiable?

1215
00:57:18,840 --> 00:57:44,520
AUDIENCE: [INAUDIBLE]

1216
00:57:44,520 --> 00:57:48,840
PROFESSOR: Yeah, so there's
actually people that, you know,

1217
00:57:48,840 --> 00:57:54,106
the size distribution, male,
female, you know, weight.

1218
00:57:54,106 --> 00:57:55,100
AUDIENCE: [INAUDIBLE]

1219
00:57:55,100 --> 00:57:58,030
PROFESSOR: And [INAUDIBLE]
the metric measurements

1220
00:57:58,030 --> 00:58:01,570
are actually being
recorded and updated.

1221
00:58:01,570 --> 00:58:03,460
People are getting
taller around the world.

1222
00:58:03,460 --> 00:58:06,050
People are getting heavier.

1223
00:58:06,050 --> 00:58:07,209
We know that, right?

1224
00:58:07,209 --> 00:58:09,250
So the way you would write
that third requirement

1225
00:58:09,250 --> 00:58:13,150
to be verifiable is that
the car shall accommodate

1226
00:58:13,150 --> 00:58:19,960
for passengers in a, say,
P10 male or P5 female

1227
00:58:19,960 --> 00:58:23,260
and a P90 male, right?

1228
00:58:23,260 --> 00:58:26,920
And if you write it that way,
then you know the actual weight

1229
00:58:26,920 --> 00:58:29,980
and dimensions of the
human body can actually

1230
00:58:29,980 --> 00:58:33,970
be traced to a database
that's pretty well known.

1231
00:58:33,970 --> 00:58:36,370
So then instead of
saying average build,

1232
00:58:36,370 --> 00:58:40,930
you say it's a PT50, P50 female.

1233
00:58:40,930 --> 00:58:43,480
And that's something
that's very verifiable.

1234
00:58:43,480 --> 00:58:45,688
Does that make sense?

1235
00:58:45,688 --> 00:58:46,596
AUDIENCE: Yes.

1236
00:58:46,596 --> 00:58:47,470
PROFESSOR: OK, great.

1237
00:58:47,470 --> 00:58:49,750
So what about the last one?

1238
00:58:49,750 --> 00:58:53,800
So by the way, here at MIT,
anybody who's been to EPFL?

1239
00:58:53,800 --> 00:58:57,580
Been to the-- [INAUDIBLE] He
has a Master's from there.

1240
00:58:57,580 --> 00:59:01,180
[INAUDIBLE] been there?

1241
00:59:01,180 --> 00:59:03,040
This is a very cool building.

1242
00:59:03,040 --> 00:59:05,470
It's pretty unusual.

1243
00:59:05,470 --> 00:59:08,180
I mean, we have some very
unusual building on campus,

1244
00:59:08,180 --> 00:59:08,680
too, here.

1245
00:59:08,680 --> 00:59:11,410
But if you get a
chance, it's pretty.

1246
00:59:11,410 --> 00:59:14,920
Unusual so let's
hear from you guys.

1247
00:59:14,920 --> 00:59:18,430
So what do you think was a
requirement for the [? Rolex ?]

1248
00:59:18,430 --> 00:59:20,790
center design?

1249
00:59:24,240 --> 00:59:27,370
AUDIENCE: The building should be
a recognizable structure, that

1250
00:59:27,370 --> 00:59:30,565
would be memorable for EPSL.

1251
00:59:30,565 --> 00:59:31,690
PROFESSOR: Ah, interesting.

1252
00:59:31,690 --> 00:59:36,100
A recognizable structure
that should be memorable.

1253
00:59:36,100 --> 00:59:39,400
So that's-- I think I know
where you're going with this.

1254
00:59:39,400 --> 00:59:41,170
But it's a little--

1255
00:59:41,170 --> 00:59:42,580
you know, recognizable.

1256
00:59:42,580 --> 00:59:45,940
Every building is recognizable
in a sense, right?

1257
00:59:45,940 --> 00:59:48,460
I think-- I like it.

1258
00:59:48,460 --> 00:59:50,050
I know where you're
going with this.

1259
00:59:50,050 --> 00:59:54,260
But it's a little
fuzzy still, right?

1260
00:59:54,260 --> 00:59:57,295
So why do you think
it has holes in it?

1261
00:59:57,295 --> 00:59:58,420
Why do you think the holes?

1262
01:00:01,535 --> 01:00:06,260
AUDIENCE: [INAUDIBLE] unique
might be a better word.

1263
01:00:06,260 --> 01:00:08,624
[INAUDIBLE]

1264
01:00:08,624 --> 01:00:09,290
PROFESSOR: Yeah.

1265
01:00:09,290 --> 01:00:11,120
So that's really.

1266
01:00:11,120 --> 01:00:14,450
The holes are, in some
sense, inefficient, right?

1267
01:00:14,450 --> 01:00:16,880
Because you're putting holes
in the middle of a building.

1268
01:00:16,880 --> 01:00:18,770
But they provide
natural lighting.

1269
01:00:18,770 --> 01:00:21,410
And there's a
symbolism there, right?

1270
01:00:21,410 --> 01:00:24,510
There's a symbolism, the
Swiss cheese symbolism.

1271
01:00:24,510 --> 01:00:28,310
So if the holes, in a sense,
have at least two functions,

1272
01:00:28,310 --> 01:00:29,030
right?

1273
01:00:29,030 --> 01:00:29,730
That's good.

1274
01:00:29,730 --> 01:00:30,400
I like that.

1275
01:00:30,400 --> 01:00:31,650
That's very good.

1276
01:00:31,650 --> 01:00:32,480
What else?

1277
01:00:32,480 --> 01:00:37,200
Another example of a requirement
for the [? Rolex ?] center?

1278
01:00:37,200 --> 01:00:39,910
AUDIENCE: You're taking and
making [INAUDIBLE] category.

1279
01:00:39,910 --> 01:01:18,960
[INAUDIBLE]

1280
01:01:18,960 --> 01:01:20,140
PROFESSOR: Good.

1281
01:01:20,140 --> 01:01:21,350
No, that was great.

1282
01:01:21,350 --> 01:01:22,970
So you have functional
requirement,

1283
01:01:22,970 --> 01:01:24,770
you had interface requirements.

1284
01:01:24,770 --> 01:01:26,600
You had a lot of
those six categories

1285
01:01:26,600 --> 01:01:28,350
we talked about, right?

1286
01:01:28,350 --> 01:01:28,850
Good.

1287
01:01:28,850 --> 01:01:29,260
Excellent.

1288
01:01:29,260 --> 01:01:29,759
Yeah?

1289
01:01:29,759 --> 01:01:30,950
Go ahead.

1290
01:01:30,950 --> 01:01:33,920
AUDIENCE: So how do you write
a requirement about something

1291
01:01:33,920 --> 01:01:38,330
aesthetic in a way that's
not fuzzy or is there a way

1292
01:01:38,330 --> 01:01:40,730
to do that?

1293
01:01:40,730 --> 01:01:42,810
PROFESSOR: That's
a great question.

1294
01:01:42,810 --> 01:01:45,800
I'm not sure I'm the best
person to answer that.

1295
01:01:45,800 --> 01:01:50,390
What I can't tell you is,
in the automotive world,

1296
01:01:50,390 --> 01:01:54,200
the aesthetics of automobiles
and how people judge

1297
01:01:54,200 --> 01:01:56,900
whether a car is
beautiful or appealing,

1298
01:01:56,900 --> 01:02:00,700
that's actually moved from being
kind of just a very fuzzy thing

1299
01:02:00,700 --> 01:02:02,060
to quite measurable.

1300
01:02:02,060 --> 01:02:04,970
You know, there's
different shapes and then

1301
01:02:04,970 --> 01:02:07,460
the building blocks of
shapes and streamlines.

1302
01:02:07,460 --> 01:02:10,430
Those things are-- that's
really a science today.

1303
01:02:10,430 --> 01:02:13,190
And eventually, you
know, there's ratings.

1304
01:02:13,190 --> 01:02:17,820
These are people rate vehicles
for aesthetics and so forth.

1305
01:02:17,820 --> 01:02:21,140
And usually, it's a five point
scale, like a Likert scale,

1306
01:02:21,140 --> 01:02:24,920
like JD Power and associates
is a very well-known marketing

1307
01:02:24,920 --> 01:02:25,860
firm.

1308
01:02:25,860 --> 01:02:30,980
And so they'll say, this
looks-- this will prob--

1309
01:02:30,980 --> 01:02:32,810
they can actually,
at this point,

1310
01:02:32,810 --> 01:02:35,890
you can show them a-- not
a sketch, but a model.

1311
01:02:35,890 --> 01:02:39,590
And based on past data and
information, they'll tell you,

1312
01:02:39,590 --> 01:02:44,190
this car will probably
score between a 4.2 and 4.4

1313
01:02:44,190 --> 01:02:48,100
on the 5 point aesthetic
scale, JD Power scale.

1314
01:02:48,100 --> 01:02:50,110
It's really pretty remarkable.

1315
01:02:50,110 --> 01:02:54,270
But I guess the bigger point is
there's some things that really

1316
01:02:54,270 --> 01:02:56,565
delight us, that have
an artistic quality

1317
01:02:56,565 --> 01:02:59,540
and surprise us in
aesthetic quality.

1318
01:02:59,540 --> 01:03:01,200
And it's true.

1319
01:03:01,200 --> 01:03:03,470
That is one of the
tensions is yet--

1320
01:03:03,470 --> 01:03:06,180
system engineering should--
you know, this very precise.

1321
01:03:06,180 --> 01:03:08,440
Write it down.

1322
01:03:08,440 --> 01:03:12,090
Make sure there is no fuzziness
there or as little as possible.

