NEC-LIST: Buried Radials

Grant,

You have asked:

Date: Sat, 13 Jun 1998 17:51:30 -0400
From: Grant Bingeman <DrBingo_at_compuserve.com>
Subject: NEC-LIST: NEC4D and buried antennas

Has anybody used NEC4D to model buried transmitting and receiving
antennas? Has there been any validation?

Grant Bingeman

Comment:

Comparing with measured data for 120 buried radials is not a very
informative test. Losses are small, and NEC predicts small losses.
Comparing impedance (computed vs. measured) is also a problem, since
the measured impedance for MF broadcast antennas is influenced by
stray effects, such as base insulator capacitance, capacitance of guy
insulators, lighting transformers, etc.

We need data which show how field strength and impedance change as
number of radials change --- and the only set of data that I know of
are the very early (primitive) data of Brown et. al, 1937.

I have tried to find agreement between measured and computed FI and
impedance, see below. I also append Jerry Burke's response to an
earlier copy of this text, in which Table 1 showed results only for
average ground (3 mS/m, 13).

I have noted George Hagn's comments.

Jack, VE2CV

Buried Radials - Revised
_________________

We have shown that we can compute the performance of MF broadcast
antennas employing elevated radials using NEC-4D [reference Belrose,
Communications Quarterly, Spring 1998 issue]. But we have yet to
validate that we can compute the performance of a monopole with buried
radials. This is a progress report, on our work, which seems to say
yes-and-no??

The ground systems used by MF broadcasters are based on the results of
the classical experimental study by Brown et. al. [1937]. Therefore,
any comment concerning the ability of NEC-4 to predict field strength
and impedance for antenna systems should look at a comparison between
prediction and the measured parameters by Brown et. al.

In an earlier study the I did not find a very close agreement in being
able to predict impedance or field strengths, in fact a discouraging
disagreement. On closer study of the paper by these authors we note
that the graphed field intensities at one mile given in the paper were
determined from a single measurement at 0.3 mile (482.8 m), and the
ground conductivity was said to be poor (a conductivity of 2 mS/m was
mentioned) and the frequency (3 MHz) well above the MF broadcast band.
Hence the value plotted is not the unattenuated field strength at one
mile, which is the usual reference for determining power radiated, and
antenna efficiency. But the measured trends should be correct.

We have for one antenna height 75-degrees (h = 20.833 m for the
frequency used 3 MHz), and one radial length (length 90 feet, or 0.274
wavelengths), computed (using NEC-4D) field strength (at 1 mile) in
exactly the way it was measured. Details of our model are given in
Note 1. We repeated the calculation for four ground conductivities, 2
mS/m, 3 mS/m, 5 mS/m and 10 mS/m. We find the best agreement between
our computed values and the measured values are for good ground
conductivity (10 mS/m, 30). This seems reasonable, since the
photographs seem to show agricultural ground. See Table 1.

Clearly, according to NEC-4, for low to average conductivities, the
radials which are a part of the antenna system, dominantly affect the
antenna's impedance, particularly the resistive component.

                                Table 1
Average ground
________________
Number Measured Predicted Antenna Impedance
of Radials mV/m mV/m Measured Predicted

15 153 137.2 29.0 - j 45 22.2 - j 52.9

30 162 157.5 24.6 - j 45 17.0 - j 54.6

60 176 174.6 23.5 - j 45 13.7 - j 56.3

113 179 184.4 22.3 - j 45 12.1 - j 57.5

PEC 22.1 - j 67.1

Good Ground
____________

15 153 158 29.0 - j 45 25.8 - j 59

30 162 169.3 24.6 - j 45 22.8 - j 60.3

60 176 178.4 23.5 - j 45 20.6 - j 61.4

113 179 183.7 22.3 - j 45 19.2 - j 62.3

Notes

1. The antenna was a 2.5 inch diameter (63.5 mm) steel pipe, mounted
on a hardwood base insulator (height not known but probably 10-12
inches, say 30 cm). We used the segment taper option available with
EZNEC pro. We used a minimum segment length of 0.3 m, such that the
minimum segment length/diameter equaled about 5 for the antenna.
Maximum segment length about 1.17 m (conservative segment length for a
frequency 3-times the reference frequency); and changing the maximum
segment length to 2.34 m had a neglible effect. Our antenna height is
measured from ground level; hence the height of the base insulator is
ignored.

