THE conditions are hostile. The air is so thin that anybody exposed to it
would lose consciousness in seconds and yet the winds are smooth and swift. The
sky is a deep velvet blue and the temperature a freezing 鈥� 65 掳C. There is
no moisture and nothing lives here.
This is not Mars but the region 30 kilometres above the Earth鈥檚 surface. Few
have visited and none have stayed for long鈥攋ust a handful of rocket plane
pilots and high-altitude balloonists.
The problem is that nobody understands the aerodynamics of flight at 30
kilometres and, consequently, aircraft cannot yet be designed to fly at this
altitude for sustained periods. Conventional airliners fly at about 10
kilometres while Concorde reaches 15 kilometres. Even the famous American U-2
spyplane only reaches 20 kilometres.
Advertisement
True, a few supersonic and hypersonic aircraft have reached this altitude. In
the 1960s, the American X-15 rocketplane forced its way to the edge of space at
an altitude of 120 kilometres, albeit for only a few minutes. Modified versions
of the Russian MiG-25 fighter plane have reached 30 kilometres for extremely
short periods of time. But extended flight at this altitude has never been
possible.
That is set to change. Researchers at NASA鈥檚 Dryden Flight Research Facility
in California are planning to hoist an uncrewed experimental aircraft to 30
kilometres using a balloon. They then plan to drop it nose first to see how it
flies.
The $5-million project is called Apex and the results will be used to
build research aircraft that can fly reliably for long periods at these
altitudes. These aircraft will be hugely important for monitoring the
atmosphere, remote sensing and surveillance. They may even be suitable for
exploring other planets such as Mars, where the aerodynamic conditions are
similar to those on Earth at 30 kilometres above the surface.
Naturally, though, the first step is to develop an aircraft that can fly in
Earth鈥檚 atmosphere. But because of the strange laws of aerodynamics that Apex
will experience, the NASA team cannot rely on conventional methods of design.
For example, models of Apex cannot be tested in a wind tunnel. Instead, the team
is relying entirely on computer predictions of what will happen. Indeed, they
hope the real flights will help improve their computer simulations. 鈥淭he worst
case scenario is that it鈥檚 a completely different flow regime and we have to go
back to the drawing board,鈥� says Mark Drela, an aerodynamics expert at the
Massachusetts Institute of Technology who designed the aircraft鈥檚 wing. Even if
the computer design is wrong, he insists that the aircraft cannot go out of
control. 鈥淚t鈥檒l just drop faster than we expect.鈥�
The lack of wind-tunnel testing may seem like an oversight. Wind-tunnel tests
are extremely useful because they allow researchers to test small-scale models
so that the design can be modified before expensive full-scale aircraft are
built. But even if Drela had wanted to test his design in a wind tunnel, he
could not. Peculiar as it may sound, the aerodynamics of flight at 30 kilometres
cannot be recreated in this way.
Wind-tunnel tests work because the flow of fluid over a scale model can be
made to mimic the flow on a larger scale by manipulating factors such as the
speed of the flow, its density and viscosity. Together these determine a factor
known as the Reynolds number and it is this that characterises the flow.
Reynolds numbers vary over many orders of magnitude. The flow around a
dirigible can have a Reynolds number in the billions, around a jet aircraft in
the tens of millions, while insects cope with Reynolds numbers in the thousands.
In simple terms, the number tells engineers how 鈥渟ticky鈥� the air is within the
layer flowing across a wing. All that a wind tunnel must do is reproduce over a
scale model the 鈥渟tickiness鈥� that an aircraft experiences in real flight.
Of course, there are limits to how successfully this can be done, especially
because the fluid of choice for most wind tunnels is air at atmospheric
temperature and density. Without changing the density or viscosity of the fluid,
engineers are limited to altering the size of their models and the speed of the
airflow. Because of these limits, most wind tunnels are good at reproducing
airflows in the range of millions to tens of millions.
But the flow over Apex鈥檚 wings will have a Reynolds number of 300 000, about
the same as a hawk鈥檚. To mimic this, a wind tunnel model would need to have a
wing span of just 2.5 centimetres and measuring the flow around such a small
model would be well-nigh impossible. 鈥淵ou just can鈥檛 put instruments on such an
itsy-bitsy little model,鈥� says Drela.