1323
01:03:12,090 --> 01:03:13,860
And then on the
other hand, we want

1324
01:03:13,860 --> 01:03:18,690
delightful, surprising things
that have an artistic nature

1325
01:03:18,690 --> 01:03:19,440
to them.

1326
01:03:19,440 --> 01:03:21,390
And that is
absolutely a tension.

1327
01:03:21,390 --> 01:03:23,980
And we just acknowledge that.

1328
01:03:23,980 --> 01:03:27,180
OK, any questions?

1329
01:03:27,180 --> 01:03:29,290
Comments before we move on?

1330
01:03:29,290 --> 01:03:30,949
AUDIENCE: I had a
question about--?

1331
01:03:30,949 --> 01:03:31,740
PROFESSOR: Hang on.

1332
01:03:31,740 --> 01:03:32,970
Just one second.

1333
01:03:32,970 --> 01:03:35,400
OK, go ahead.

1334
01:03:35,400 --> 01:03:37,675
At EPFL?

1335
01:03:37,675 --> 01:03:40,050
AUDIENCE: And to which extent
can we refer [? to norms ?]

1336
01:03:40,050 --> 01:03:42,301
and [? loads ?] in
the requirements?

1337
01:03:42,301 --> 01:03:43,800
PROFESSOR: Yeah,
you should do that.

1338
01:03:43,800 --> 01:03:48,570
So compliance, being compliant
with standards and just

1339
01:03:48,570 --> 01:03:51,390
to make clear, standards
and regulations

1340
01:03:51,390 --> 01:03:53,320
are not the same, right?

1341
01:03:53,320 --> 01:03:57,290
A standard, like an IEEE
standard or an ISO standard,

1342
01:03:57,290 --> 01:03:58,300
it's not a law.

1343
01:03:58,300 --> 01:03:59,760
It's not a legal thing.

1344
01:03:59,760 --> 01:04:04,140
It's-- a standard it is
something that maybe a group

1345
01:04:04,140 --> 01:04:07,320
of companies or a group of
organizations have agreed to.

1346
01:04:07,320 --> 01:04:09,960
This is how we
will do it, right?

1347
01:04:09,960 --> 01:04:13,320
The IEEE Wi-Fi
standard, what is it?

1348
01:04:13,320 --> 01:04:17,050
802.11g.

1349
01:04:17,050 --> 01:04:18,030
And that's not a law.

1350
01:04:18,030 --> 01:04:19,260
That's a standard.

1351
01:04:19,260 --> 01:04:22,720
And if you're going to
be 802.11g compliant,

1352
01:04:22,720 --> 01:04:25,410
you write that in
the requirements.

1353
01:04:25,410 --> 01:04:28,460
The environmental emissions
standards we talked about,

1354
01:04:28,460 --> 01:04:30,360
those are actually laws.

1355
01:04:30,360 --> 01:04:33,570
If you are going to sell
a vehicle in country x,

1356
01:04:33,570 --> 01:04:36,750
it has to comply legally
with these regulations.

1357
01:04:36,750 --> 01:04:39,300
But you're absolutely
right, if you

1358
01:04:39,300 --> 01:04:41,460
need to comply
with these things,

1359
01:04:41,460 --> 01:04:43,620
it needs to be part of
the requirements set,

1360
01:04:43,620 --> 01:04:48,680
because otherwise, it will
not happen just automatically.

1361
01:04:48,680 --> 01:04:49,965
[? Weston? ?]

1362
01:04:49,965 --> 01:04:51,840
AUDIENCE: Yeah, it's
kind of along that vein.

1363
01:04:51,840 --> 01:04:55,151
If you have to
comply, say, with ADA,

1364
01:04:55,151 --> 01:04:57,150
instead you have to be
called out, specifically,

1365
01:04:57,150 --> 01:04:59,940
or do you say must comply with
building codes or building

1366
01:04:59,940 --> 01:05:00,450
laws?

1367
01:05:00,450 --> 01:05:02,280
And it's sort of a
blanket requirement?

1368
01:05:02,280 --> 01:05:03,904
PROFESSOR: No, you
have to be specific,

1369
01:05:03,904 --> 01:05:07,320
because building codes are,
first of all, a lot of these

1370
01:05:07,320 --> 01:05:09,280
are local, as well.

1371
01:05:09,280 --> 01:05:11,670
And it's very chaotic.

1372
01:05:11,670 --> 01:05:15,730
So and some of these
might be conflicting.

1373
01:05:15,730 --> 01:05:18,960
So you should be as
specific as possible.

1374
01:05:18,960 --> 01:05:21,900
OK, let's move on here.

1375
01:05:21,900 --> 01:05:25,060
My sense is you're getting it
just for what requirements are.

1376
01:05:25,060 --> 01:05:27,190
And why they're important.

1377
01:05:27,190 --> 01:05:31,290
So let me talk briefly about
requirements, decomposition,

1378
01:05:31,290 --> 01:05:32,850
allocation, and validation.

1379
01:05:32,850 --> 01:05:35,490
This is a figure from
the NASA handbook.

1380
01:05:35,490 --> 01:05:39,750
And basically, what it talks
about is the high level system

1381
01:05:39,750 --> 01:05:43,050
functional requirements
are broken down

1382
01:05:43,050 --> 01:05:45,120
into the performance
requirements.

1383
01:05:45,120 --> 01:05:47,040
And then as you make
design decisions,

1384
01:05:47,040 --> 01:05:50,460
as you decompose your system
into different subsystems,

1385
01:05:50,460 --> 01:05:53,850
each subsystem will have its
own functional and performance

1386
01:05:53,850 --> 01:05:55,020
requirements.

1387
01:05:55,020 --> 01:05:58,650
And then the important thing
is the difference between

1388
01:05:58,650 --> 01:06:01,290
allocated and
derived requirements.

1389
01:06:01,290 --> 01:06:03,930
So allocated requirements
are requirements

1390
01:06:03,930 --> 01:06:08,220
that you choose to allocate.

1391
01:06:08,220 --> 01:06:10,620
And then derived
requirements are calculated

1392
01:06:10,620 --> 01:06:13,590
based on the dependent
requirements,

1393
01:06:13,590 --> 01:06:15,240
based on the allocated
requirements.

1394
01:06:15,240 --> 01:06:18,120
And I'll get into
this in a minute.

1395
01:06:18,120 --> 01:06:19,740
So requirements
are hierarchical.

1396
01:06:19,740 --> 01:06:21,330
We talked about this.

1397
01:06:21,330 --> 01:06:24,480
Functional performance
requirements are allocated.

1398
01:06:24,480 --> 01:06:28,800
And then from these, we can
further decompose and derive

1399
01:06:28,800 --> 01:06:29,637
requirements.

1400
01:06:29,637 --> 01:06:31,470
And then the total set
of these requirements

1401
01:06:31,470 --> 01:06:34,380
needs to be verified
and then validated

1402
01:06:34,380 --> 01:06:38,500
against the stakeholder
expectations.

1403
01:06:38,500 --> 01:06:41,550
So let me briefly talk
about requirements margins

1404
01:06:41,550 --> 01:06:42,690
management.

1405
01:06:42,690 --> 01:06:46,290
So because you don't
know everything upfront,

1406
01:06:46,290 --> 01:06:47,640
there's uncertainty.

1407
01:06:47,640 --> 01:06:50,820
We typically build in reserves
into our requirements.

1408
01:06:50,820 --> 01:06:52,270
And those are called margins.

1409
01:06:52,270 --> 01:06:55,200
So we put in margins
for mass, power, maybe

1410
01:06:55,200 --> 01:06:57,460
memory in computer systems.

1411
01:06:57,460 --> 01:07:00,290
So margins are
essentially reserves

1412
01:07:00,290 --> 01:07:02,940
that are not allocated
to particular subsystems,

1413
01:07:02,940 --> 01:07:05,580
but are controlled by
the project managers

1414
01:07:05,580 --> 01:07:07,270
or at a higher level.

1415
01:07:07,270 --> 01:07:10,140
So the idea is that you
write the requirement

1416
01:07:10,140 --> 01:07:14,070
in a way that is a little bit
more stringent than it really

1417
01:07:14,070 --> 01:07:16,810
needs to be.

1418
01:07:16,810 --> 01:07:19,230
And then by being
more stringent,

1419
01:07:19,230 --> 01:07:21,840
you've built some
reserve into the system

1420
01:07:21,840 --> 01:07:24,360
to handle unexpected things.

1421
01:07:24,360 --> 01:07:27,540
So I'll just give you the
example with mass growth.

1422
01:07:27,540 --> 01:07:30,480
You know, this is very
typical in aerospace vehicles.

1423
01:07:30,480 --> 01:07:34,560
Mass growth can range
between, here, 10 to 60%.

1424
01:07:34,560 --> 01:07:35,920
I'll show you some examples.

1425
01:07:35,920 --> 01:07:38,910
And a lot of it depends on
the novelty of the project.

1426
01:07:38,910 --> 01:07:42,930
So a typical guideline,
specifically for mass margins,

1427
01:07:42,930 --> 01:07:50,220
is about 30% reserve at
the SRR, 20% PDR, 10% CPR,

1428
01:07:50,220 --> 01:07:55,170
and keep about 5% right
before you operate the system.

1429
01:07:55,170 --> 01:07:58,380
IOC is initial
operating capability.

1430
01:07:58,380 --> 01:08:00,960
So here's some historical data.

1431
01:08:00,960 --> 01:08:05,400
This is for manned
or crude vehicles,

1432
01:08:05,400 --> 01:08:07,710
starting with Mercury,
Gemini, Apollo,

1433
01:08:07,710 --> 01:08:09,840
Skylab, and then the shuttle.

1434
01:08:09,840 --> 01:08:14,760
You can see the mass growth
from the concept stage, which

1435
01:08:14,760 --> 01:08:18,990
is phase A or prephase A,
all the way to operations,

1436
01:08:18,990 --> 01:08:21,750
you know, between 10 and 60%.

