like I've been here before all right so
00:12
I'm going to tell you a bit about how I
00:15
end as I mentioned twenty-two of my
00:17
closest friends at Makani power are
00:19
inventing a new kind of wind turbine and
00:21
I'm also going to tell you a bit about
00:22
the process we've gone through to
00:24
develop this device because I think it's
00:26
kind of informative of the process of
00:27
hardware innovation so when you're
00:31
trying to make something new it's good
00:34
to look at what's out there and actually
00:36
there's this place called Altamont Pass
00:37
which is about a 90 minute drive from
00:40
our offices in the Bay Area in
00:42
California and you can go out there you
00:44
can see all these old wind turbines and
00:45
you can see all the different things
00:46
people have tried and they tried just
00:48
tons of different designs so there's
00:50
three bladed down wind turbines there's
00:52
up wind turbines there's these things
00:53
that look kind of like egg beaters I
00:55
call Darius turbines there's this thing
00:56
that you see in the lower right-hand
00:57
corner here
00:59
it has wires that hold the blades in
01:01
place and what looks like little
01:02
pinwheels that keep the the rotors
01:05
pointed into the wind so people tried
01:07
all these different things and some of
01:09
them worked and some of them didn't and
01:10
the people that made the things are
01:12
worse they kept doing that though the
01:13
ones that didn't those people copy the
01:14
ones who are doing things that works
01:15
they probably added more innovations and
01:17
what you get to in the end is all this
01:21
work in this collaboration everyone
01:23
converged on to one design and that's
01:25
the three bladed up wind turbine like I
01:27
have in the picture here and I actually
01:30
think these are great devices and right
01:34
now they're the cheapest form of
01:35
renewable energy so to do better than
01:38
that is quite a it's quite a task and I
01:43
guess I can say on this on this process
01:45
of innovation this is by the way very
01:47
similar to any other complex projects
01:49
we're at an Aeronautical University so
01:51
let's bring up commercial aircraft
01:52
you look at these they're all the same
01:54
people aren't building things that look
01:56
the same because they're lazy they're
01:58
doing it because simulating and modeling
02:01
while an important part of developing
02:02
new things isn't enough the world does
02:05
not we are not so smart that we fully
02:08
understand the world so you know people
02:12
want to make something that works at the
02:13
end of the day you can only do so by
02:15
building on experience so back to wind
02:20
wind power is great
02:22
I said however it has this problem that
02:24
is still just slightly more expensive
02:26
than all the non-renewable sources out
02:28
there so I have this plot here in red
02:31
are all the years where there was not
02:32
this small subsidy for wind and in blue
02:35
or all the years where there was a
02:36
subsidy so the little subsidy subsidy
02:39
wind does great the wrong people don't
02:41
agree in Congress and it just dies and
02:43
actually we had this happen again this
02:45
year so we've gone from an estimated
02:47
eight thousand megawatt installed last
02:51
year to projected in 2013 zero so this
02:58
brings us to the Makani power Makani was
03:01
founded when the smart forward-looking
03:04
and I say somewhat techno philic guys at
03:08
Google said hey we love wind power but
03:10
could you just go ahead and make it half
03:12
the price and twice as reliable so
03:16
that's a bit of a tall order and so we
03:19
we approach the problem basically
03:21
thinking about a wind turbine if you
03:25
look at a wind turbine you make most of
03:27
the power with the outside segment of
03:28
the blade so if you have a 50 meter
03:30
blade the outer 25 meters are making
03:32
most of the power now
03:35
this conversion of aerodynamic energy of
03:39
kinetic energy in the wind
03:41
into electrical energy can only be done
03:44
by carrying that force through the blade
03:46
so the blade has to be really heavy to
03:48
carry that load to the hub and then the
03:50
hub has to be really heavy and the
03:52
gearbox and the nacelle have to be
03:53
really heavy to carry that power into
03:54
the generator and carry the other loads
03:56
into the tower the tower has to be
03:58
really big and heavy to keep the whole
03:59
thing from tipping over and to support
04:02
all this loads and then you have to put
04:04
that whole thing on a really big
04:05
foundation and on top of that let's say
04:09
you want to put these things offshore
04:11
where the wind is more consistent and
04:12
where you don't have to look at them
04:14
there's plenty of space out there as we
04:16
saw in the earlier talks today well
04:18
water is kind of slippery compared to
04:19
dirt so you need a really big foundation
04:22
to to do that and again if you look at
04:26
this turbine the part that's actually
04:28
doing work is just this little bit just
04:31
the tips of the blades and all the rest
