Polar Analysis from flight logs?

Robert Ehrlich wrote:
"Lars P. Hansen" wrote:

OK, here we go: Could devices like this not also be used to detect
thermals?
The description in the link below about how the laser "sees" minute
dust
particles in the air seems to be well suited to thermals.


I don't buy the explanation they give in the cited url
(http://www.navysbir.brtrc.com/succes...navsea_p3.html)
They pretend the device measures the speed and direction of dust
particles
from the shift in the frequency of reflected light, this is well
known
as Doppler effect and can only give the radial component (toward or
away
from the sensor) of the speed, not its value and direction. For
thermals
we are interested in the speed component which is nearly
perpendicular
to the measured component, so this would be of little interest. Of
course
whith several such devices on the ground, all the 3 compenents of
airmass
speed could be measured, maybe this in the intended use of the device
as it
is advertised, but in a glider you don't have sufficient vertical
distance
for putting 2 devices which could provide an accurate value for the
vertical
component of the speed.

It does work, but they use a little different technique.
The Doppler is only measured on particles at the focal length of the
optics.
The assumption is that the airmass (at least locally) is all the same,
and that the Doppler measurement is taken far enough away so the
effects of the airplane on the airmass are negligible.
So you send out two beams -- say, one forward at 45 degrees, one aft at
45 degrees.
It turns out that if you sum the signals from the two beams, you get
the vertical component of velocity, and if you difference the two
signals, you get the horizontal component.
Since we're measuring frequency, we can get sum and difference
frequencies from a mixer, though I have no doubt it runs through a DSP
somewhere.

So you only need one sensor head (though it puts out multiple beams).

By sending out two more beams, to each side, you can also pick up
sideslip information.
The clever thing is that they're using components developed for the
communications field, which helps to keep costs down.

Tim Ward

Robert Ehrlich wrote:

After this discussion we are far from the original question, i.e.
can we infer anything about the polar of a glider just from GPS
fligth logs of this glider?

The obvious answer several people gave was: no because the airmass
movement is unknown.

Well, as a former mathematician, I would say this is just what the
name says: an unknown. Would we be able to determine it?

The problem is this is not a single unknown, it is an infinity of them.
And despite the fact that each track log point gives 3 equations, this
would not be sufficient for determining an infinite number of unknown.
Even when considering the finite number of unknowns consisting of the
3 components of the airmass speed at each point of the log, we have
more unknowns than equation since we have also the unknown polar we
want to determine.

So is there a solution?

We can do for other unknowns just what we do with the polar data,
which are also an infinity of unknowns: reduce their number by
assuming a simple model depending of a small number of unknowns, this
is usually done for the polar by assuming a quadratic approximation
depending only on 3 parameters.

In the same way we can assume that the horizontal components of
the wind are constant on the flight area at a given altitude, and
that the evolution with altitude could be carcterized with a few
parameter, e.g. wind speed at 3 given altitudes and polynomial
interpolation between them.

For the vertical component of airmass movement, we can assume that
the pilot is following some speed-to-fly rule caracterized by
a MC setting and the 3 parameters of the glider polar that was
used for making the MC ring or programming the flight computer.
Even these values may be considered as unknowns.

So we have now a small number of unknowns and a comparatively
large number of equations from the flight log, plus the few
ones from the physics relating position to speed, airspeed and
wind to ground speed, sink to height, sink to airspeed according
to the speed to fly rule and so on. Our system is no more undetermined
but overdetermined, probably there is no exact solution, but there
are methods for determining a most likely solution, i.e. values
for the unknowns that minimize the way the equations are not
satisfied (least square method or others).

As I am only a *former* mathematician, I will not go further
in this way, maybe somebody who is a real mathematician will
complete the job, or somebody who is teaching maths will
propose that as a project for students, this is the lazy method
I would have used when I was myself teaching :-)

Yeah- but would you believe the answer when you got it?

Not to mention another unknown- airflow separation during pull-ups and
pushovers, and effect of varying g-forces.

Why is it not possible to refine the process that is being used by
Dick Johnson by using the gadgets that are available.
That would refine the data acquisition and would make the
resolution much finer. The work load is very high during the process.
Dick invited me to a test flight from 12000ft and not much useable
data was obtained from my only attempt. It sure takes practice.

One certainly has to chew gum, use a stop watch, make notes of time,
speed and temperature and flap settings, maintain speeds, change
speed and stabilize it.

This every 500ft, At the same time Fly a big circle around the airport,
so you have a place to land when finished. I may have left things out.

What electronic aids could be used that would help in this matter?
Udo






"Richard Brisbourne" wrote in message
...
Robert Ehrlich wrote:

After this discussion we are far from the original question, i.e.
can we infer anything about the polar of a glider just from GPS
fligth logs of this glider?