1437
01:08:21,750 --> 01:08:25,740
The new Orion spacecraft is
not yet included on this chart.

1438
01:08:25,740 --> 01:08:28,020
So what you do is
you essentially write

1439
01:08:28,020 --> 01:08:31,979
the requirement, knowing
that this growth will happen

1440
01:08:31,979 --> 01:08:33,450
during the design process.

1441
01:08:33,450 --> 01:08:36,569
That's the basic
idea of margins.

1442
01:08:36,569 --> 01:08:38,010
And then the next
thing you get is

1443
01:08:38,010 --> 01:08:41,100
that you monitor
the satisfaction

1444
01:08:41,100 --> 01:08:44,020
of the requirements
during the design process.

1445
01:08:44,020 --> 01:08:47,510
And so let's say you have,
in a performance requirement,

1446
01:08:47,510 --> 01:08:49,830
you say that the
system shall not

1447
01:08:49,830 --> 01:08:54,090
emit more than x number
of [? NOX ?] emissions,

1448
01:08:54,090 --> 01:08:58,000
or the systems shall not be
heavier than such and such.

1449
01:08:58,000 --> 01:08:59,270
That's on the y-axis here.

1450
01:08:59,270 --> 01:09:01,649
That's your technical
performance measure.

1451
01:09:01,649 --> 01:09:04,020
And then you move through
time and, you know,

1452
01:09:04,020 --> 01:09:06,450
as the design gets
more detailed,

1453
01:09:06,450 --> 01:09:07,620
usually it gets heavier.

1454
01:09:07,620 --> 01:09:09,779
You add more things.

1455
01:09:09,779 --> 01:09:11,430
And you monitor that.

1456
01:09:11,430 --> 01:09:14,340
And then you have your
final current estimate

1457
01:09:14,340 --> 01:09:17,430
of where you will be at
the end of the project.

1458
01:09:17,430 --> 01:09:21,630
And as you do that, your limits,
your reserves, your margins,

1459
01:09:21,630 --> 01:09:24,420
and here there is an upper
and lower margin shown,

1460
01:09:24,420 --> 01:09:29,560
is going to be narrowing down.

1461
01:09:29,560 --> 01:09:31,080
I'll give you a quick example.

1462
01:09:31,080 --> 01:09:36,390
When I was at McDonnell Douglas,
the F18 EF version, the Super

1463
01:09:36,390 --> 01:09:38,250
Hornet, was being developed.

1464
01:09:38,250 --> 01:09:42,240
And the key thing
there was GTOW,

1465
01:09:42,240 --> 01:09:45,899
GTOW, gross take off weight.

1466
01:09:45,899 --> 01:09:50,350
It's basically the weight of
the plane with crew, fuel,

1467
01:09:50,350 --> 01:09:53,960
any payloads all in.

1468
01:09:53,960 --> 01:09:57,450
The gross take off weight of
the airplane, very important,

1469
01:09:57,450 --> 01:09:59,370
because it determines the range.

1470
01:09:59,370 --> 01:10:00,840
It determines a lot of things.

1471
01:10:00,840 --> 01:10:03,480
And the number, not
to exceed number,

1472
01:10:03,480 --> 01:10:08,020
which is shown here on this
chart as the maximum contract

1473
01:10:08,020 --> 01:10:11,200
or allocated
requirements threshold,

1474
01:10:11,200 --> 01:10:12,780
was contractually specified.

1475
01:10:12,780 --> 01:10:17,160
Not only that, but there were
penalties, financial penalties,

1476
01:10:17,160 --> 01:10:21,960
associated with every kilogram
that you would be over.

1477
01:10:21,960 --> 01:10:23,160
So this was a big deal.

1478
01:10:23,160 --> 01:10:25,170
And they had a
huge the wall chart

1479
01:10:25,170 --> 01:10:27,000
in the hallway,
the main hallway,

1480
01:10:27,000 --> 01:10:28,680
of the engineering building.

1481
01:10:28,680 --> 01:10:32,580
And every day somebody would
actually manually update

1482
01:10:32,580 --> 01:10:35,569
that day's best estimate of
what the gross take off weight

1483
01:10:35,569 --> 01:10:36,610
of the airplane would be.

1484
01:10:36,610 --> 01:10:38,890
And it's like a
Brownian motion thing.

1485
01:10:38,890 --> 01:10:42,320
And as soon as you hit
some critical threshold,

1486
01:10:42,320 --> 01:10:44,070
you would see there'd
be people with would

1487
01:10:44,070 --> 01:10:46,590
come together and say,
we've got to take weight out

1488
01:10:46,590 --> 01:10:49,290
of the airplane
somehow, again, right?

1489
01:10:49,290 --> 01:10:52,410
And then basically
try to get the design

1490
01:10:52,410 --> 01:10:55,720
to comply with that requirement
as you move through the design

1491
01:10:55,720 --> 01:10:56,400
process.

1492
01:10:56,400 --> 01:10:59,280
You can't do that with too
many of the requirements.

1493
01:10:59,280 --> 01:11:01,870
But the most important
technical performance

1494
01:11:01,870 --> 01:11:03,870
measures, that's what you do.

1495
01:11:03,870 --> 01:11:06,510
It's a big deal.

1496
01:11:06,510 --> 01:11:09,609
So here's the flow chart,
basically, for requirements.

1497
01:11:09,609 --> 01:11:11,400
I'm not going to go
through this in detail.

1498
01:11:11,400 --> 01:11:14,940
But the basic idea is the
inputs come from the stakeholder

1499
01:11:14,940 --> 01:11:18,410
expectations, the stakeholder
work we talked about a lot

1500
01:11:18,410 --> 01:11:19,440
last time.

1501
01:11:19,440 --> 01:11:21,840
You go through the
requirements definition process

1502
01:11:21,840 --> 01:11:25,860
and outcomes invalidated set
of technical requirements,

1503
01:11:25,860 --> 01:11:27,930
measures of performance
that you can measure,

1504
01:11:27,930 --> 01:11:29,820
and then these
technical performance

1505
01:11:29,820 --> 01:11:36,330
measures that you can then track
and validate against later.

1506
01:11:36,330 --> 01:11:38,850
Here's a question that
I often get asked.

1507
01:11:38,850 --> 01:11:41,010
Well, OK, so we write
these requirements.

1508
01:11:41,010 --> 01:11:43,440
You guys write them on
paper or your tablet.

1509
01:11:43,440 --> 01:11:47,250
But how do these requirements
actually get recorded, right?

1510
01:11:47,250 --> 01:11:48,360
And managed.

1511
01:11:48,360 --> 01:11:50,917
So I would say
there's sort of two--

1512
01:11:50,917 --> 01:11:52,500
there's the low cost,
and then there's

1513
01:11:52,500 --> 01:11:55,640
sort of the professional
version of doing this.

1514
01:11:55,640 --> 01:11:58,230
The easy way to do it
is you just write them,

1515
01:11:58,230 --> 01:12:00,420
you capture them in a document.

1516
01:12:00,420 --> 01:12:04,530
So that means Microsoft
Word, Excel, Google

1517
01:12:04,530 --> 01:12:09,930
Docs, just a document, a well
written, organized document.

1518
01:12:09,930 --> 01:12:12,060
And then you capture and
revise your requirements.

1519
01:12:12,060 --> 01:12:14,640
And the one thing I
strongly recommend

1520
01:12:14,640 --> 01:12:18,110
is using hyperlinks
to link requirements.

1521
01:12:18,110 --> 01:12:22,890
And this is the idea
that every requirement

1522
01:12:22,890 --> 01:12:25,044
has to be linked to
some other requirement.

1523
01:12:25,044 --> 01:12:27,210
So if it's a low level
requirement, you want to ask,

1524
01:12:27,210 --> 01:12:29,530
well, where does that come from?

1525
01:12:29,530 --> 01:12:31,020
Why did we write
that requirement?

1526
01:12:31,020 --> 01:12:33,240
Well, it has a
parent requirement.

1527
01:12:33,240 --> 01:12:35,250
You want to have
a hyperlink there

1528
01:12:35,250 --> 01:12:37,270
to get you from
one to the other.

1529
01:12:37,270 --> 01:12:39,030
And I have a little
example for that.

1530
01:12:41,860 --> 01:12:47,430
And I think this is OK for
smaller projects, where

1531
01:12:47,430 --> 01:12:50,070
you have dozens or a few
hundred requirements, right?

1532
01:12:50,070 --> 01:12:52,590
And so here's my rule
of thumb for this.

1533
01:12:52,590 --> 01:12:54,720
Do you remember
the magic number 7?

1534
01:12:54,720 --> 01:12:57,300
So we talked last
time, where does

1535
01:12:57,300 --> 01:13:00,000
the world of really
complex systems start.

1536
01:13:00,000 --> 01:13:03,000
And the argument was, well, if
you need more than three levels

1537
01:13:03,000 --> 01:13:04,110
of the decomposition.

1538
01:13:04,110 --> 01:13:05,140
What does that mean?

1539
01:13:05,140 --> 01:13:08,580
Well, 7 plus, minus
2 to the third power

1540
01:13:08,580 --> 01:13:13,890
is somewhere between
125 and 729, right?

1541
01:13:13,890 --> 01:13:16,980
So if you're sort of in
that world or fewer, right?

1542
01:13:16,980 --> 01:13:21,020
If a few hundred, a few dozen,
or few hundred requirements

1543
01:13:21,020 --> 01:13:23,640
kind of what we have in CanSat.

1544
01:13:23,640 --> 01:13:26,130
It's OK to do it this way.

1545
01:13:26,130 --> 01:13:28,241
It's still going to be
a lot of requirements.

1546
01:13:28,241 --> 01:13:29,490
And you have to do a good job.

1547
01:13:29,490 --> 01:13:31,530
But you can do it that way.