04:33
of this is kind of just long for the
04:34
ride so
04:36
what can we do to get rid of the rest of
04:39
the turban and the idea that we happened
04:42
upon is to resolve those forces with the
04:47
tether
04:47
so rather than convey a load we have
04:48
this blade rather than to bend the blade
04:50
to convey a load aerodynamic loads
04:52
through the blade through the nacelle
04:53
through the tower we convey it through a
04:55
tether and then we tie that tether to
04:57
something in the ocean or on the ground
04:59
and it's not the perfect solution there
05:01
still are other things besides this
05:04
segment of blade that's converting
05:05
aerodynamic energy into mechanical
05:07
energy but it's much better than then
05:10
the then what's out there right now so
05:15
the first way that that we tried this
05:18
was actually with soft kites and this is
05:20
the sort of technology that I think
05:21
Silicon Valley types will really like
05:22
it's very iterative we could actually
05:24
get a new prototype in a week for $300
05:27
on a totally new design and it's very
05:33
easy to prototype so we can take these
05:35
kites and hand launch them it's very
05:37
easy to get up and testing and get out
05:39
and experience the real world and see
05:40
the challenges that that really exists
05:42
outside of the computer and it also has
05:45
the advantage of being really soft and
05:46
fluffy kind of visually appealing and
05:47
Makani had a lot of success with this
05:51
and actually managed to fly some 36-hour
05:56
totally autonomous power generating
05:58
flights with soft wings but there's also
06:00
some challenges to this sort of
06:01
technology one is that it's very hard to
06:03
model so one of the things that happens
06:05
is you build your 10 square metre kite
06:08
and get it working great and you want to
06:09
go build something that's three hundred
06:10
square meters and big enough to actually
06:12
address some of this you know real
06:15
energy output and you do that and it
06:17
behaves totally differently and it's
06:18
extremely hard to figure out how will
06:20
behave so once you scale up all bets are
06:22
off and you're kind of just doing the
06:23
same thing you know the same development
06:25
process again it also has this problem
06:28
that when conditions are not good
06:31
maybe the winds little high they're a
06:33
little fragile so here's one that just
06:35
exploded midair not even reeling it in
06:37
or trying to load it up just because the
06:39
winds were too high for the device so
06:42
after that point we realized okay this
06:44
this this approach it's very good to
06:46
experiment with it we don't think it's
06:48
going to work
06:49
and the next thing that we tried was we
06:51
actually tried a kind of the opposite
06:54
approach and we were in a very aerospace
06:55
approach we designed and built a very
06:58
high-performance rigid wing in
07:00
simulation sorry I should say we did all
07:02
the design work in simulation we
07:04
actually built in real you know you can
07:06
only yeah only build in real life and we
07:10
we built this thing with active controls
07:15
it has to have everything working to get
07:18
this extremely high performance out but
07:20
on paper theoretically extremely good
07:22
but what happens with all projects like
07:24
this well inevitably something goes
07:27
wrong and it can be a sensor being off
07:31
in this case it was one integrator that
07:32
was not zeroed in the control system and
07:34
then the things that happen are not so
07:37
great so at this point we took a
07:44
different approach and that was to go
07:49
back to this sort of iterative very
07:51
classic aircraft design approach which
07:53
says why don't you go and you you copy
07:56
what everyone else does because it
07:57
already works the problem is no one out
07:58
there builds airborne wind turbines so
08:00
we can't very well copy that so we had
08:02
to do it ourselves and what we set out
08:04
to do was to iteratively experiment and
08:07
of course you know modeling and
08:09
simulating as well but experiment and
08:11
and build up the knowledge base the
08:12
intuition the design sense to make these
08:15
devices and make them actually work and
08:17
we started a very simple design now this
08:19
this wing it just looks like a big RC
08:22
plane it's about 12 or 15 feet across
08:24
but this is designed as an airborne wind
08:28
turbine and it's actually focused on one
08:29
thing which is very robust power
08:32
generating loops and in fact when we
08:34
designed it we didn't even know how
08:35
we're going to put it in the air and
08:37
this is actually I think exemplifies
08:38
some of the problems you come across
08:40
when you're trying to make a new
08:41
hardware which is that 99% of the
08:43
problems you solve are not the problems
08:45
that you're going to have you know in
08:46
the final product they're the things you
08:47
have to do to learn the challenges so
08:50
what we did was we invented this this
08:51
pneumatic launcher to to launch the wing
08:55