Low Reynolds numbers cause other problems. In these conditions, the viscous
properties of air make it much more difficult for turbulence to form. Since
turbulence plays an extremely important role in reducing the amount of drag that
aircraft experience, without it an aircraft must overcome huge forces to move
through the atmosphere.
鈥淭he flow around Apex is rather like the flow around a golf ball,鈥� says
Drela. 鈥淚n fact, they have similar Reynolds numbers.鈥� Purely spherical golf
balls are notoriously poor fliers. When hit, they do not travel far because the
flow around them is smooth. So to create drag, golf balls are dimpled. 鈥淚t鈥檚
dramatic. Turbulence can reduce drag by a factor of 2,鈥� says Drela.
This is one of the reasons why wind tunnel tests are futile. Small amounts of
turbulence form in the flow as the air passes through fans and veins inside a
tunnel. 鈥淭his kind of small-scale turbulence never occurs in the atmosphere,鈥�
says Drela.
Most of the time it can be ignored. When an airliner鈥檚 wing is tested, for
example, the 鈥渇ree stream鈥� turbulence is insignificant compared with the
turbulence created by the wing. But for low Reynolds numbers, the free-stream
turbulence is large compared with that created by the wing. It can even trigger
more turbulence, an effect known as 鈥渢ripping鈥�, and this can dramatically reduce
the drag. 鈥淏ut you鈥檇 never know whether the drag was being reduced by the wing
design or the free stream turbulence,鈥� says Drela. This problem lies at the
heart of Apex鈥檚 wing design鈥攚hich must be carefully shaped to encourage
the formation of turbulence.
At the same time, the design must tackle another problem caused by the air鈥檚
viscosity. 鈥淭he thick, sluggish boundary layer is prone to separate from the
upper surface of the wing,鈥� says Al Bowers, another aerodynamics expert at NASA.
When this happens, the aircraft suffers the catastrophic loss of lift known as a
stall.
Engineers accept that there is little they can do to prevent the boundary
layer bubbling away from the front of the wing. But there is a neat way out of
the problem: the trick is to taper the aerofoil so that the boundary layer
reattaches several centimetres down the wing. To pull this off, Apex must fly at
about two-thirds the speed of sound, just fast enough to accelerate the air
flowing over the top of the wing beyond the speed of sound.
This creates a shock wave that sends ripples of turbulence through the
boundary layer. This has the double benefit of reattaching the boundary layer
and creating the turbulence necessary to reduce drag. Meanwhile, the rest of the
airflow is subsonic. 鈥淵ou are between a stall where there is not enough lift and
the region where you get shock waves,鈥� says Jeff Bauer, chief engineer on the
project. 鈥淵ou鈥檙e flying up there on a knife edge.鈥�
Drela can only hope that his design will do the trick. The aircraft has the
appearance of an ungainly glider with a wingspan that is nearly twice the length
of its 6.7-metre fuselage. Nevertheless, the design has been fine-tuned using
computer models with low Reynolds number airflows, a technique known as computer
fluid dynamics or CFD. 鈥淐FD modelling is common,鈥� he explains. 鈥淏ut usually the
models are calibrated using wind-tunnel tests,鈥� an option that is not possible
with Apex.
Instead, NASA will use the results from the real flights of Apex to calibrate
the CFD simulations. A section of the aircraft鈥檚 wing will be riddled with holes
connected to sensors that will measure the distribution of pressure over the
wing. At the same time, an array of metal tubes called a 鈥渨ake rake鈥� will
protrude behind the trailing edge of each wing to measure the amount of
turbulence as it flies.
So after the flight, this data will help physicists to reconstruct the
airflow to find out where the boundary layer has separated and the amount of
turbulence that has formed. It will also calibrate the team鈥檚 CFD models,
anchoring the data. Hence the nickname NASA engineers have given to Apex of the
鈥渨ind-tunnel in the sky鈥�.
And if the problems with aerodynamics weren鈥檛 bad enough, building a machine
that can function at 30 kilometres is equally tough. With the temperature at
鈥� 65 掳C, it might seem that the challenge would be to keep Apex warm. In
fact, it is just the opposite. The mechanical actuators that move Apex鈥檚
ailerons and flaps generate heat, but the air is so thin at this altitude that
there aren鈥檛 enough molecules to draw this heat away.