1548
01:13:31,530 --> 01:13:34,710
If you have more than that,
and that typically means

1549
01:13:34,710 --> 01:13:37,740
more than 1,000
requirements to handle,

1550
01:13:37,740 --> 01:13:41,880
and there are big projects
that have 5,000, 10,000

1551
01:13:41,880 --> 01:13:46,260
requirements, there is no way
you can manage that effectively

1552
01:13:46,260 --> 01:13:48,720
in a kind of document based way.

1553
01:13:48,720 --> 01:13:51,180
So what you need
then is a database.

1554
01:13:51,180 --> 01:13:52,980
You basically capture
the requirements

1555
01:13:52,980 --> 01:13:55,620
in a relational
database where that

1556
01:13:55,620 --> 01:13:57,400
allows you to link
each requirement

1557
01:13:57,400 --> 01:13:58,480
to other requirements.

1558
01:13:58,480 --> 01:14:02,430
And so this is not meant
as an advertisement,

1559
01:14:02,430 --> 01:14:06,570
but one of the most heavily
used requirements tools

1560
01:14:06,570 --> 01:14:08,880
out there is called, DOORS.

1561
01:14:08,880 --> 01:14:14,100
And this was relatively
recently bought by IBM.

1562
01:14:14,100 --> 01:14:15,360
This was a separate company.

1563
01:14:15,360 --> 01:14:17,280
It was bought by
IBM and included it

1564
01:14:17,280 --> 01:14:23,250
in a suite of software tools
called IBM Rational for System

1565
01:14:23,250 --> 01:14:24,540
Development.

1566
01:14:24,540 --> 01:14:26,400
And so DOORS allows you.

1567
01:14:26,400 --> 01:14:28,170
It's a database,
relational database,

1568
01:14:28,170 --> 01:14:30,450
that allows you to write
requirements, share

1569
01:14:30,450 --> 01:14:31,110
requirements.

1570
01:14:31,110 --> 01:14:33,540
And the latest version
of DOORS is actually

1571
01:14:33,540 --> 01:14:36,570
web based, so you can
have people in India,

1572
01:14:36,570 --> 01:14:39,990
in Europe, in the US, you're
co-developing a system.

1573
01:14:39,990 --> 01:14:42,390
You can all write
requirements and manage them

1574
01:14:42,390 --> 01:14:43,890
on this common database, right?

1575
01:14:43,890 --> 01:14:46,340
Because if you have a
document, very quickly

1576
01:14:46,340 --> 01:14:48,000
it's going to be
confusing, as what's

1577
01:14:48,000 --> 01:14:51,690
the latest version, who
has the latest update.

1578
01:14:51,690 --> 01:14:53,501
Version management
becomes a nightmare.

1579
01:14:54,000 --> 01:14:55,890
So just so you're aware of this.

1580
01:14:55,890 --> 01:14:58,037
We're not going to be
using DOORS in this class.

1581
01:14:58,037 --> 01:14:59,370
We'll just do it document based.

1582
01:14:59,370 --> 01:15:01,620
But you know the
rule of thumb here

1583
01:15:01,620 --> 01:15:03,240
is, more than
1,000 requirements,

1584
01:15:03,240 --> 01:15:06,370
you've got to go to some
professional solution.

1585
01:15:06,370 --> 01:15:09,600
So here's a very sort
of trivial example

1586
01:15:09,600 --> 01:15:12,330
of hierarchical
requirements with links.

1587
01:15:12,330 --> 01:15:14,020
Requirement one,
the systems shall

1588
01:15:14,020 --> 01:15:17,700
fit into a volume not
exceeding one cubic meter.

1589
01:15:17,700 --> 01:15:19,160
And then we have
sub-requirements.

1590
01:15:19,160 --> 01:15:24,230
The system's width shall
be between 0.5 and 1 meter.

1591
01:15:24,230 --> 01:15:26,190
The height, the depth.

1592
01:15:26,190 --> 01:15:28,290
The system shall
be made entirely

1593
01:15:28,290 --> 01:15:30,660
from aluminum 60/60 alloy.

1594
01:15:30,660 --> 01:15:32,550
A sub-requirement
here is the system

1595
01:15:32,550 --> 01:15:35,770
shall not contain any
internal voids or cavities.

1596
01:15:35,770 --> 01:15:39,690
Requirement three, the shape
of the system must be a cube.

1597
01:15:39,690 --> 01:15:42,390
And then a sub-requirement,
the angles between the sides

1598
01:15:42,390 --> 01:15:45,670
shall be 90 degrees
plus or minus 1 degree.

1599
01:15:45,670 --> 01:15:48,660
And so I did this here
just in the slides.

1600
01:15:48,660 --> 01:15:52,030
But if you click on this,
it'll actually transport you

1601
01:15:52,030 --> 01:15:53,570
into another requirement.

1602
01:15:53,570 --> 01:15:55,080
This is the requirement four.

1603
01:15:55,080 --> 01:15:58,890
The mass of the system shall not
exceed 2,700 kilograms, right?

1604
01:15:58,890 --> 01:16:04,110
Aluminum has a density
of about 2,700 kilograms

1605
01:16:04,110 --> 01:16:05,550
per cubic meter.

1606
01:16:05,550 --> 01:16:07,710
So I click back, it
transports me back

1607
01:16:07,710 --> 01:16:09,620
to the earlier requirement.

1608
01:16:09,620 --> 01:16:15,150
So the fact that it's made of
aluminum and the volume cannot

1609
01:16:15,150 --> 01:16:18,840
exceed one cubic meter,
then this requirement four,

1610
01:16:18,840 --> 01:16:20,040
the mass shall not exceed.

1611
01:16:20,040 --> 01:16:21,670
It's not an independent
requirement.

1612
01:16:21,670 --> 01:16:24,330
It's an dependent or
derived requirement

1613
01:16:24,330 --> 01:16:25,980
based on the first two.

1614
01:16:25,980 --> 01:16:27,260
And therefore, they're linked.

1615
01:16:27,260 --> 01:16:29,170
Do you see how this works?

1616
01:16:29,170 --> 01:16:32,430
And to really manage
requirements well and then link

1617
01:16:32,430 --> 01:16:34,530
them, use these hyperlinks.

1618
01:16:34,530 --> 01:16:37,930
It's very, very effective.

1619
01:16:37,930 --> 01:16:40,990
So what would be an object that
satisfies these requirements?

1620
01:16:40,990 --> 01:16:43,830
So here's our one
cubic meter envelope.

1621
01:16:43,830 --> 01:16:46,500
So an aluminum cube
with a side length

1622
01:16:46,500 --> 01:16:51,000
of 60 centimeters this
volume and this mass

1623
01:16:51,000 --> 01:16:52,590
will satisfy the requirements.

1624
01:16:52,590 --> 01:16:53,730
It's not the only thing.

1625
01:16:53,730 --> 01:16:58,014
There's a lot of
other geometries

1626
01:16:58,014 --> 01:16:58,930
that would satisfy it.

1627
01:16:58,930 --> 01:17:01,230
But this particular
instantiation

1628
01:17:01,230 --> 01:17:03,700
would satisfy
these requirements.

1629
01:17:03,700 --> 01:17:06,840
So that's the idea
hierarchical requirements,

1630
01:17:06,840 --> 01:17:10,560
linking them to hyperlinks.

1631
01:17:10,560 --> 01:17:13,770
OK, so let me talk briefly
about the challenges now

1632
01:17:13,770 --> 01:17:15,150
of requirements definition.

1633
01:17:15,150 --> 01:17:16,840
And there's-- this is not easy.

1634
01:17:16,840 --> 01:17:18,670
There's a lot of challenges.

1635
01:17:18,670 --> 01:17:21,000
The first one is
requirements allocation.

1636
01:17:21,000 --> 01:17:23,310
You know, there's composition
and flowing requirements

1637
01:17:23,310 --> 01:17:24,930
to the lower levels.

1638
01:17:24,930 --> 01:17:27,510
And then the idea
is that, whatever

1639
01:17:27,510 --> 01:17:30,210
requirements you derive
at a lower level,

1640
01:17:30,210 --> 01:17:33,990
if you satisfy the lower
level requirements, that

1641
01:17:33,990 --> 01:17:37,320
should guarantee that you then
automatically satisfy a higher

1642
01:17:37,320 --> 01:17:40,170
level requirements from
which the lower level

1643
01:17:40,170 --> 01:17:42,180
requirements were derived.

1644
01:17:42,180 --> 01:17:45,590
That's the basic idea of
requirements allocation.

1645
01:17:45,590 --> 01:17:47,390
And so the way you
can think of this,

1646
01:17:47,390 --> 01:17:50,020
graphically, is we start
out at the high level.

1647
01:17:50,020 --> 01:17:51,290
Here's our stakeholders.

1648
01:17:51,290 --> 01:17:53,330
Stakeholder needs requirements.

1649
01:17:53,330 --> 01:17:54,830
This is level 0.

1650
01:17:54,830 --> 01:17:56,660
You set the system boundary.

1651
01:17:56,660 --> 01:17:58,430
What's the lifecycle?

1652
01:17:58,430 --> 01:17:59,730
And then we apply it.

1653
01:17:59,730 --> 01:18:01,640
That's our first application.

1654
01:18:01,640 --> 01:18:04,370
And then we apply it by
decomposing the system

1655
01:18:04,370 --> 01:18:07,080
into its function.

1656
01:18:07,080 --> 01:18:09,087
And so that's the
second application.

1657
01:18:09,087 --> 01:18:10,670
And then you say,
well, how do these--

1658
01:18:10,670 --> 01:18:13,770
how will these companies
be carried out?

1659
01:18:13,770 --> 01:18:15,980
So we define subsystems.

