At 13:53 98-07-25 +0100, Mr Phil Ede wrote:
>Hi.
Bonjour!
>I use NEC to generate the radiation pattern of transmit and receive
>antennas for use with ICEPAC,
I know an ICEPAC thermal modeling package... I'm sure there's no
relationship.
>which I now need to use to predict the
>coverage performance expected by an air-traffic control ground
>station. Can anyone please advise me of the approach I should use to
>characterise the antenna performance expected on a typical large
>airliner?
The performance will not extremely be dependant on the exact shape of
the aircraft elements, especially at the lowest frequencies. But the
general dimensions of the fuselage, the tail, and of the wings,
relative to the wavelength will.
Since you're only interested in general behaviour, may I suggest
creating a simplified wire grid generic model, say of a bottom wing
design such as a 737, which would be frequency swept.
Suppose you're trying to model a 190 ton Iliouchine IL86 aircraft. (I
take the data I have in hand :-) I think this is a rough equivalent to
a DC10.
>From the drawing, I estimate a 7m fuselage diameter, a 60m body
length, a tail height of 6m, and a wing span of about 45m.
Taking the measurement frequency to be 20MHz (lambda=15m), and using
12 segments to the wavelength, or 2m/segment [2], we have by a crude
approximation:
Surface of fuselage ~= pi*7*60/2 = 659m^2.
Surface of wings ~= 45*3 = 135m^2
Surface of tail ~= 7*3 = 21m^2
Or a total wire grid area of about 815m^2.
For a square of one wavelength (15*15=225m^2), we need 2*12*12=288
segments (the "2" is for horizontal and vertical grid segments)
For the entire surface, this boils down to the order of
288*815/225=1043 segments - a few minutes on a "normal" computer
configuration, but the number of segments will take their toll in a
frequency sweep. Symmetry unfortunately can't be used, as the
important tail section would sit exactly on the cutting plane. Some
thinking is required in the processing of radiation patterns, and
their storage on disk.
As usual, the amount of storage is proportional to the square of the
number of segments, which itself is proportional to the square of the
problem size in wavelengths. And the solution time is roughly
proportional to the cube of the number of segments.
The solution time is therefore proportional to the 5th power of the
product of the top frequency and the size of the aircraft to be
modeled... Not an entirely trivial task.
Many things have to be decided or checked in the model:
1) Conductor thickness (equal area criterion)
2) constraints on the diameter of meeting conductors
3) number of conductors meeting at a particular junction
Data from other types of aircrafts could be scaled from the results.
Using one model for all scales means that you will need to sweep over
a broader bandwidth, which may not be possible with a single
model. Supposing that you want to sweep from 4 to 21 MHz (5.05:1) and
for aircrafts of relative size 3:1, then the model meshing will need
to be valid over a 15:1 frequency scale... But you can't have too many
segments per wavelength, either.
Perhaps someone here already has a model he/she could share. I've seen
a few papers on aircraft HF antennas. I've also seen work about the
Orion P3C wire antennas. A commercial airliner will usually have a
kind of loop HF antenna on the leading edge of the tail.
>To complete the job I also need a figure to characterise the expected
>receive noise level.
On the ground or on the aircraft?
At HF, even on an inefficient aircraft antenna, I think that the noise
figure is entirely dominated by the environment, and not the
receiver. You need for that an evaluation of the atmospheric noise
distribution depending on the angle of arrival, which would be
weighted with the antenna patterns. See [1] for references.
Do you also need to consider ionospheric propagation models? IMO, it
would start to get a bit complicated.
Another alternative would be to try to contact an airframe
manufacturer... A company like Boeing or Airbus must have done this
type of work already for its own aircraft designs. Maybe they could
share a few radiation patterns.
Alexandre
[1] A.A. Kolosov, "Over-the-Horizon Radar", Artech House, 1987
[2] I'm presently doing a small modeling job involving a square
diffracting plate... I'd be curious to hear what people have to say on
this choice of sampling interval. A paper by Miller and Poggio suggest
between "10 and 20" segments/wavelength, but I don't have the original
Pergamon book cited.
Received on Thu Jul 30 1998 - 11:39:18 EDT
This archive was generated by hypermail 2.2.0 : Sat Oct 02 2010 - 00:10:38 EDT