34C3 - Drones of Power: Airborne Wind Energy

34C3 preroll music

Christoph Sieg: The idea is now to go from
space back to earth and try to use drones

– so autonomous flying vehicles – for
power generation. So this is the second

part. So the outline here is …
Applause

Christoph: Thank you very much.
Applause

Christoph: So the outline is that first I
will introduce the source here and

motivate why it is a good idea to harvest
high altitude winds and produce energy

from them. The technological part will
come in the second part here. This is

about the technology which is called
airborne wind energy. And in a third part

I want to show how you can build a wind
drone for low cost for yourself and

experiment with this kind of technology.
So let's start with the first part. And

here as a reminder is the conventional
energy supplier wish list, so probably

what your global players in conventional
energy would think about it or tell you:

They would say that is a surely clean-
enough resource and, meaning on timescales

here, it is exploitable of the order of
one human life expectancy, it's

controllable especially economically and
politically, it is depreciable

economically and it leads so to a very
high profit for some players.

Unfortunately there's also the
technological part and here sometimes it's

driven by hope, saying it will be OK. But,
as we know, it might be mostly harmless.

So as we see here for instance there are
catastrophes like Chernobyl. This is after

the catastrophe where you have the
memorial for the people who died. Then you

have scenarios during the catastrophe
here. This is Deepwater Horizon being like

desperately tried to extinguish the fire
by the US Coast Guard and Fire Brigades.

And of course – what I don't have to
mention here, but in times of fake news

it's important to mention – we are before
the catastrophe. So this is here a plot of

the carbon dioxide concentration in the
atmosphere taking from ice. And as you see

here the ice ages give this variations
over 500,000 years, and now we are at this

spot here that points up. And if you
resolve this into the time scale, extent

this time scale from the last thousand to
2,000 years here – so we are here at this

spot at 2000, year 2000 – then you see
that this rise has started at the

industrialization. So it's a clear sign
that we have to do something. And we have

to do it quickly.

Applause

Christoph: So now let's try to propose
something which can be part of the

solution, namely sustainable energies. And
here's a wish list of what probably you

would think it should be: It should be
sustainable, ubiquitous, continous,

accessible and profitable at the very end.
So does such a source exist? And first I

should define what it means. So
sustainable means it should serve present

needs without compromising the future –
and this is clearly not what we are doing

now – so it should be available on
timescales which are like the lifetime of

our central star if possible. It should be
ubiquitous, meaning that it should be

present almost on any location on earth so
that we can without a very complicated

long-range infrastructure have access to
the energy. It should be continuous,

meaning it should be present at almost any
day time and seasons, so that we can plan

of what we produce. And of course it
should be accessible, meaning it can be

tapped by the technology and lead to a
significant contribution to our energy

mix. And profitable should of course also
be. So does it exist? And the answer is

yes and I want to show that this airborne
wind energy can be a big part of it. So

here I have a table of some sustainable
energy sources and the wishlist items are

written here and I put some of the
sustainable sources. So there is fusion,

there is solar energy – terrestrial and
also the spacial energy which was

presented by Anja and by Stefan before –
hydro energy, geothermal energy and

conventional wind energy; where by
conventional wind energy I mean wind

energy up to approximately 100 meter which
is the hub height of wind turbines,

approximately. And as you can see some of
these items here are not fulfilled by all

these different approaches. So for example
the spacial energy is clearly not

ubiquitous, because you have this beam as
we heard which is just like basically

hitting a certain spot on the earth and
there are transferred into energy, so you

have to distribute this energy. Also it is
not yet accessible. On the other hand wind

energy – here conventional wind energy –
is not ubiquitous, because you can only

select certain spots. And it is not
continuous, because you cannot really plan

when the wind is blowing and when it's not
blowing. So let's add to this list what is

called high altitude wind. And high
altitude wind is clearly sustainable,

because it's also wind energy – so it's
like driven as all the other wind energy

as well. And high altitude here means to
go to heights which are above 200 meters

and try to drain energy from these winds.
So let me argue why it is a ubiquitous

source. And for this Philip who is also
here and part of the team – I'm very happy

he has made this very nice plot here which
shows the western part of Europe and it