1660
01:18:15,980 --> 01:18:19,250
And essentially, then,
the subsystem requirements

1661
01:18:19,250 --> 01:18:21,320
are derived from the
functional requirements.

1662
01:18:21,320 --> 01:18:25,510
And you put numbers, key
system performance parameters,

1663
01:18:25,510 --> 01:18:26,350
behind these.

1664
01:18:26,350 --> 01:18:28,490
And depending on the
complexity of the system,

1665
01:18:28,490 --> 01:18:30,790
you may have to go
multiple layers down.

1666
01:18:30,790 --> 01:18:33,440
But that's the basic
requirements allocation

1667
01:18:33,440 --> 01:18:35,810
process.

1668
01:18:35,810 --> 01:18:37,250
And I think I said this already.

1669
01:18:37,250 --> 01:18:40,820
It's difficult to do and
stay solution neutral

1670
01:18:40,820 --> 01:18:44,030
the deeper you go into this.

1671
01:18:44,030 --> 01:18:46,910
Let me briefly
mention, this will not

1672
01:18:46,910 --> 01:18:48,990
be, we will not sort
of test you on this.

1673
01:18:48,990 --> 01:18:51,870
But I want to make
you aware of this.

1674
01:18:51,870 --> 01:18:54,620
A methodology called
ISO performance.

1675
01:18:54,620 --> 01:18:59,990
And that was actually the
topic of my dissertation.

1676
01:18:59,990 --> 01:19:04,400
How do you allocate requirements
to lower level parameters

1677
01:19:04,400 --> 01:19:08,000
in systems, when a higher
level requirement is defined,

1678
01:19:08,000 --> 01:19:09,110
is clear.

1679
01:19:09,110 --> 01:19:13,550
So the idea is that you have a
vector of desired performance

1680
01:19:13,550 --> 01:19:14,960
requirements.

1681
01:19:14,960 --> 01:19:17,300
And you want to
understand, first of all,

1682
01:19:17,300 --> 01:19:20,120
you want to understand are
those requirements feasible.

1683
01:19:20,120 --> 01:19:21,920
And if they are
feasible, it usually

1684
01:19:21,920 --> 01:19:24,740
means there is more than one way
that these requirements could

1685
01:19:24,740 --> 01:19:26,990
be satisfied, that the higher
level requirements could

1686
01:19:26,990 --> 01:19:28,230
be satisfied.

1687
01:19:28,230 --> 01:19:32,810
So find different non-unique
feasible combinations

1688
01:19:32,810 --> 01:19:36,260
to satisfy the high
level requirements.

1689
01:19:36,260 --> 01:19:38,870
And so this is one
of the readings.

1690
01:19:38,870 --> 01:19:40,850
Very quickly, we have
our design space.

1691
01:19:40,850 --> 01:19:42,930
We have our objective space.

1692
01:19:42,930 --> 01:19:46,100
And then we have this
cost risk objective space.

1693
01:19:46,100 --> 01:19:48,410
And so the idea is that,
in the performance space,

1694
01:19:48,410 --> 01:19:49,910
you have the shall statements.

1695
01:19:49,910 --> 01:19:52,516
You shall perform at that level.

1696
01:19:52,516 --> 01:19:53,640
And that's this point here.

1697
01:19:53,640 --> 01:19:55,490
That's your performance target.

1698
01:19:55,490 --> 01:20:00,740
And if it's in this gray
area, it's actually feasible.

1699
01:20:00,740 --> 01:20:04,170
And it usually means there's
more than one way to do it.

1700
01:20:04,170 --> 01:20:07,310
So if you can then map
backwards to the design space,

1701
01:20:07,310 --> 01:20:11,180
these points here are
all ISO performance,

1702
01:20:11,180 --> 01:20:15,620
meaning they all
satisfy and provide

1703
01:20:15,620 --> 01:20:16,730
this level of performance.

1704
01:20:16,730 --> 01:20:18,860
But they do it in
different ways.

1705
01:20:18,860 --> 01:20:22,130
And then in order to
select the final design

1706
01:20:22,130 --> 01:20:24,080
or final requirements
set, you want

1707
01:20:24,080 --> 01:20:25,400
to look at other objectives.

1708
01:20:25,400 --> 01:20:27,980
This is where the should
statements come in.

1709
01:20:27,980 --> 01:20:31,310
And these are typically cost
and risk related objectives.

1710
01:20:31,310 --> 01:20:34,400
So let me give you,
this sounds pretty

1711
01:20:34,400 --> 01:20:35,780
abstract and theoretical.

1712
01:20:35,780 --> 01:20:38,120
Let me give you a
very specific example.

1713
01:20:38,120 --> 01:20:42,080
This was my case study,
the Space Telescope.

1714
01:20:42,080 --> 01:20:45,890
This is the actually this has
now become the James Webb Space

1715
01:20:45,890 --> 01:20:46,760
Telescope.

1716
01:20:46,760 --> 01:20:48,300
Hasn't launched yet.

1717
01:20:48,300 --> 01:20:50,220
This is 20 years ago.

1718
01:20:50,220 --> 01:20:52,440
We've been working on the
James Webb for a long time.

1719
01:20:52,440 --> 01:20:54,680
But if it works, it
will be exciting,

1720
01:20:54,680 --> 01:20:58,100
because it's going to get us
very, very close to the Big

1721
01:20:58,100 --> 01:20:58,830
Bang, right?

1722
01:20:58,830 --> 01:21:03,190
That's what James Webb will
look in the highly redshifted

1723
01:21:03,190 --> 01:21:06,380
infrared to really see the
formation of the earliest

1724
01:21:06,380 --> 01:21:08,130
proto-galaxies in the universe.

1725
01:21:08,130 --> 01:21:11,340
So if this works, it
will be very exciting.

1726
01:21:11,340 --> 01:21:15,120
So here's, essentially, a
model of the spacecraft.

1727
01:21:15,120 --> 01:21:17,600
This is a precursor
to James Webb.

1728
01:21:17,600 --> 01:21:20,330
And the big thing
you need to do is

1729
01:21:20,330 --> 01:21:26,420
you need to point and be very,
very stable for a long time

1730
01:21:26,420 --> 01:21:29,100
to take in these very,
very faint images.

1731
01:21:29,100 --> 01:21:32,030
So we have wavefront error
phasing requirements.

1732
01:21:32,030 --> 01:21:33,905
And then we have these
pointing requirements.

1733
01:21:33,905 --> 01:21:37,640
You've got to a point
very stable and in order

1734
01:21:37,640 --> 01:21:40,820
to achieve this optical
pointing performance,

1735
01:21:40,820 --> 01:21:45,740
we have the structure
of the spacecraft.

1736
01:21:45,740 --> 01:21:47,030
We have reaction wheels.

1737
01:21:47,030 --> 01:21:48,430
We have controllers.

1738
01:21:48,430 --> 01:21:51,290
And there is noise, different
kinds of noise sources

1739
01:21:51,290 --> 01:21:55,420
that are trying to basically
prevent us from pointing

1740
01:21:55,420 --> 01:21:57,350
with this precision.

1741
01:21:57,350 --> 01:22:01,970
So very quickly, this is what
this Nexus precursor spacecraft

1742
01:22:01,970 --> 01:22:05,240
looks like in the
deploy configuration,

1743
01:22:05,240 --> 01:22:10,740
here on the upper right in
the stowed configuration.

1744
01:22:10,740 --> 01:22:12,480
Initial performance assessment.

1745
01:22:12,480 --> 01:22:14,660
So this is really trying
to define the requirements

1746
01:22:14,660 --> 01:22:19,540
for deriving the requirements
for the structure,

1747
01:22:19,540 --> 01:22:21,650
for the optics,
for the controller,

1748
01:22:21,650 --> 01:22:23,600
knowing that this is the point.

1749
01:22:23,600 --> 01:22:26,330
If you want to have
this science happen,

1750
01:22:26,330 --> 01:22:28,310
you've got a point
with that precision.

1751
01:22:28,310 --> 01:22:28,910
That is known.

1752
01:22:28,910 --> 01:22:31,970
So let's flow that down.

1753
01:22:31,970 --> 01:22:35,400
So if we look at the pointing,
for example, here on the right,

1754
01:22:35,400 --> 01:22:37,460
this is kind of fuzzy furball.

1755
01:22:37,460 --> 01:22:38,930
What does that represent?

1756
01:22:38,930 --> 01:22:42,400
That's a time domain simulation
of the centroid of the image,

1757
01:22:42,400 --> 01:22:42,900
right?

1758
01:22:42,900 --> 01:22:46,440
The telescope is now
observing a part of the sky,

1759
01:22:46,440 --> 01:22:49,620
trying to get these
early proto-galaxies.

1760
01:22:49,620 --> 01:22:52,110
And it's flexible.

1761
01:22:52,110 --> 01:22:53,740
It's a very
lightweight telescope.

1762
01:22:53,740 --> 01:22:56,160
It has these reaction
wheels are turning,

1763
01:22:56,160 --> 01:22:57,690
trying to keep it stable.

1764
01:22:57,690 --> 01:23:00,130
You have electronic
sources of noise.

1765
01:23:00,130 --> 01:23:04,020
So the blue furball here
predicts that we would not

1766
01:23:04,020 --> 01:23:05,130
meet the requirements.

1767
01:23:05,130 --> 01:23:07,350
This is some initial
design, some initial set

1768
01:23:07,350 --> 01:23:08,520
of requirements.

1769
01:23:08,520 --> 01:23:12,210
We need to find a way
to get down to the--

1770
01:23:12,210 --> 01:23:14,520
this is about 15 microns--

1771
01:23:14,520 --> 01:23:16,210
[? root mean ?] square error.

1772
01:23:16,210 --> 01:23:20,010
So it's about three times worse
than what we need to achieve.

1773
01:23:20,010 --> 01:23:21,900
And the requirement
is we need to get down

1774
01:23:21,900 --> 01:23:25,450
to 5 microns of pointing
or jitter precision.