and here you see us trying it out before
08:57
we put the wing on it
09:02
that was a fun day
09:04
and then you know we we solved these
09:07
problems these sort of everyday
09:08
challenges that come across and then we
09:11
use them so here's the wing being
09:13
launched this pneumatic launcher to get
09:14
it up into crosswind flight so that we
09:15
can actually learn about the problem and
09:17
gain the experience necessary to build
09:19
real hardware and the goal with this
09:22
wing was to make it very reliable it was
09:24
to make it extremely robust you want to
09:27
make it so you can remove almost all the
09:28
sensors you can remove almost all the
09:31
servos totally change the wind direction
09:33
and have a fly on just as happily as
09:35
ever and right now this video is this is
09:37
very exciting because this was our first
09:39
crosswind flight with this and it worked
09:41
perfectly and that's not so much because
09:43
we did a great job of making everything
09:46
squeezing every ounce of performance out
09:49
of it we did a great job of covering all
09:51
our bases and making sure that when
09:52
things break or don't work exactly as
09:54
you model that it still behaves the way
09:56
you want it to so this is great we made
10:02
a lot of progress with that model but
10:03
there's one thing it doesn't do which is
10:05
generate power the way a wind turbine
10:08
generates power is there's force on the
10:11
blades there's aerodynamic loads and the
10:12
blades are spinning at some velocity and
10:14
some component of this force is towing
10:16
the blades around
10:17
it's just torquing it around in a circle
10:19
and this is turning the generator at the
10:21
hub well if you remember we we erased
10:24
that part of the turbine so we can't do
10:26
that anymore
10:28
so what we can do is put these little
10:33
turbines on the wing and so the wing
10:35
this wing is still just a segment of a
10:37
blade and still getting pulled around
10:38
this loop by the air and instead of
10:41
turning some generator at a hub we put
10:43
small turbines on the wing because the
10:46
thing is flying fast is prior it's like
10:47
trying to generate power in a hurricane
10:49
you need a very small turbine to get the
10:51
power up so each of these generates is
10:53
generators is generating power and
10:57
shipping it down the tether at a high
10:59
voltage so there's our solution of
11:01
course then we have to go build it so
11:03
this is our prototype and you can see us
11:05
take this prototype and test it so this
11:09
one we still didn't have a way to get it
11:11
in the air that we thought would be
11:12
viable as a prod
11:14
but what we were able to do with this is
11:17
get it up in the air and really learn
11:18
about power generation and learn about
11:19
all the problems that have to be solved
11:21
and all the optimizations you can do to
11:23
make the power output higher so and this
11:28
still isn't the end okay so we solved
11:29
this problem of power generation and the
11:32
next thing is how to launch and land a
11:34
wind turbine you can turn it off and it
11:36
sits there that's a pretty easy way to
11:39
to handle no wind with us we can't do
11:43
that it's it's a flight vehicle so you
11:45
have to bring it back to the ground
11:46
somehow and what we do is hit the wrong
11:49
button there we go at what we do it's
11:55
very complicated control I have like a
11:57
different button for every slide just to
11:58
keep my toes I guess what we do is we
12:04
actually use these turbines and we
12:06
design them to provide static thrust so
12:08
they're a little less efficient as
12:09
turbines we lose about five percent
12:10
efficiency but it gives us the ability
12:12
to actually put power back up the tether
12:13
and then provide thrust so we return the
12:17
wing to hover when we want to land and
12:19
then we just reel it in and land it on a
12:22
perch so here you see I gave you a
12:31
preview before but now you can actually
12:32
see the this is a prototype doing this
12:36
and we actually managed to build this
12:38
prototype and test all these flight
12:39
modes you saw it just transition in the
12:41
crosswind flight and now it's going to
12:43
fly a few loops and when it's done when
12:45
we say okay it's time to land we can
12:47
actually bring it back and stick it back
12:49
and hover and reel it back in so here
12:52
you see it's going to slow down and it
12:55
just returns to hover right there
12:57
now it's hovering again and it can be
12:59
reeled right back in so that's great
13:03
we've done all the different things we
13:04
have to do on separate vehicles but
13:07
we're not done we have to put it all
13:10
together and furthermore there are other
13:12
challenges that should be solved by wind
13:13
turbines these things they don't build
13:15
them all the same because it's hard to
13:16
come up with a device that's going to do
13:18
better when you have perfect conditions
13:20
is a sunny day and you have exactly 12
13:22
meters a second wind they build them all
13:24
the same because of all the strange