shows the ratio of wind power which you
can extract at an optimal height which

should be below 1,000 meter – so this is
just an arbitrary at the moment limit – to

say that we can have a system which can
basically get up to thousand meter height

and compare it to the wind energy which is
basically available at hundred meter. And

in this plot you can see at the coastline
there is a line here and this line is the

line where in the interior you have
already a doubling of the wind power. So

meaning at the coast line itself if you go
to higher altitude you have the double

wind power available then at hundred
meter. Even better, directly at the coast

line there is another line which is a
factor of four better. So as soon as you

put your wind turbines on land side, you
will be a factor of, you have access to a

factor of four higher wind power. And
here, in the region slightly south of

Leipzig, there's another line, this is a
factor of eight where you become better in

wind power, in high altitudes wind power.
So, seeing that the coastal regions have

already a factor of four in this ratio
better and the inland between four and

eight. Oh, the wrong sign. Sorry, they
should be reversed of course. So, saying

that here the site of conventional wind-
energy harvesting, which are now very

limited, and where you put for instance
all the wind turbines in the north, they

become much more accessible if you go to
higher heights. Because there you can

basically use all the land sites. So this
is where you have more sites available

when you harvest at optimal height. And
here, as an example about why it is a

continuous source, you see a time
distribution of the wind velocity in

January 2016 in Leipzig. The wind velocity
is here increasing from yellow to red, and

the altitude is displayed here, and this
is the time scale of the month. And what

you can see is, at hundred meter height
you have almost like only in the lower

parts you have winds, whereas, if you go
to higher heights you have the reddish

parts where you have high wind velocities.
So this shows that continuity is already

improved if you go to higher altitude,
especially for land sites. And this is

almost impossible for conventional wind
turbines. You would have to build a mast

higher and much, much bigger structures.
And also, what is displayed here is the

optimal harvesting height. So this is the
height, again below thousand meter, where

it would be optimal to harvest wind at a
certain time, displayed over the whole

month. And if one goes from this plot to
the histograms, so to the time

distribution of the different wind
velocities, you get this picture here. So

this are the spots the histograms of 100,
170, 500, 1000 meter, and of the optimal

height, so if you adjust your height. And
one of the things that you can see is that

the mean is clearly shifted to higher wind
velocities if you increase the height. And

also, if you harvest at optimal altitude
you shift the whole probability

distribution to the right. So and what
increases there is that the fraction of

time below five meter per second, which is
like the the time where the cut in wind

speed for a wind turbine, so you would
like starting produce energy, the

probability to have such winds is
increased from 76% to 87%, which is quite

a lot of increase. So adjusting to varying
optimal harvesting height is not only

almost, but is really impossible for
conventional wind turbines. So one has to

find another technology, which is better
and can give you access to this higher

altitude winds. So this is the plot again
from before. So I have now a little bit

motivated why the source is ubiquitous and
continuous. Now the question is, is it

accessible, and how it is accessible. And
this is the technological part which is

called airborne wind energy. So how do we
access these high altitude winds. So on

for these, let's come back to the design
challenges, which would be necessary to go

to higher height. So high altitude means,
that you just cannot just increase your

tower, and have more torque on your
foundation, and just scale up the system.

So you should avoid proliferation of mass
and proliferation of the tower and

foundation. And also, varying altitude
means you shouldn't have passive,

stabilizing, static structures, but find
something which can vary. So just as an

example here this is the sky walk in the
Grand Canyon, and this is already a quite

scary lever arm which you have. And if, in
comparison, you take your modern wind

turbine, you rotate it by 90 degrees, and
compared it in size to this, you can see

what kind of torque will be, like will act
on the foundation. So this is already a

very big piece of technology you have
here. So we have to do better, and this is

the second part, namely airborne wind
energy, so the technology itself. So the

first slide is probably the most important
of this part because it explains the

idea behind this technology. So you take
autonomous drones, which are the most

flexible connected to the ground via
tethers, and extract wind energy via these

drones. So how does it work? So look at
this conventional wind turbine here. You

have most of the energy is produced by the
outer part of the wings. They are rotating

with the highest velocity, and at the same
time you have the highest the largest

lever arm. So you produce most of the
energy in the outer part. The inner part

is more or less passive, stabilizing
structure. So you remove that structure

and replace it by something which is
flexible, and the first which comes to

mind probably is a tether with which you
attach it to the ground. And then you have

just the active part here, which is now an
aircraft, moving in this circle, which

before was circulated by the wing tips, to
extract your energy. This is the

principle. So how do we bring down the
power when circulating this aircraft? So

we have to, in some way, transform it to
electric power. So there are, which are

not shown in the picture before, lighter
than air systems. So you just basically

take a balloon, you put your wind turbine
at high altitude, and extract the power.