The obvious answer several people gave was: no because the airmass
movement is unknown.

Well, as a former mathematician, I would say this is just what the
name says: an unknown. Would we be able to determine it?

The problem is this is not a single unknown, it is an infinity of them.
And despite the fact that each track log point gives 3 equations, this
would not be sufficient for determining an infinite number of unknown.
Even when considering the finite number of unknowns consisting of the
3 components of the airmass speed at each point of the log, we have
more unknowns than equation since we have also the unknown polar we
want to determine.

So is there a solution?

We can do for other unknowns just what we do with the polar data,
which are also an infinity of unknowns: reduce their number by
assuming a simple model depending of a small number of unknowns, this
is usually done for the polar by assuming a quadratic approximation
depending only on 3 parameters.

In the same way we can assume that the horizontal components of
the wind are constant on the flight area at a given altitude, and
that the evolution with altitude could be carcterized with a few
parameter, e.g. wind speed at 3 given altitudes and polynomial
interpolation between them.

For the vertical component of airmass movement, we can assume that
the pilot is following some speed-to-fly rule caracterized by
a MC setting and the 3 parameters of the glider polar that was
used for making the MC ring or programming the flight computer.
Even these values may be considered as unknowns.

So we have now a small number of unknowns and a comparatively
large number of equations from the flight log, plus the few
ones from the physics relating position to speed, airspeed and
wind to ground speed, sink to height, sink to airspeed according
to the speed to fly rule and so on. Our system is no more undetermined
but overdetermined, probably there is no exact solution, but there
are methods for determining a most likely solution, i.e. values
for the unknowns that minimize the way the equations are not
satisfied (least square method or others).

As I am only a *former* mathematician, I will not go further
in this way, maybe somebody who is a real mathematician will
complete the job, or somebody who is teaching maths will
propose that as a project for students, this is the lazy method
I would have used when I was myself teaching :-)

Yeah- but would you believe the answer when you got it?

Not to mention another unknown- airflow separation during pull-ups and
pushovers, and effect of varying g-forces.

"Udo Rumpf" wrote in message
...
Why is it not possible to refine the process that is being used by
Dick Johnson by using the gadgets that are available.
That would refine the data acquisition and would make the
resolution much finer. The work load is very high during the process.
Dick invited me to a test flight from 12000ft and not much useable
data was obtained from my only attempt. It sure takes practice.

One certainly has to chew gum, use a stop watch, make notes of time,
speed and temperature and flap settings, maintain speeds, change
speed and stabilize it.

This every 500ft, At the same time Fly a big circle around the airport,
so you have a place to land when finished. I may have left things out.

What electronic aids could be used that would help in this matter?
Udo


A data logger (could be a PDA) attached to a differential carrier phase GPS
It should also be able to record accurate airspeed indicator and
temperature, or a true-airspeed reading device like the laser anemometer .
The GPS data can then be post-processed to give altitude information down to
10-15 cm or so.
The airspeed data can be be used post-flight to see just how steady-state
the data collection was.
With increased accuracy in the vertical measurement, it might be possible to
get sufficiently accurate data in say, 300 feet of altitude, allowing more
data points on a single flight.

That will make the flight card a little busier, perhaps, but fewer tows
should make the testing cheaper.

With samples every second, or even every tenth of a second, the additional
data should make it possible to improve the repeatability.

The hardware isn't cheap, but it's getting cheaper, and the data reduction
post-flight isn't a trivial problem, either, but it's doable.

Tim Ward

Udo Rumpf wrote:
Why is it not possible to refine the process that is being used by
Dick Johnson by using the gadgets that are available.
That would refine the data acquisition and would make the
resolution much finer. The work load is very high during the process.
Dick invited me to a test flight from 12000ft and not much useable
data was obtained from my only attempt. It sure takes practice.

One certainly has to chew gum, use a stop watch, make notes of time,
speed and temperature and flap settings, maintain speeds, change
speed and stabilize it.

This every 500ft, At the same time Fly a big circle around the airport,
so you have a place to land when finished. I may have left things out.

What electronic aids could be used that would help in this matter?

In the '80s, Rudy Allemann and I did some Johnson style flight tests but
recording the time, ASI, altitude, and temperature with a camera aimed
at the panel. It made the data acquisition easier, and the pilot could
spend more time on flying, and less taking notes.

Though this probably isn't the "electronic aid" Udo had in mind, using a
digital camera would make it even easier today. It could be aimed to
include the flap handle position, and it has it's own clock to
time-stamp the pictures. Many more pictures could be taken for each run
instead of being limited to the 36 a film camera has.