13:25
things you have to deal with the big
13:27
storms the
13:28
electrical outages so one good example
13:30
is high winds and the first thing to
13:34
note with this is that okay our system
13:37
works the same as a wind turbine that's
13:38
what I've been emphasizing so let's
13:40
examine how a kite behaves in a gust
13:44
compared to a wind turbine so you have
13:47
some apparent wind that the kite and
13:49
wind turbine are seeing this is a
13:51
combination of the speed that they're
13:52
moving at and the wind coming into the
13:55
field so when a gust hits it's very hard
13:57
for a wind turbine to pitch it's very
14:00
slow the blades are massive the way tons
14:02
and they're on these huge bearings so
14:04
with the kite because we have a tail
14:06
plane because we designed them to be
14:08
passively stable any good aircraft
14:10
designer should know they pitch in to
14:12
the wind they they actually relieve load
14:14
passively and so when a gust hits we
14:17
actually don't have to deal with it as
14:18
much as wind turbines do and the first
14:21
system where we really incorporated this
14:22
is part of our design concept and it
14:24
brought all these other aspects of
14:27
performance together was winged 7 we're
14:30
not very creative of the naming and this
14:32
wing so this is the first wing that can
14:34
do everything together it can it can
14:36
hover launch and land it can take off
14:37
and land it approach it can transition
14:39
into an outer these power generating
14:41
loops the crosswind flight and it can do
14:44
so in a big range of wind conditions and
14:47
here you get to see it do that
14:59
so here you see a take off from the
15:01
perch when the wind is good good enough
15:04
to generate power you launch it from the
15:05
perch it hovers out to full length
15:07
tether it accelerates up into crosswind
15:09
flight and then it starts generating
15:11
power and as it flies the loops you can
15:18
see it's almost perpendicular to the
15:20
wind and this is because this is the
15:22
orientation where we can slow the wind
15:25
down in the most our goal is just to
15:26
slow wind down that's how we we want to
15:28
squeeze all the energy we can out of it
15:29
so after it flies these loops let's say
15:32
the wind gets gets bad or let's say we
15:34
need to bring it down to do maintenance
15:35
whatever the situation is then what we
15:39
can do is we can actually bring it
15:43
around and return to hover again and
15:46
here we go it's going to slow down and
15:48
it just returns to hover right there we
15:50
reel the tether in and then we land on
15:53
the perch
16:03
and there we go so there's a fully
16:07
functional prototype and it seems pretty
16:10
great but I should say things don't just
16:13
happen where everything works and this
16:15
is a great example where also we didn't
16:18
say okay we've solved the problem let's
16:19
just build it it's going to go you can
16:21
see in the top the upper left hand
16:24
corner the screen here this is our first
16:26
version of the wing and on the lower
16:27
left upper left lower right hand corner
16:30
we have a later version and you can see
16:33
it looks a little different the pylons
16:34
are different and actually what we had
16:35
was we we made a mistake
16:38
we designed the problem the pounds in a
16:40
way we didn't anticipate to be a problem
16:41
but they they didn't work they vastly
16:43
decreased performance and one of the
16:45
philosophies we took with this wing was
16:48
to build it kind of like a truck so we
16:49
make it so everything's really easy to
16:50
bolt off bolt on really simple to build
16:53
fast turnaround and this allowed us to
16:55
very quickly build a new design and put
16:59
it on and test it and and find that it
17:02
works and so the wing we have now is
17:06
absolutely phenomenal we're very happy
17:08
with the performance so the next thing
17:10
to do obviously is to scale up so what
17:12
we're doing next is we're starting to
17:13
work on utility scale devices here I
17:16
have in white solid the wing we have now
17:19
and then the m600 the wing we're working
17:21
on now I think we're going to encounter
17:23
a lot of challenges along the way as we
17:25
build this wing I think it would be
17:27
foolish to say that it will be easy and
17:29
it's just a matter of going from the
17:30
experience we have now but our secret
17:33
weapon to that is our team so I have to
17:37
say this is the thing that allows us to
17:39
do all these innovations and make all
17:41
this progress is we have a team of about
17:44
22 people who are absolutely excited to
17:46
work on this project and they're not
17:48
really in it for the Paycheck they're in
17:49
it because they want to see a future
17:51
with these devices powering our grid and
17:54
they're in it for the challenge that
17:56
comes with putting your energy into this
17:59
sort of problem and after lots of
18:02
perseverance seeing results so that's
18:04
that's what I wanted to say thank you
18:06
for having me here and good night
18:15
you
No comments:
Post a Comment