For our model the radial wires were buried 20 cm, Number 8 wire. It
is not clear how deep the wires were buried for the experiment, since
there is a misprint in the paper, but it might have been 6 inches (15
cm). For our radial wire model end 1 of the first segment (length 0.3
m) was at the surface of the ground, and end 2 was sloped to the
buried depth. The first segment on the long horizontal part of the
buried radial was also 0.3 m in length.

Reference

G.H. Brown, R.F. Lewis and J. Epstein, "Ground Systems as a Factor in
Antenna Efficiency", Proc. IRE, 25, June 1937, pp. 753-787.

Date: Fri, 17 Apr 1998 17:06:07 -0800 (PST)
From: BURKE_at_FLAME.LLNL.GOV
Subject: Re: On Validation of NEC-4 for Buried Radials
To: john.belrose_at_crc.ca

Jack,

I tried comparing with Brown, Lewis and Epstein's results back in
1983, when I first got the NECGS program, optimized for radial-wire
ground screens, going. I did not find much agreement, but do not
think the problem is in NEC. Back in 1937 they had relatively crude
measurement equipment and little understanding of the problem. They
obviously had little idea of the ground parameters, and the conditions
of the measurements are not well documented. I guess you noticed that
in extending their field strength measurements at 0.3 miles to 1 mile
they assumed a 1/R rate of change. Norton's work was published the
same year as Brown, Lewis and Epstein's paper, so they may not have
been aware of ground wave behavior.

I had hoped to compare NEC results with their measurements for the
decrease in current along a radial, but the agreement was not good.
The NEC results looked more reasonable. NEC results for current have
been compared with the analytic solution of Olsen et al. for an
infinite wire above ground, and with Wait's solution for an infinite
buried wire with excellent agreement. With those results, it seems
that if NEC is not right for these simple problems Maxwell's equations
are not right. Radial-wire screens are a little more complicated than
a single wire, but a radial in a sparse screen at a distance from the
junction looks somewhat like an isolated wire, and the rate of current
decay did not agree with the old measurements.

Ray King was working here at the time, and he had worked with Jim Wait
on the Compensation Theorem approximation for the input impedance of a
radial-wire screen (Collin, Antenna Theory, Part II.) Ray thought it
was a waste of time to compare with such old measurements. NEC is in
good agreement with the Compensation-Theorem solution for conditions
where that approximation is expected to be valid. That requires a
screen with enough wires that the surface-impedance approximation is
valid and the condition:

screen radius/wavelength > 1/Sqrt(Abs(rel. complex permittivity)).

Another check on NEC was to model the radial-wire screen buried in a
dielectric ground and integrate around the sphere at infinity to get
the radiated power. The far-field power was in good agreement with
the input power computed at the source. This test involves the
junction penetrating the interface, and the agreement would be
unlikely to occur if there was any major inaccuracy in the solution.
To integrate the radiated power you need a command such as

RP0,721,3,1002,0.,0.,0.25,0.75

for 120 radials. Very small increments are needed in elevation angle,
since there is a very sharp spike of radiation into the ground at the
critical angle for internal reflections. An azimuthal angle increment
of 1/4 of the angle between radials is good, since there may be
significant power in the horizontal polarization at 1/4 of the way
from one radial to the next, while horizontal power is zero in the
plane of a radial or half way between radials by symmetry. With a
lossy ground you can integrate over the upper hemisphere. Then half
the "average gain" will be the radiation efficiency, and that times
input power will be the radiated power. Comparing that radiated power
with the input power can give a measure of the ground-loss resistance.

If people are still relying on the Brown, Lewis and Epstein
measurements it seems like some modern measurements are needed. Of
course, that would probably upset a lot of FCC rules. I am sure there
have been a lot of measurements of antennas on radial-wire screens,
but do not know how many have been published. Maybe scale model
measurements would offer a better chance of getting results under
known conditions.

Jerry Burke
LLNL

_____________________________________________
John S. (Jack) Belrose, PhD Cantab, VE2CV
Senior Radioscientist
Radio Sciences Branch
Communications Research Centre
PO Box 11490 Stn. H
OTTAWA ON K2H 8S2
CANADA
TEL 613-998-2308
FAX 613-998-4077
e-mail <john.belrose_at_crc.ca>
_____________________________________________
Received on Thu Jun 18 1998 - 10:03:18 EDT