The common thing you can do at home, is to disconnect
your sensitive equipment from the power grid, at the first
sign of an electrical storm. All the "must remain operational"
setups switch to locally generated power or can operate
from true protected isolation transformers (Big honking
things) sometimes both. The TelCo has functioned on
large banks of batteries in their exchange buildings with
isolation from the grid, ever since they first hooked up to
someone else's power grid and weren't generating their
own power on site.
If you disconnect power also disconnect signal - particularly
telephone.
The idea that one cannot protect against lightning induced surges is
at odds with what I have read. (This does not include a direct hit to
a house.)
The vast majority of the energy of a strike is of course dissipated in
3 miles of sky as heat and light.
If a strike hits primary (high voltage) power lines more is dissipated
in lightning arresters. On a power line the strike will be conducted
in all available directions and dissipated at multiple points.
François Martzloff was the NIST guru on surges until he recently
retired. He wrote the NIST guide as well as may technical papers
including:
http://www.eeel.nist.gov/817/pubs/spd-anthology/files/Neutral earthing.pdf
"The Effect of Neutral Earthing Practices on Lightning Current
Dispersion in a Low-Voltage Installation"
One of the test scenarios was a US system with 3 buildings connected
by separate service drops to a single transformer. All 3 houses had
service panel surge protection devices. One of the houses was hit with
a 100,000A surge with rise time of 10 microseconds and duration of 350
microseconds. This is a large hit with a very long duration. The hit
was to the neutral service connection at the house. In total this is
close to a worst possible case.
*1* The result was
21,000A was conducted to earth at the building (this lifts the ground
potential at the house)
33,000A flowed away from the building on the service neutral to the
transformer and other 2 buildings
23,000A went through each of the hot-neutral protectors and flowed
away from the building on each hot service conductor to the
transformer and other buildings.
Of the 100,000A surge, 2 of the 3 service panel surge protection
devices had a peak current of 23,000A
The dissipation at the building surge protector was 3500J
*2* If the same surge hit one of the other buildings the dissipation
at the surge protectors at this building was 840J.
*3* If the surge was shorter - 20 microseconds - which is probably
more realistic, the surge protector dissipation would be 200J, or if
the hit was to one of the other building the surge protector
dissipation at this building would be 80J
For comparison, the *plug-in* surge protector I recently bought had
rating between each pair of wires of 30,000A and 590J (1180J for both
hots to ground). But my plug-in protector was rated at a higher
current than occurred in any case, and the energy rating was higher
than occurred except the strong hit the building (*1*). Service panel
protectors would likely be rated higher. And this is near worst case.
The point is that you can survive a near strike. It is not practical
to protect against a direct hit to your building (unless you install
lightning rods), but direct hits are very rare.
==================================================
Another Martzloff paper looks at a MOV (simulating a plug-in
suppressor) at the end of a 10-50 meter branch circuit. The surge is
2,000-10,000A, and I believe 25 microseconds. There is an arc-gap at
the source end with a breakdown voltage of 6,000V. US systems will
have arc-over at panels and receptacles at about 6,000V. Arc-over
dumps a large percentage of the surge to earth. Branch circuit
impedance greatly limits the surge current to the MOV.
In all cases the energy dissipated at the MOV was less than 1J except
for a 10M branch circuit and, ironically, the lower current surges
below 5,000A. Contrary to intuition, at all branch circuit lengths the
energy dissipation at the MOV was lower as the surge current went up.
That was because the MOV acted to clamp the voltage at the source
spark gap. With the short branch circuit and lowest surge currents,
the MOV prevented the gap from arcing over at all. Higher current
surges forced the voltage at the gap up faster causing it to break
down faster and dump more of the energy to earth.
I don't remember the branch circuit current was indicated, but it
would be far below the max surge current of 10,000A. The max energy
dissipation at the MOV was 22J. Plug-in suppressors are readily
available with ratings higher.
===========
In another guide Martzloff said "In fact, the major cause of TVSS
[surge suppressor] failures is a temporary overvoltage, rather than an
unusually large surge."
What I have read, including the IEEE and NIST guides, indicates
surviving even relatively close lightning induced surges is quite
possible with adequately sized surge protectors. The IEEE guide list
of effective measures is:
earthing
single point ground
service panel surge protector
plug-in surge protector.