And here the tether can clearly serve as
the power line. But what we can also do is

crosswind flight, which was shown in the
picture before. So here you have a moving

aircraft, which can move in something
which is called the drag mode, meaning

that you have onboard generators on the
aircraft. So essentially it's a propeller

aircraft, but the propellers are reversed
in repeller mode, so that the repellers

produce energy for you. And then the
tether serves as power line. So this

principle is shown here. So here you can
see the generators and then the power is

brought down by the tether. In the second
part, second strategy, is using the so

called lift mode So here you have ground
based generators and the tether itself

transmits the power, there are no power
lines in the tether. So here you use that

the power is given by the pulling force
times the reel out velocity of the tether.

So you circulate in some patterns with
your aircraft and you use the lift force

acting on the aircraft to unreal this
tether from a drum, and at the drum, on

the base station, there's a generator
attached which helps you to get the

energy, to transform the energy into
electrical energy. And of course, at some

point the tether is maximally reeled out
and then you have to have to go to a reel

in phase, where with minimal energy you
reel in the tether again, and start

periodically this phase again. So these
are the concepts, and there's a whole zoo

of airborne wind energy devices and
proposals, which show that this technology

is still in a very early stage of being
developed. So you have people here flying

figure-of-eight patterns with the
aircraft. So some things are lighter than

air turbines, which look very exotic like
this one, probably this one you have seen

in media already. Proposals like this
here. There are quad copters, which

produce the energy by rotating of their,
of the propellers here. And all kind of

exotic lever arm and aircrafts which you
can use. So let's bring a little bit of

more order into the technology, into the
proposals. And one of the things I want to

discuss, which is very promising, is what
is called crosswind flight. So here as a

example is a comparison of a conventional,
lighter-than-air system with the big wheel

in London. So this is one of the biggest
wind turbines. And the harvesting area is,

so the effective area of such a wind
turbine is the swept area of your

propellers, essentially. So now let's look
what happens if you move an aircraft

instead through the wind. Then the picture
of before is like of that size. And if you

take an aircraft, which has the same wing
area as the wing areas of the propeller

here, you're harvesting area is of that
size. It's much bigger. And the reason for

this is, that the effective area is now
given by the wing area times a

coefficient, which is the square fraction
of the lift to drag coefficient of the

aircraft times the lift coefficient
itself. And this factors of the order of

200. So it increases the efficiency of
your of your wings dramatically. This was

already found by Loyd in 1980. And you can
now ask "Why does it take 30 years from

this idea to first systems?". And the
answers is, in this community for is

probably a very interesting is "Why are
these prototypes are appearing only 30

years later?". It's because sufficient
computer power. So for the control

algorithms, which allow you to control
such flight modes, was not available. So,

as an example, here's an illustration of
one of the current leaders in the field

called AMPYX POWER, showing a crosswind
airborne wind energy system versus a

conventional system. So here's the
conventional wind turbine for two

megawatts. And the conventional this is a
conventional system. And the airborne wind

system is, this is the ground station, and
this is the aircraft. So one of the things

which are, I mean, visible in this picture
is that it has much less like even sight

impact in the environment. So having
something like this is much less

disturbing from the even from the
aesthetic point of view, than this huge

wind turbine. So now the next step would
be to look closer to the technology and

see what are the AWE system components
that you need, that you need to build such

a device. So first of all, there is the
drone or the fixed-wing aircraft. We have

seen that it's very good to have large
lift and small drag coefficients, so you

need something which is like a rigid
glider, more or less. On board you need

sensors, like accelerometer, gyroscope,
GPS, receiver, barometer, and a pitot tube

to measure the air the air speed. And this
is to determine the system state, that

then is like reacted on by the control
surfaces, in the case of an aircraft by

ailerons flaps and the rudder. Moreover,
you need of course a microcontroller and

