At 12:23 PM 1/4/2007, K4SAV wrote:
>..."does anyone run simulations or calculations on lightning strikes. A
>direct stike is 100 kA with a risetime of 8 us, correct? If so, using an
>inductance of 200 nH/ft, the voltage drop along 1 foot of wire is 2.5
>kV. And we can
>calculate the low frequency impedance of a ground system. Let's say it is 1
>ohm. Now we have 100 kV between the tower base and earth ground. ".....
>A common lightning strike may be 100KA but that's not considered
>typical. The typical number used is now 20KA (used to be 17 KA).
>However I saw an article that said the maximum current ever recorded was
>500KA, so that probably means there are a few that are even larger than
>that but never recorded. Of course if you are designing for protection,
>you probably want to use something much greater than typical, and even
>100KA may be a little low, depending on how safe you would like to be.
>There are a couple of large hurtles standing in the way of doing good
>simulations of the currents and voltages present in a tower system
>during a strike. Most everything in a tower system can be modeled fairly
>accurately in SPICE with the exception of the earth. The non-linear
>conduction that happens during the strike where underground arcing and
>current saturation take place is very difficult to model, mainly because
>I don't think anyone really understands this event very well. If a model
>for this event was available, SPICE could calculate all the currents and
>voltages present due to conduction.
>The one thing SPICE can't do is calculate the induced voltages. For
>this you need a field solver, or something like NEC that can do
>calculations for electromagnetic fields. NEC unfortunately can't handle
>all of the models and complex structures needed to simulate the whole
>system. It seems to me that what is needed is a tool that combines the
>capability of SPICE with that of NEC. I am not aware of anything that
>does this. Maybe someone else can comment on this subject. A field
>solver used in PC board design has some of these capabilities but I
>don't know if any of these programs can handle a general environment, or
>are relegated to PC board use.
The literature shows several folks doing a NEC model with a frequency
sweep, so they can determine a coupling model (vs frequency). Then,
they take the lightning current waveform (which is well known) and do
a fourier transform, multiply by the couplings, and get a induced
current vs frequency, then do a reverse transform to get the waveform.
There's also some people who have used a model of the lightning that
actually propagates. That is, they break the stroke up into little
chunks, use NEC to model the interaction of each chunk with the
antenna or victim system, then sum all the contributions.
Either way, that gives you a time domain model of the source
voltage/current that you can put into a circuit model. I haven't
seen SPICE used, but that might be what they use.
>Another special capability that would be needed in this tool would be
>the ability to predict arcing. Since your tower design should be such as
>to prevent arcing, maybe the tool wouldn't have to actually calculate
>values due to arcing, for that you just change the design, but it would
>have to predict them.
There are spark gap models for SPICE that work fairly well. I think
you'd be looking more for a general approach, and do sensitivity
analysis, rather than try to model every little detail.
>So you can do some simple calculations either manually or using SPICE,
>including some large assumptions, and maybe get a gross estimate of some
>of the voltages and currents, but the answers are likely to have large
We in the modeling and measurement basis prefer the term
"uncertainty" rather than error, but you're right.
TowerTalk mailing list