[TowerTalk] Modeling, ground, etc.

[Jim Lux]

Folks interested in how NEC was validated might find the following LLNL 
report interesting:

G.J. Burke, "Recent Advances to NEC: Applications and Validation", 3 March 
1989, UCRL-100651

http://www.llnl.gov/tid/lof/documents/pdf/210389.pdf

This one is particularly interesting because it gives some history (that 
NEC-3 was partly developed to address things like ground rods and EMP, and 
structures that are small in comparison to wavelength (NEC-3VLF)).

It also makes the telling comment (page 3-20 (22 in the pdf)):
"... The most difficulty has been encountered in modeling ground screens 
floating above the ground. ..." and goes on to how to construct your model 
to deal with it.



Another LLNL report, particularly relevant to wires close to (or in) the 
ground, etc. is:

E.K. Miller & F.J. Deadrick, "Analysis of Wire Antennas in the Presence of 
a Conducting Half Space: Part III - The Buried Antenna", 25 Feb 1977, UCRL-5228

http://www.llnl.gov/tid/lof/documents/pdf/175262.pdf

One take home message from the above is that if you're within, say, 0.2 
wavelength of the ground, you'd better be using the Sommerfeld ground analysis.



Another useful work (not online, unfortuately) is Hagn's paper on measuring 
ground constants:

G.H. Hagn, "HF Ground Constant Measurements at the Lawrence Livermore 
National Laboratory (LLNL) Field Site", Applied Computational 
Electromagnetics Society (ACES)(I think), 1988

Of some interest is that the relative epsilon (permittivity) varied with 
frequency.  At one test station, the best estimate ranged from 182 at 2MHz 
to 17 at 30 MHz.  The measurement uncertainty is quite large for the lower 
frequencies. At 7MHz, the best estimate was 62, with a lower bound of 38 
and an upper bound of 217.

Quite a ways from assuming 13/.005, isn't it?


They also assessed the horizontal homogeneity (which is of interest to us, 
when considering how to model ground radials and ground screens).  "The 
small scale variations seemed rather large, but the variation seemed to 
decrease with increasing measurement frequency."  Very interestingly, (or 
not so suprising, given the crew doing the measurements, most of whom, if 
not all, were hams) they did some surveys at 7,14, and 30 MHz.

In general, they concluded that variations were about an order of magnitude 
(10:1) for conductivity, and half an order of magnitude (3.16:1?) for epsilon.

One of the comments from this paper is that the "uniform slab" model cannot 
be used if the water table is within one skin depth of the surface.  (The 
Livermore ground constants work out to a skin depth on the order of a 
meter, and the water table was some 80 feet down).