The solution the Apex engineers have come up with is to place small metal
plates near the actuators to conduct heat away and radiate it out into the
atmosphere. The technique has been used before but never at this altitude, says
Bauer. The plates will be installed on the underside of each wing so they are
out of the sunlight. 鈥淲e鈥檇 also hate to disturb the flow on the top surface,鈥� he
adds.
Then there are the aircraft鈥檚 electronic systems, such as the flight-control
computer and the electronics associated with the experiments鈥攚hich also
generate heat. 鈥淲e鈥檙e especially concerned about the balloon ride up,鈥� Bauer
says. Over the first 10 kilometres, the instruments have to be kept warm as the
temperature plummets, then at higher altitudes the problem becomes one of
cooling. Just how this will be solved isn鈥檛 yet clear.
Nevertheless, engineers began constructing the outer shells of two Apex
aircraft in February. For this job, NASA has turned to a small company near Los
Angeles called Advanced Soaring Concepts, which builds everything from gliders
to components for Formula 1 racing cars and amusement park rides.
Settling on a final design has not been easy. The aircraft must be extremely
light鈥攐nly 270 kilograms, about the weight of three men鈥攁nd yet cope
with forces five times greater than gravity. 鈥淲e鈥檙e fighting weight like crazy
to fly at that altitude,鈥� says John Del Frate, who oversees the Apex project. At
the same time, the wings must remain rigid because any twisting will influence
the pressure measurements. A twist of more than half a degree could ruin the
results. 鈥淲e鈥檙e at the edge of what is doable,鈥� adds Del Frate.
Initially, the structure was built from carbon composite. But engineers at
Advanced Soaring Concepts had their doubts when experiments showed that the
carbon composite wings would twist by as much as 4 degrees. The only way to make
them stronger would be to use more material, but that would exceed the weight
limit. 鈥淲e had to inform NASA it wouldn鈥檛 work. Boron composites are the only
materials that can maintain that stiffness at that weight,鈥� says engineer Tor
Jensen, president of Advanced Soaring Concepts.
The boron material is formed by running tungsten filaments through a boron
trichloride gas and then weaving them into a lightweight blanket that is
hardened with resin. The process is expensive鈥擜dvanced Soaring Concepts
buys the material for $1000 per kilo. The entire wing, including the
internal spar, will be made from the material as well as the tail.
Ironically, this new breed of aircraft could leave NASA with a painful choice
about high-altitude flight. The US hopes to build a fleet of supersonic aircraft
known as High Speed Civil Transports, that will travel faster and higher than
Concorde. Some environmentalists fear these planes will pollute the stratosphere
and, in particular, damage the ozone layer.
But because few measurements have been made in the middle stratosphere,
nobody knows what effect aircraft will have on the atmosphere at this height.
杏吧原创s can make isolated measurements using balloons, but these take samples
only in one place.
What scientists want is a way to fly around in the middle stratosphere taking
large numbers of measurements in different places and even bringing samples back
to Earth. Apex and its successors will help to determine what chemistry occurs
in the upper stratosphere and what the risks will be.
Even further in the future, the Apex data could help build a plane to fly
above Mars where the density of the atmosphere is similar to that at 30
kilometres above Earth. An uncrewed Martian plane could be armed with infrared
and optical cameras to survey more territory than a surface rover and take
higher-resolution photos than an orbiting probe.
鈥淵ou could see sand dunes and ripple patterns. You could look for dry lake
beds that might have accommodated life,鈥� says NASA scientist Larry Lemke, who
dreamt up the Mars application. You could even land at the most promising site,
he adds. Once on the ground, a robot arm could grab a soil sample and analyse it
using microscopes or spectrometers.
Lemke and a team of engineers at Ames Research Center in Palo Alto,
California, are trying to convince NASA managers to launch just such a mission
when Mars comes within reach next in 2001. They have pitched the
$200-million project as part of NASA鈥檚 interplanetary Discovery
series.
If it is accepted, the plane could provide a short cut for scientists
studying the history of life on Mars. 鈥淚n one mission you would be able to do
discovery and exploration,鈥� he says.
The first flight tests for Apex are planned for next year. Only then will
Lemke discover how good a Martian flyer it could be.