1775
01:23:25,450 --> 01:23:27,720
So how do we do this?

1776
01:23:27,720 --> 01:23:30,780
So what ISO
performance does is it

1777
01:23:30,780 --> 01:23:33,150
looks at the sensitivities
in the system.

1778
01:23:33,150 --> 01:23:35,400
What are all the
things that influence

1779
01:23:35,400 --> 01:23:38,190
the pointing performance
of the telescope.

1780
01:23:38,190 --> 01:23:40,600
And you can see here.

1781
01:23:40,600 --> 01:23:42,000
These are the two
key performance

1782
01:23:42,000 --> 01:23:46,530
measures on the left, wavefront
error, and then line of sight.

1783
01:23:46,530 --> 01:23:49,320
And the bars
essentially tell you

1784
01:23:49,320 --> 01:23:51,940
what are the really
sensitive parameters that

1785
01:23:51,940 --> 01:23:53,285
drive performance.

1786
01:23:53,285 --> 01:23:56,880
So disturbance parameters,
planned structural parameters,

1787
01:23:56,880 --> 01:23:58,280
optics, and then controls.

1788
01:23:58,280 --> 01:24:00,510
So it's really
multi-disciplinary.

1789
01:24:00,510 --> 01:24:02,640
And what's neat
about it is you can,

1790
01:24:02,640 --> 01:24:04,680
if you know these
sensitivities, you

1791
01:24:04,680 --> 01:24:07,950
can essentially calculate
the Jacobian a matrix.

1792
01:24:07,950 --> 01:24:09,750
The Jacobian matrix
is essentially

1793
01:24:09,750 --> 01:24:12,720
the partial derivatives
of your performance, which

1794
01:24:12,720 --> 01:24:15,210
is your higher level
requirement, with respect

1795
01:24:15,210 --> 01:24:18,570
to these lower level
parameters or requirements.

1796
01:24:18,570 --> 01:24:23,520
And you can then find, using
this essentially the null space

1797
01:24:23,520 --> 01:24:26,010
of the Jacobian matrix,
that will tell you

1798
01:24:26,010 --> 01:24:30,520
how do you move in that space
to keep the performance fixed

1799
01:24:30,520 --> 01:24:33,100
at the requirement level.

1800
01:24:33,100 --> 01:24:37,020
Let me explain this just
using two parameters.

1801
01:24:37,020 --> 01:24:40,140
And this was a big deal in
the Hubble Space Telescope.

1802
01:24:40,140 --> 01:24:41,820
So the two parameters
I'm going to use

1803
01:24:41,820 --> 01:24:46,810
are KR ISO and UD, the
dynamic wheel and balance.

1804
01:24:46,810 --> 01:24:49,080
So these spacecrafts
have reaction wheels

1805
01:24:49,080 --> 01:24:52,620
that are turning to
point the spacecraft

1806
01:24:52,620 --> 01:24:55,610
and counteract any momentum
that you get from, for example,

1807
01:24:55,610 --> 01:24:57,090
from solar pressure.

1808
01:24:57,090 --> 01:25:02,010
So UD is the amount of imbalance
that you have in these wheels.

1809
01:25:02,010 --> 01:25:04,860
If you have imbalance,
that will cause chatter.

1810
01:25:04,860 --> 01:25:08,010
It will cost torques, which
will cause that jitter.

1811
01:25:08,010 --> 01:25:14,460
And then KR ISO is the vibration
isolation of the reaction wheel

1812
01:25:14,460 --> 01:25:15,250
assembly.

1813
01:25:15,250 --> 01:25:18,780
How stiff or soft is
the vibration isolation.

1814
01:25:18,780 --> 01:25:20,340
And what you can
see in this clock

1815
01:25:20,340 --> 01:25:21,870
is that, our initial design--

1816
01:25:21,870 --> 01:25:23,820
this is based on a simulation--

1817
01:25:23,820 --> 01:25:26,970
does not satisfy a
higher level requirement.

1818
01:25:26,970 --> 01:25:29,940
In order to satisfy that, we
need to go down to the blue

1819
01:25:29,940 --> 01:25:35,490
line, labelled with 5
microns appointing precision.

1820
01:25:35,490 --> 01:25:39,060
And what you can see is that
there are different ways

1821
01:25:39,060 --> 01:25:41,370
to achieve that 5 microns.

1822
01:25:41,370 --> 01:25:44,580
We could go over here.

1823
01:25:44,580 --> 01:25:46,530
This is labeled as HST.

1824
01:25:46,530 --> 01:25:49,110
That's what Hubble
Space Telescope is.

1825
01:25:49,110 --> 01:25:51,240
The Hubble Space
Telescope basically

1826
01:25:51,240 --> 01:25:55,470
went for ultra, ultra,
ultra quiet reaction wheels,

1827
01:25:55,470 --> 01:25:59,260
very, very low
dynamic imbalance.

1828
01:25:59,260 --> 01:26:02,910
So [? Marissa, ?] you
talked about buying sensors

1829
01:26:02,910 --> 01:26:04,830
from a supplier and
putting them in.

1830
01:26:04,830 --> 01:26:06,240
These reactors,
there's companies

1831
01:26:06,240 --> 01:26:08,580
that really specialize
in making reaction wheels

1832
01:26:08,580 --> 01:26:10,270
and things like that.

1833
01:26:10,270 --> 01:26:12,480
So if you're going to
go for this point here.

1834
01:26:12,480 --> 01:26:15,990
Let me point that out again,
so you guys can see it at EPFL.

1835
01:26:15,990 --> 01:26:19,140
If you're going to go, if you're
going to derive and allocate

1836
01:26:19,140 --> 01:26:21,840
this requirement, it means
you put a lot of pressure

1837
01:26:21,840 --> 01:26:25,650
on your supplier to achieve
that level of dynamic wheel

1838
01:26:25,650 --> 01:26:27,380
imbalance.

1839
01:26:27,380 --> 01:26:28,830
If they can do it, that's great.

1840
01:26:28,830 --> 01:26:30,600
But it could be very expensive.

1841
01:26:30,600 --> 01:26:35,850
But it makes your job easier
as the spacecraft integrator.

1842
01:26:35,850 --> 01:26:40,740
Or you can go straight down to
this point here called spec.

1843
01:26:40,740 --> 01:26:45,720
So you're essentially tolerating
a noisier reaction wheel.

1844
01:26:45,720 --> 01:26:50,680
But then you need a very, very
soft, very capable isolation

1845
01:26:50,680 --> 01:26:53,790
stage, very soft isolator.

1846
01:26:53,790 --> 01:26:57,462
What's the disadvantage of
having very soft isolation?

1847
01:26:57,462 --> 01:26:58,920
What do you think
is the big issue?

1848
01:27:01,794 --> 01:27:05,630
AUDIENCE: [INAUDIBLE]

1849
01:27:05,630 --> 01:27:06,810
PROFESSOR: So the launch.

1850
01:27:06,810 --> 01:27:10,030
You probably have to lock
it down during launch.

1851
01:27:10,030 --> 01:27:11,950
You have a big displacements.

1852
01:27:11,950 --> 01:27:14,860
You need, if you have a very
soft isolation stage, you

1853
01:27:14,860 --> 01:27:17,140
need volume,
because the isolator

1854
01:27:17,140 --> 01:27:19,320
is going to move a lot, right?

1855
01:27:19,320 --> 01:27:21,760
And you may not
have that volume.

1856
01:27:21,760 --> 01:27:24,760
Or you could do a
little bit of each.

1857
01:27:24,760 --> 01:27:27,070
So this test point here.

1858
01:27:27,070 --> 01:27:30,430
We're going to have quieter
wheels, but not ultra quiet.

1859
01:27:30,430 --> 01:27:32,110
And we're going to
have a soft isolator.

1860
01:27:32,110 --> 01:27:34,970
But we have the right
combination of the two.

1861
01:27:34,970 --> 01:27:37,870
And we're going to go
for that point there.

1862
01:27:37,870 --> 01:27:40,150
So now if you go for
that point, you're

1863
01:27:40,150 --> 01:27:42,590
going to satisfy your
pointing requirement,

1864
01:27:42,590 --> 01:27:45,160
bu you've allocate at the
lower level requirement,

1865
01:27:45,160 --> 01:27:48,910
in this case, to the isolation
stage and the dynamic wheel

1866
01:27:48,910 --> 01:27:50,890
imbalance in a kind
of balanced way.

1867
01:27:50,890 --> 01:27:53,920
So everybody's job is
roughly equally difficult.

1868
01:27:53,920 --> 01:27:55,400
That's the sort of idea here.

1869
01:27:55,400 --> 01:27:56,650
Yes?

1870
01:27:56,650 --> 01:27:59,560
AUDIENCE: Does this model assume
that all of these variables

1871
01:27:59,560 --> 01:28:03,130
are independent
and, I guess, kind

1872
01:28:03,130 --> 01:28:08,000
of continuous in terms of you
can incrementally work better?

1873
01:28:08,000 --> 01:28:10,990
PROFESSOR: So they're not
independent in the sense

1874
01:28:10,990 --> 01:28:13,620
that they're coupled through
the performance of the system.

1875
01:28:13,620 --> 01:28:15,370
But you can choose--
the assumption is you

1876
01:28:15,370 --> 01:28:18,550
can choose these independently.

1877
01:28:18,550 --> 01:28:20,170
But they're coupled,
because you need

1878
01:28:20,170 --> 01:28:21,820
to achieve that requirement.

1879
01:28:21,820 --> 01:28:23,860
And yes, in this case,
they're continuous.

1880
01:28:23,860 --> 01:28:26,960
But you can think of this
as a discrete, as well.