algorithms which do the state estimation.
So from the sensor data they compute the

state of the system, meaning it's
position, altitude, velocity. And you have

to navigate. So and of course you might
need something like a propeller for

takeoff, landing, and energy generation in
case of drag mode. The second thing is of

course the ground station. So here you
need the drum for tether wind-up. You need

a motor which eventually has to be
transformed into generator mode if you

have the lift mode. You need power
converters, also microcontrollers and

algorithms which synchronize your ground-
station operation with the drone; and you

need a runway, catapult or something alike
for takeoff and landing. So far it looks

quite simple, but the devil is in the
detail. And here I found a nice quote a

colleague of mine – (uninteligble name) –
has done in one of his talks,

and I liked it very much because it
displays very well what challenges have to

be still overcome. So it starts with
"Theory is when nothing works but everyone

knows why." and to demonstrate this let's
have a look at this video here which is

one of the flight attempts of one of the
companies: So the aircraft lifts off,

there's no sound … yet. Now there is
sound.

Background music of shown video

Speaker in shown video:
Abort! Abort! Abort!

Soft laughter

Christoph: Yeah. And the desperation of
the founder was clearly hearable at the

end. And you could see that the tether
ruptured. And then there was no way to

recover that most of the aircraft was
lost. Second: "Sometimes practice is when

everything works but no one knows why." So
there are also positive surprises. And

here is a launch, a catapult launch, for
an aircraft which now uses weight.

Background noise of shown video
Laughter

Christoph: So a positive surprise for a
test. And finally, sometimes if you

combine theory and practice then "nothing
works but no one knows why". This is where

the complication really is: The devil is
in the detail. And here you can see a

video from a flight which is crosswind
flight: Everything seems normal …

Dramatic background music of shown video
Christoph: … and then the prototype is

again lost. So this is complicated. So but
there is a lot of progress and so I want

to come closely, very quickly introduce
the current industrial status. So I focus

on three companies which work on that: So
one of them is Enerkite in Berlin, and

they have now a system which is basically
stationed on such a truck and this is a

crosswind system of a passive wing. So it
steered via three tethers and it produces

up to 30 kilowatts of energy. Then you
have Ampyx Power. They have here the

launching site in the Netherlands and they
are currently producing this aircraft

here. This type which is a crosswind
system in lift mode. And at the end will

produce up to 250 kilowatts of power. This
is under construction. And, finally, there

is Google X Makani in California. And they
have built a drag-mode aircraft – here –

which is flying. And I can show you a
video that they have on their home page –

very nicely. Where they show a flight so
that you can see that the 600 kilowatt

system is working. Here you see the
onboard propellers. You can see the

tether. Down here this is from the tether
attachment point. So the things are

working. There are prototypes. But one of
the things which are important is: one has

to "test, test, test" and get experience
from tests. So "experience is what you get

when you were expecting something else".
You really. So what does it mean? So we

have to test, analyze, adapt the systems.
So because many – as you could see from

this design variations in the zoo which
I've shown – many of the concepts are

still open. So for example the design of
the airframe. If you use a biplane, a

flying wing or anything alike – or
something totally different – is still

open. The tether construction – what kind
of materials to use – is still open. The

materials in itself is still open for the
aircraft etc. etc.. The mode of operation

– that means takeoff, landing and direct
versus lift mode – is still an open

question. What is the best thing to
realize for industrial products? And then

control hardware and software algorithms
have to be tested thoroughly. Of course

it'd have to be certified by the aerospace
agencies, of course. You want to have a

failsafe. So what you have to do is you
want to even, I mean, have total losses in

experiment. You want to do the experiments
wich would lead to a total loss of your

system. So here comes the idea that
instead you should build a cheap and

disposable test platform instead of a
largely scaled-up system first, before you

build the expensive prototype and do tests
on them. And this brought us to the idea

to provide a low-cost open-source test
platform where everybody at home can build

his own wind drone. And this is the third
part of the talk. So the do-it-yourself

wind drone. What are the ingredients here?
So first you need a drone, so here I want

to show the airframe and reinforcement
hack which is necessary to prepare your

airframe for the additional forces by
adding the tether. Then there's a ground

station and here I want to motivate why
the drone is essentially behaving like a

fish – in this case a barracuda. The next
thing is navigation on curved manifold is

very important because you have like a
constraint coming from the tether. And

finally you need something for control
which is the autopilot. So in this case

it's the ardupilot open-source project
which we adapted. So let's come to the

airframe-reinforcement-hack. So what you
use: Take your favourite polystyrene

airframe – so in this case it's an Easy
Star II – and glue the wings together.