While this makes the data recording easy, the ASI (including the pitot
and static if you want to compare between gliders) and altimeter still
need to be carefully calibrated.


--
Change "netto" to "net" to email me directly

Eric Greenwell
Washington State
USA

Tim Ward wrote

The hardware isn't cheap, but it's getting cheaper, and the data reduction
post-flight isn't a trivial problem, either, but it's doable.


Tim,
My knowledge is very limited when it comes to the points you made above.
My L Nav produces the data that is now taken manualy in the test flights.
The GN II is hooked up to my LNav making some of the data available.
All this is set up to be down loaded into my PC flight software.
How much effort would it take for some one experienced in writing this type
of software to get
the remaining data out of the LNav into the flight recorder, like
temperature, altitude, airspeed?
Regards
Udo

Eric wrote
Though this probably isn't the "electronic aid" Udo had in mind, using a
digital camera would make it even easier today. It could be aimed to
include the flap handle position, and it has it's own clock to time-stamp
the pictures. Many more pictures could be taken for each run instead of
being limited to the 36 a film camera has.

While this makes the data recording easy, the ASI (including the pitot and
static if you want to compare between gliders) and altimeter still need to
be carefully calibrated.

This is a good idea as the data is displayed digitally. I can not remember
if speed, altitude
and temperature can be displayed simultaneously on the L Nav.
I wonder if a small digital movie camera would have a good a enough
resolution to do the job.
Regards
Udo

"Udo Rumpf" wrote in message
...
Tim Ward wrote

The hardware isn't cheap, but it's getting cheaper, and the data
reduction
post-flight isn't a trivial problem, either, but it's doable.


Tim,
My knowledge is very limited when it comes to the points you made above.
My L Nav produces the data that is now taken manualy in the test flights.
The GN II is hooked up to my LNav making some of the data available.
All this is set up to be down loaded into my PC flight software.
How much effort would it take for some one experienced in writing this
type
of software to get
the remaining data out of the LNav into the flight recorder, like
temperature, altitude, airspeed?
Regards
Udo

As I understand it (and mind you, I was only there once,blindfolded, and it
was a dark and stormy night), Johnson's method relies on the repeatable
vertical accuracy of a mechanical barometric altimeter and a stopwatch.
That gives him an average sink rate over the period of time. If he's
recording temperature, then maybe he post-processes that to get the true
sink rate and true airspeed.
I don't know what kind of vertical accuracy he gets out of that combination
of panel shaker and altimeter. I know I once had an occasion where the
altimeter indication in a 2-33 dropped about 150 feet when I tapped it with
a finger.
If the vertical accuracy on the LNAV is as accurate as he can get, or
better, then it's probably worthwhile to try to pull the other data. I
suspect that it's not, if it's just consumer-grade GPS altitude information,
or Johnson would be using it.

I haven't looked at the LNAV programming interface, so I don't know how hard
it would be to get the information logged. The developers of Soaring Pilot
seem to be pretty amiable about adding features to the program, though.

Tim Ward

I think that a Cambridge 302 DDV has enough data in
its proprietary GPS sentence to be able to give you
the data you need. This is a cut and paste from the
manual:
----------------------
!w
The !w sentence is a proprietary format sentence that
contains air data and
instrument settings
The format is:
!W,1,2,3,4,5,6,7,8,9,10,11, 12,13*hhCRLF
1 Vector wind direction in degrees
2 Vector wind speed in 10ths of meters per second
3 Vector wind age in seconds
4 Component wind in 10ths of Meters per second + 500
(500 = 0, 495 = 0.5 m/s
tailwind)
5 True altitude in Meters + 1000
6 Instrument QNH setting
7 True airspeed in 100ths of Meters per second
8 Variometer reading in 10ths of knots + 200
9 Averager reading in 10ths of knots + 200
10 Relative variometer reading in 10ths of knots +
200
11 Instrument MacCready setting in 10ths of knots
12 Instrument Ballast setting
-------------------------------

Enough data here ?

John

John Ferguson wrote:
I think that a Cambridge 302 DDV has enough data in
its proprietary GPS sentence to be able to give you
the data you need. This is a cut and paste from the
manual:
-snip-

I recall from a conversation with Dave Ellis a few years ago, that the
302 can dump a lot of flight parameters at fairly high data rates over
the serial interface.

I don't know the details, but I beleive all one needs to do is connect
a terminal program, send the appropriate query, and the 302 starts
spilling its guts. A quick call to CAI might provide the answer.

I've often thought it would be cool if one could make a request of the
302 to produce a "super" IGC log with perhaps several records per
second containing airspeed, temperature and calculated windspeed. The
IGC spec allows for such data as comments or extra fields. It
certainly has the horsepower, and the results might provide some
interesting atmospheric information.

-Tom