1881
01:28:26,960 --> 01:28:29,440
So for example, the
reaction wheels could be,

1882
01:28:29,440 --> 01:28:33,040
you're picking from a
catalog, in which case

1883
01:28:33,040 --> 01:28:36,940
now you're really picking
specifications for components

1884
01:28:36,940 --> 01:28:40,310
that are off the shelf.

1885
01:28:40,310 --> 01:28:42,010
OK, so let me--

1886
01:28:42,010 --> 01:28:43,580
I'm going to skip here for time.

1887
01:28:43,580 --> 01:28:47,200
So basically, following this
ISO performance process,

1888
01:28:47,200 --> 01:28:50,310
you can go from some initial
requirements, derived

1889
01:28:50,310 --> 01:28:53,940
requirements that
are infeasible, to,

1890
01:28:53,940 --> 01:28:57,170
in blue here I'm showing
you the results of this.

1891
01:28:57,170 --> 01:28:58,547
It's very close to 5.

1892
01:28:58,547 --> 01:29:00,130
There's some numerical
tolerance here.

1893
01:29:00,130 --> 01:29:03,970
But by but essentially
rebalancing the requirements

1894
01:29:03,970 --> 01:29:06,550
within the system, we're
achieving a higher level

1895
01:29:06,550 --> 01:29:07,310
requirement.

1896
01:29:07,310 --> 01:29:09,430
But in a way that
this [INAUDIBLE]

1897
01:29:09,430 --> 01:29:13,940
sort of the challenge
equitably across sub-systems.

1898
01:29:13,940 --> 01:29:15,060
Does that make sense?

1899
01:29:15,060 --> 01:29:17,710
At EPFL, did you
guys follow this?

1900
01:29:17,710 --> 01:29:19,030
I know this was a bit fast.

1901
01:29:19,030 --> 01:29:23,080
But that's the basic
idea of ISO performance.

1902
01:29:23,080 --> 01:29:24,402
Any questions on your side?

1903
01:29:28,340 --> 01:29:31,620
AUDIENCE: [INAUDIBLE]

1904
01:29:31,620 --> 01:29:33,690
AUDIENCE: Maybe I have one.

1905
01:29:33,690 --> 01:29:35,810
So given that you have
[? distance ?] requirements

1906
01:29:35,810 --> 01:29:39,580
and you probably have
[INAUDIBLE] for each one,

1907
01:29:39,580 --> 01:29:41,725
is there a way that we
would integrate all those

1908
01:29:41,725 --> 01:29:46,550
[? lines, ?] that you could
see the [INAUDIBLE] for those?

1909
01:29:46,550 --> 01:29:48,300
If you don't mind,
maybe it's a little bit

1910
01:29:48,300 --> 01:29:50,674
[? softer, ?] but maybe there's
another requirement where

1911
01:29:50,674 --> 01:29:55,284
[INAUDIBLE]

1912
01:29:58,220 --> 01:30:02,508
PROFESSOR: [INAUDIBLE]

1913
01:30:02,508 --> 01:30:04,472
AUDIENCE: We don't hear you.

1914
01:30:21,190 --> 01:30:23,530
PROFESSOR: Sorry, guys,
ran out of battery here.

1915
01:30:23,530 --> 01:30:24,979
Can you hear me again?

1916
01:30:27,880 --> 01:30:29,240
Can you hear me?

1917
01:30:29,240 --> 01:30:30,540
Yeah?

1918
01:30:30,540 --> 01:30:34,260
So the question was,
well if you have multiple

1919
01:30:34,260 --> 01:30:37,230
of these performance, ISO
performance lines, that's

1920
01:30:37,230 --> 01:30:38,090
exactly right.

1921
01:30:38,090 --> 01:30:41,220
If you have a vector of
higher level requirements,

1922
01:30:41,220 --> 01:30:44,039
you're going to have ISO
performance surfaces.

1923
01:30:44,039 --> 01:30:45,830
And so it gets, you
know, the more of these

1924
01:30:45,830 --> 01:30:49,660
you have to satisfy at the
same time, the more complex

1925
01:30:49,660 --> 01:30:50,310
this gets.

1926
01:30:50,310 --> 01:30:53,010
But we can handle
that computationally.

1927
01:30:53,010 --> 01:30:55,920
So you have to do some
simulation and computation

1928
01:30:55,920 --> 01:30:58,140
upfront to make sure you pick.

1929
01:30:58,140 --> 01:31:02,170
These requirements are
not just wild guesses.

1930
01:31:02,170 --> 01:31:05,360
They're actually based on some
calculation and simulation.

1931
01:31:05,360 --> 01:31:07,590
OK?

1932
01:31:07,590 --> 01:31:08,220
All right.

1933
01:31:08,220 --> 01:31:10,620
So let's do this.

1934
01:31:10,620 --> 01:31:12,990
Let's do this in a
kind of simplified way.

1935
01:31:12,990 --> 01:31:17,200
This is going to be our last
concept question for today.

1936
01:31:17,200 --> 01:31:21,030
Here's the higher level
requirement is a balloon.

1937
01:31:21,030 --> 01:31:23,941
Those of you that have done
unified engineering here,

1938
01:31:23,941 --> 01:31:24,940
you've done this, right?

1939
01:31:24,940 --> 01:31:26,460
We did this in unified.

1940
01:31:26,460 --> 01:31:29,850
A balloon shall lift the payload
of 1,000 kilograms, which

1941
01:31:29,850 --> 01:31:33,030
includes its own mass.

1942
01:31:33,030 --> 01:31:36,780
You can use either helium,
which has a density of 0.2

1943
01:31:36,780 --> 01:31:38,260
kilograms per cubic meters.

1944
01:31:38,260 --> 01:31:40,200
This is in standard
atmosphere conditions.

1945
01:31:40,200 --> 01:31:42,240
Or hydrogen, 0.1.

1946
01:31:42,240 --> 01:31:45,230
I've rounded these
numbers, as a lift gas.

1947
01:31:45,230 --> 01:31:50,400
The standard density
of air is 1.3.

1948
01:31:50,400 --> 01:31:54,360
Which of the following
requirements is infeasible?

1949
01:31:54,360 --> 01:31:56,340
A, the balloon
shall have a radius

1950
01:31:56,340 --> 01:31:58,920
of 6.1 meters and
the balloon shall

1951
01:31:58,920 --> 01:32:02,190
use 99.9 percent pure helium.

1952
01:32:02,190 --> 01:32:05,800
B, a radius of 5.9
and the balloon

1953
01:32:05,800 --> 01:32:11,310
shall use 99.99 point percent
hydrogen as a lift gas.

1954
01:32:11,310 --> 01:32:16,920
And then 5.9 meters helium,
6.1 meters hydrogen,

1955
01:32:16,920 --> 01:32:19,620
all these requirements
actually are OK.

1956
01:32:19,620 --> 01:32:23,320
Or none of these
requirements are feasible.

1957
01:32:23,320 --> 01:32:28,080
OK, so I'll give you I'll
give you three minutes

1958
01:32:28,080 --> 01:32:31,110
to try and figure this out.

1959
01:32:31,110 --> 01:32:34,090
And then I'll show
you the solution.

1960
01:32:34,090 --> 01:32:34,590
Don't cheat.

1961
01:32:34,590 --> 01:32:35,714
Don't go to the next chart.

1962
01:32:39,610 --> 01:32:42,910
So think about this the
high level requirement is

1963
01:32:42,910 --> 01:32:45,520
the balloon shall
lift 1,000 kilograms,

1964
01:32:45,520 --> 01:32:47,410
including its own mass.

1965
01:32:47,410 --> 01:32:51,010
And we're trying to allocate
lower level requirements

1966
01:32:51,010 --> 01:32:54,970
for the size of the
balloon and the gas

1967
01:32:54,970 --> 01:32:58,275
that we're going to use.

1968
01:32:58,275 --> 01:32:59,772
So try to figure this out.

1969
01:33:05,699 --> 01:33:07,240
All right, I think
we probably need--

1970
01:33:07,240 --> 01:33:08,740
who needs more time for this?

1971
01:33:11,440 --> 01:33:15,740
OK, so we're going to leave
this as a cliffhanger, OK?

1972
01:33:15,740 --> 01:33:18,530
We'll do the solution next time.

1973
01:33:18,530 --> 01:33:21,380
And take a little time
to figure this out.

1974
01:33:21,380 --> 01:33:23,990
Try not to look at
the next slide, OK?

1975
01:33:23,990 --> 01:33:27,440
So I'm going to skip
this just for time.

1976
01:33:27,440 --> 01:33:30,950
And I want to talk briefly about
the SRR, what it is, and then

1977
01:33:30,950 --> 01:33:32,630
kick off assignment 2.

1978
01:33:32,630 --> 01:33:33,890
So SRR.

1979
01:33:33,890 --> 01:33:36,500
So the idea is you've
really been working

1980
01:33:36,500 --> 01:33:38,990
on these requirements
hard, you know, as a team,

1981
01:33:38,990 --> 01:33:40,309
with your customers.

1982
01:33:40,309 --> 01:33:42,100
And this takes-- this
is not a quick thing.

1983
01:33:42,100 --> 01:33:43,558
This is not just
two or three days.

1984
01:33:43,558 --> 01:33:46,520
Typically, writing
your level 0, level

1985
01:33:46,520 --> 01:33:51,470
1 requirements is a process
that takes weeks, months,

1986
01:33:51,470 --> 01:33:54,980
sometimes at least a year.

1987
01:33:54,980 --> 01:33:58,490
But typically, it's on the
order of three to six months

1988
01:33:58,490 --> 01:34:00,620
in many projects, roughly.

1989
01:34:00,620 --> 01:34:03,570
And once you have your
high level requirements,

1990
01:34:03,570 --> 01:34:07,180
SRR is the milestone where
you review those, right?

1991
01:34:07,180 --> 01:34:09,710
So here's an example.