This is the lower side of the wings. You
put in there a carbon rod – here in this

part – and you stabilize it with racks
which you glue into the slits we can see

here. And then you wrap carbon in the
forward part of it where the most of the

aerodynamic force is attached. Then you
have the carbon ???? ???? wind around your

tether. And you install additional tubes
for fixing the wings on the fuselage. So

the fuselage is here. We cut off the
engine blocks, included additional carbon

rods. So you can put these carbon rods on
these carbon rods here, and fix everything

with screws. So to show you how that looks
like and what are the size of this model

is: So here is the original-size aircraft
with carbon. And you can later – if you

want – pass by the assembly area and look
at it and have a look at it and touch it.

So this is how it looks decomposed into
different components: So again wings and

so on and so on, the servos for the
control surfaces. The central unit here is

the Pixhawk autopilot. So there's a
microcontroller which contains some of the

sensors: You have a GPS sensor, in
addition you have a telemetry antenna for

data – for data transfer to the ground
station. And you have RC control for

manual control when you switch out of auto
mode to have manual control in emergency

situations – or if you want to make other
kind of flight tests. So now this is the

drone itself. So now is the question what
to do with the ground station. And here

let's look why the drone behaves as a
fish: Because what it does is, like in

fishing, you would need a free-moving
tether; it has to be fast and fail-safe

reeled in and reeled out; and it should
remain twist free so that it doesn't give

any knots if it is not under tension. And
the thing which we came up with wich best

serves for our needs at the moment is an
off-shore fishing reel.

Applause
Christoph: So and you need offshore here

because the drum has to be perpendicular
to the rod: This guarantees you like

reload phases twist free on the
tether. Other fishing rods have the drum

aligned with the rod, and then you
accumulate twist on the tether which can

lead to knots, lead to knots and then …
it's not a good idea. It will destroy your

tether. So and this is the first flight
test. So we were very enthusiastic and

started the first flight test. And here it
is.

Indistinct voice in shown video
Voice in shown video: OK. … Hinterher?

Laughter
Visv: Versuch mal rauszugehen. Manual?

Achtung, Achtung! Versuch ihn zu fangen.
Na gut.

Beeping in shown video
Christoph: OK.

Laughter
Christoph: So unfortunately it did not

work.
Applause

Christoph: So what happens? This was the
result: The tail was broken. And because

the tether apparently wrapped around the
back of the aircraft and then it became

uncontrollable. So we came up with what do
we do: If you don't know any further, any

better, use carbon! So we put some carbon
on the lower part of the of the fuselage

to reinforce it. And then of course you
have to think about writing your

navigation code, to navigate if you are
under tethered flight. So here is the

receipe for how to do it: So first you
take one git clone of ardupilot – this

autumn open-source software. You
take one curved 2-dimensional manifold –

it's essentially giving us a hypersurface
embedded in 3-dimensional Euclidean space.

In case of constant tether length this is
just a semi-hemisphere as to which is

centered around your ground station. Then
you take a planar curve which you want to

fly along – or curved segments – and a
pinch of Differential Geometry to wrap it

on the sphere, to make this curve
appearing on the sphere. You take a little

bit of Classical Mechanics for the flight
control to transfer the curve

accelerations into actually control-
surface motions. And then you need, of

course, 12 dozen coffee for doing so. You
put everything together into – of course

not the coffee – into the computer algebra
system and stir well, and let the CPU bake

it at 100 degrees, and then you come up
with a smooth – at least C¹ – curve.

Applause
Christoph: So the curve is shown here. So

it's … this is one part of a figure-8
pattern. So the other part would be behind

here. It's composed of two geodesic
segments and one turning segment and they

are C¹ glued together here. And these are
the equations: So you can find in the

paper – I don't want to go into detail. So
now you have to modify the source code of

this ardupilot project. So here there are
highlighted the patterns which you

basically have to … where you have to do
modifications: You have to implement new

flight modes and change some of the
control algorithms. And then you come up

with the next flight test. And here is the
next attempt.

