24
engine testing
Lasers are also used in laser induced florescence (LIF)
testing. “We use laser-induced fluorescence to excite ions,
wait for them to relax and watch the light that is
emitted,” says Frieman. A passive detector detects the
luminescence created by the laser exciting the atoms. LIF
can measure how fast the ions are moving, their speed
and distribution. This helps understand the performance
of the rocket in accelerating the ions.
Unfortunately, LIF is unsuitable for very long
duration tests. Spectroscopy can also be used to measure
the light emission from the plasma, “which tells you,”
Freiman says, “the argon’s purity, if it is argon ions or
neutrals or argon ions with two electrons removed”.
Another method for analyzing plasmas that uses
directed energy is microwave interferometry. This sends
a microwave beam through the plasma stream. As the
plasma slows the passage of the microwave, the change
in the microwave, when it is detected on the other side,
will tell you how dense the plasma is. Related to this is
terahertz interferometry. “Those beams penetrate
through certain thruster material so they can shine right
through the thruster,” says Frieman. “It allows you to get
into the inner channel where you see the plasma being
formed, like a real-time x-ray.”
DEEP LEARNING
The terabytes of data that the probes, optical diagnostics
and directed energy measurement methods generate
have to be analyzed. The only practical way of doing this
is machine learning. “One of the areas we’re looking at is
using artificial intelligence, specifically deep learning,”
says Kingston University astronautics senior lecturer,
Peter Shaw.
Deep learning is where an artificial
intelligence (AI) is provided with huge
quantities of data structured in a way so
that the AI can find patterns. That huge
quantity of data could be the images from
optical diagnostics, for example. Shaw
says, “Your inputs could be one of the
pictures of that frame. Just through
training on the video data that you feed it
100 to 200 different videos, the AI will
then create those patterns itself.”
Shaw believes that by training a
neural network in this way an AI can
then make predictions about how plasma
will behave. A recent research project at
Imperial College London in the UK saw
researchers obtain accurate data from a
plasma model which took 13 hours to
compile into a simulation.
The researchers created a
“rudimentary neural network,” and fed it
the plasma data. “We ran the data and we
got some pretty decent matches in 283
milliseconds. It wasn’t a perfect match,
but it was significantly reasonable.” Shaw
points out the dramatic difference
between 13 hours and 283 milliseconds,
just over a quarter of a second. “That
could really speed up the design loops in
modelling, especially plasma modelling.”
JUNE \\ AEROSPACETESTINGINTERNATIONAL.COM
200kW
of power from the VASIMIR
rocket engine
700,000 liters
per second of pumping capacity
4 // A hall effect thruster
is positioned for testing
at the NASA Glenn
Research Center
VASIMIR
The most recent continuous testing of an electric rocket
has been conducted by Houston, Texas-based Ad Astra
Rocket Company (AARC) for its Variable Specific Impulse
Magnetoplasma Rocket (VASIMR) technology. Spun out
of NASA in 2005, VASIMIR was invented by former
NASA astronaut Franklin Chang Díaz, who is AARC’s
president and chief executive officer.
While the Hermes electric rocket has a maximum
power output of 13kW, the VS-200SS VASIMIR prototype,
which has been tested, has 200kW. The VASIMIR
technology is being developed to provide a propulsion
system for a human Mars mission, because it is more
efficient propulsion method than chemical rockets.
AARC’s research vice president Jared Squire says that
his team want to test for longer durations, and believes
the chamber they have can be used for hundreds of
hours. However, the major challenge is the installation of
more cryogenic pumps, large plates inside the chamber
that are cooled down to cryogenic temperatures. These
capture the plume particles to prevent contamination. All
electric propulsion testing vacuum chambers have this
problem. At NASA’s Glenn Research Center, during a
10,000-hour test they will have to shut down the test
every 1,000 hours to clean their cryogenic pumps.
Terahertz x-ray machines, laser induced
luminescence, IR high-speed cameras, neural networks,
standardized physical probes, the testing technology for
the next generation of space propulsion systems is varied
and complex. It reflects the nature of space and the long
duration missions they will endure. \\
4
/AEROSPACETESTINGINTERNATIONAL.COM