1992
01:34:09,710 --> 01:34:11,300
So it's a social.

1993
01:34:11,300 --> 01:34:13,220
It's a peer review process.

1994
01:34:13,220 --> 01:34:16,700
And the main goal of SBAR
is to vet the requirements

1995
01:34:16,700 --> 01:34:19,100
as they were written
to see if you have

1996
01:34:19,100 --> 01:34:22,770
any missing, misstated,
redundant, or otherwise

1997
01:34:22,770 --> 01:34:24,870
unsatisfactory requirements.

1998
01:34:24,870 --> 01:34:29,690
This is a picture from JPL,
from MSL, actually, mission.

1999
01:34:29,690 --> 01:34:32,090
And I want to just point
out a good friend of mine

2000
01:34:32,090 --> 01:34:35,280
who is a graduate here of our
department, Richard Kornfeld.

2001
01:34:35,280 --> 01:34:38,030
[? Voelker ?] knows him,
met him last summer.

2002
01:34:38,030 --> 01:34:40,220
He's been working on
three different Mars

2003
01:34:40,220 --> 01:34:42,021
missions over the last decade.

2004
01:34:42,021 --> 01:34:43,520
And we're going to
hear from Richard

2005
01:34:43,520 --> 01:34:47,180
later this semester about
verification of requirements,

2006
01:34:47,180 --> 01:34:49,400
because he verified the
entry descent and landing

2007
01:34:49,400 --> 01:34:51,770
requirements for MSL.

2008
01:34:51,770 --> 01:34:57,040
So here's essentially what
it says in the NASA handbook.

2009
01:34:57,040 --> 01:35:01,490
SRR happens during phase A,
before you go into phase B.

2010
01:35:01,490 --> 01:35:04,940
And you need at least your top
two level requirements written

2011
01:35:04,940 --> 01:35:06,685
before you can do the SRR.

2012
01:35:09,410 --> 01:35:12,050
I do want to mention the
second reading post reading.

2013
01:35:12,050 --> 01:35:14,600
This is about
requirements volatility.

2014
01:35:14,600 --> 01:35:19,160
So just because you
past the SRR doesn't

2015
01:35:19,160 --> 01:35:22,010
mean that the requirements
are completely frozen.

2016
01:35:22,010 --> 01:35:24,860
First of all, you don't have a
lot of lower level requirements

2017
01:35:24,860 --> 01:35:25,400
yet, right?

2018
01:35:25,400 --> 01:35:28,610
Those are going to
be added post SRR.

2019
01:35:28,610 --> 01:35:30,020
and even some of the--

2020
01:35:30,020 --> 01:35:33,302
you pointed out, some of
you, the requirements creep

2021
01:35:33,302 --> 01:35:34,010
and these things.

2022
01:35:34,010 --> 01:35:38,540
So this second reading is about
requirements volatility, which

2023
01:35:38,540 --> 01:35:40,220
you can actually quantify.

2024
01:35:40,220 --> 01:35:42,890
What are the sources of
requirements volatility?

2025
01:35:42,890 --> 01:35:44,810
That's what this
figure shows you.

2026
01:35:44,810 --> 01:35:47,810
And then what is the impact
of requirements volatility?

2027
01:35:50,580 --> 01:35:52,610
What is the impact of
requirements volatility

2028
01:35:52,610 --> 01:35:53,990
on a project?

2029
01:35:53,990 --> 01:35:57,200
So specifically, on the
number of system requirements,

2030
01:35:57,200 --> 01:36:00,490
on rework, on project
schedule, and so forth.

2031
01:36:00,490 --> 01:36:03,410
Really, it's a very,
very recent paper

2032
01:36:03,410 --> 01:36:06,860
on requirements volatility.

2033
01:36:06,860 --> 01:36:09,380
So let me just kick
off assignment 2.

2034
01:36:09,380 --> 01:36:10,670
And then we'll be done.

2035
01:36:10,670 --> 01:36:15,110
So assignment 2 goes out
today and is due in two weeks.

2036
01:36:15,110 --> 01:36:17,270
So that's October 9.

2037
01:36:17,270 --> 01:36:19,640
And it's very focused
on requirements.

2038
01:36:19,640 --> 01:36:23,600
Basically, what I want you-- and
it's the same teams that you're

2039
01:36:23,600 --> 01:36:27,020
in for all the assignments.

2040
01:36:27,020 --> 01:36:29,540
Is essentially, first
of all, first task

2041
01:36:29,540 --> 01:36:32,810
is find a project
or program where

2042
01:36:32,810 --> 01:36:34,730
poorly written or
managed requirements

2043
01:36:34,730 --> 01:36:36,090
were a major problem.

2044
01:36:36,090 --> 01:36:40,670
So as a team, discuss an
example and discuss that example

2045
01:36:40,670 --> 01:36:42,050
as a team.

2046
01:36:42,050 --> 01:36:46,100
The second task is look
at those CanSat 2016

2047
01:36:46,100 --> 01:36:49,580
requirements, those
47 base requirements

2048
01:36:49,580 --> 01:36:51,740
and analyze those critically.

2049
01:36:51,740 --> 01:36:55,730
So that means are they
feasible, are they well-written.

2050
01:36:55,730 --> 01:36:56,990
Classify them.

2051
01:36:56,990 --> 01:36:58,670
What type of requirement is it?

2052
01:36:58,670 --> 01:37:00,230
Is it an interface requirement?

2053
01:37:00,230 --> 01:37:02,150
Is it a performance requirement?

2054
01:37:02,150 --> 01:37:05,690
And then figure out whether
there's a hierarchy there.

2055
01:37:05,690 --> 01:37:09,080
And basically, then, from
that, generate your own set

2056
01:37:09,080 --> 01:37:12,300
of requirements for
the CanSat competition.

2057
01:37:12,300 --> 01:37:15,020
So you're going to either
rewrite or organize

2058
01:37:15,020 --> 01:37:17,600
these requirements in
a way that's better

2059
01:37:17,600 --> 01:37:20,180
and that works
for you as a team.

2060
01:37:20,180 --> 01:37:22,520
And then the fourth
requirement is

2061
01:37:22,520 --> 01:37:27,530
to figure out where
you want your margins,

2062
01:37:27,530 --> 01:37:31,700
where do you think you need
to have reserves and margins

2063
01:37:31,700 --> 01:37:33,410
in these requirements.

2064
01:37:33,410 --> 01:37:36,620
So it's very much
looking and reorganizing

2065
01:37:36,620 --> 01:37:43,930
and critically analyzing the
CanSat 2016 requirements set.

2066
01:37:43,930 --> 01:37:46,120
So let me summarize.

2067
01:37:46,120 --> 01:37:48,010
Good requirements
are really essential.

2068
01:37:48,010 --> 01:37:50,950
It's the starting point
for system design,

2069
01:37:50,950 --> 01:37:53,380
system engineering.

2070
01:37:53,380 --> 01:37:55,150
It is a challenging
thing, especially

2071
01:37:55,150 --> 01:37:58,610
the flow down is challenging.

2072
01:37:58,610 --> 01:38:01,090
There are some methods and
commercial tools for doing

2073
01:38:01,090 --> 01:38:02,590
formal requirements management.

2074
01:38:02,590 --> 01:38:04,840
So I mentioned ISO performance.

2075
01:38:04,840 --> 01:38:09,290
And then I mentioned DOORS,
these commercial tools.

2076
01:38:09,290 --> 01:38:13,480
And then the last point here
is just because you passed SRR,

2077
01:38:13,480 --> 01:38:15,850
and you have high level
requirements approved,

2078
01:38:15,850 --> 01:38:18,310
doesn't mean that's
the end of the story.

2079
01:38:18,310 --> 01:38:23,650
The requirements will continue
to be updated, refined.

2080
01:38:23,650 --> 01:38:26,230
But you have to be you
have to be very disciplined

2081
01:38:26,230 --> 01:38:27,370
to keep that in check.

2082
01:38:27,370 --> 01:38:30,410
If your requirements
volatility is too high,

2083
01:38:30,410 --> 01:38:34,420
then bad things can
happen to your project.

2084
01:38:34,420 --> 01:38:36,280
And I really
recommend the reading

2085
01:38:36,280 --> 01:38:40,400
on requirements volatility, some
very, very recent data on that,

2086
01:38:40,400 --> 01:38:41,510
OK?

2087
01:38:41,510 --> 01:38:45,880
So that's the end of
the session for today.

2088
01:38:45,880 --> 01:38:49,520
I'm going to be online on the
WebEx in about 10 minutes.

2089
01:38:49,520 --> 01:38:52,000
If any of you have questions,
or any comments, or you

2090
01:38:52,000 --> 01:38:54,550
want to dive into this further.

2091
01:38:54,550 --> 01:38:57,160
I know this is not a great
time for you at EPFL,

2092
01:38:57,160 --> 01:38:59,540
because it's Friday
night for you guys.

2093
01:38:59,540 --> 01:39:02,210
So I think what
we're going to do,

2094
01:39:02,210 --> 01:39:05,020
we'll figure this out with
[? Leighs ?] and [? Johana. ?]

2095
01:39:05,020 --> 01:39:07,480
We'll see if we want
to do a separate time,

2096
01:39:07,480 --> 01:39:10,780
a different time during the
week for the office hours.

2097
01:39:10,780 --> 01:39:12,460
I'm happy to do this
at a better time,

2098
01:39:12,460 --> 01:39:14,590
if this doesn't
work for you guys,

2099
01:39:14,590 --> 01:39:16,270
because it's happy
hour time for you.

2100
01:39:16,270 --> 01:39:18,580
So anyway.

2101
01:39:18,580 --> 01:39:19,600
OK, so great.

2102
01:39:19,600 --> 01:39:20,440
Have a great week.

2103
01:39:20,440 --> 01:39:22,770
And we'll see you next Friday.