Music and propellor sounds
Voice in shown video: Beim Auswerten

müssen wir sehen, ob wir dann verschiedene
wählen.

Music ends
Applause

Christoph: The whistling sound you have
heard at the end is the tether being

dragged through the air. So there was
really tension on the tether. And you can

also see this if you do a data analysis on
the flight data later. So yes for example

multiple possibilities. You have a lot of
data which is possible to analyze. So the

autopilot this was very very it's very
very nicely done in this open-source

project: So they have a data file with all
primary and secondary data you can use for

your analysis. So for instance this is the
flight curve of different flight modes

which we used. You have the altitude of
the aircraft, you can look to deviations

in radial and transverse directions. You
can look to tether tension – or like a

measure for tether tension – by looking to
the length variation of the tether. And

you can of course do time series analysis
of how your figure-8 pattern has flown

along. And that is what you can do with
this very very nice autopilot open-source

software which is available when … written
by many many people on the internet. So

the question which remains is: After all
of this is, will it be a fail-safe to

100%? And the answer is nope, it will not!
It will … there will be of course

accidents happen. But the thing is:
Nothing is failsafe. And so here's a

standard wind turbine and look for
yourself.

Laughter
Christoph: You see there is no 100%

guarantee, but we have to try very hard to
get it as failsave as possible. So yeah

this is essentially it. That was the talk.
So what I want to say is that the current

status of airborne wind energy can be seen
here by a nice book on the Springer page

which you can download here. And we are
very very happy to have any kind of

critical remarks, input to help in
developing the system further. So please

if you want, look to this web page,
there's a lot of information including a

paper and we will be very happy for any
kind of help. And finally I would again

stress that we could rely on this
tremendous work of the open-source

community working on this autopilot
project that has helped us to realize this

project in very short time; so very happy
about this. And I want to thank of course

Phillip Bechtle, who is here, and Thomas
Gehrmann and Maximillian Schulz-Herberg,

the students, and Udo Zillmann, who can
not be here, for working on this project

and putting so much work also into it.
Thank you very much for your attention!

Applause

H: We can have two more on the microphones

here and here – one and five – so two
questions. The first one, please!

Question: So you talk, so you talked a lot
about powered – and not powered –, but

controlled flight. How does it compare –
energy wise – to uncontrolled flight?

Basically putting a propellor on a kite?
Answer: So the thing is the propellor on

the kite … with kite you mean, I guess,
non-rigid structures. So meaning that the

first question is how do you want to put a
propeller on a kite if it's non rigid. So

that is a question which goes back to you.
So because that is something is not clear

to me. But in any case rigid air
frame is harder to control than a

kite. So there are people who work with a
kite. And by kite surfing or if you do

like steer normal kites from the ground.
You know it's like moving not that fast in

the wind field, so it's easier to control.
This is a big benefit of kites. And also

the weight is a big benefit. But the power
output – because of the bad or worse lift-

to-drag coefficient – is unfortunately not
that efficient as a rigid aircraft. So you

want to go to the rigid air craft.
H: If you leave the room now, please be

quiet because we have questions and
answers here! Number three please, and

that is the last question I'm afraid. But
you can ask questions after the talk.

Q: I want to go back to the space part. I
was wondering … there are some ideas about

bootstrapping like a solar station on the
moon and then like shipping, I don't know,

hydrogen or like pre-charged lithium
batteries back to earth and back and

forth. Is it like realistic or not really?
A by Anja Kohfeldt (previous talk): I

think also this approach would be quite
expensive. And you have to install this

infrastructure on the moon first, and you
have to establish the flight base back and

forward. Realistic is a thing, you know.
At the end that's a question of money and

investment. And I'm not sure whether this
would pay out, but we haven't analyzed

this kind of approaches, yet.
H: Thank you! So thank you very very much

Stefan, Anja and Christoph! Give them a
warm applause again please!

Applause
Stefan: Thank you!

Outro music

subtitles created by c3subtitles.de
in the year 2019. Join, and help us!