[Discuss] Eclipses Re: Great talks last night, however...

[grg-webvisible+blu at ai.mit.edu]

On Sat, Jul 22, 2017 at 12:31:08PM -0400, Richard Pieri wrote:
> On 7/22/2017 1:14 AM, Gregory Galperin wrote:
> > You seem to be claiming here that the lithium ion batteries Musk is selling
> > have less than a 10% charge/discharge efficiency.  But when I look for a
> > number on charge/discharge efficiency of lithium ion batteries I find
> > numbers in the range 80%-90% efficient depending on battery age (with
> > measured numbers like 86%).  This says you'd need ~15% more solar power to
> > compensate for the losses, not 1000% more.  Can you point me at something
> > which explains this discrepancy?
> 
> Certainly. Musk's figures are marketing. Actual Tesla, Leaf and Volt
> owners see much lower figures, as low as 50% for newish battery packs
> (75-80% is typical). As a Li-ion battery wears and its charge density
> decreases the thermal waste increases.

OK, so here you're saying that instead of a <10% charge/discharge
efficiency, batteries actually have a 75%-80% charge/discharge efficiency?
If so, we now pretty much agree on this point.

And FWIW I wasn't citing Musk's numbers at all; every university research
source I see says 80%-90%, starting at 90% for newer batteries and dropping
to 80% as they age, with a life-averaged value of about 86%.  Our ranges
overlap at 80%, so let's just call it 80%.  So you need to generate 20% more
solar power to compensate, rather than 1000% more.


> > OK, so the 10,000 km^2 would have to be 11,500 km^2 if you put all of it up
> > at the Canadian border above NY & VT.  Not sure why anyone would do that
> > instead of using some area in TX, but even if someone did, this doesn't
> > seem like a big deal -- 11,500 km^2 is 0.15% of the land area of the
> > continental US, rather than 0.13% of it for 10,000 km^2.  Hard to see how
> > that 0.02% would be a dealbreaker even if Texas did secede.
> 
> This assumes that the land up north is flat like Texas. Hilly terrain
> creates shade which reduces the time per day that PV panels are

See my previous email that there's more flat Great Plains in the North than
in Texas.  Given that, seems like the "north has flat land" assumption is
safe ;)


> effective. While there is flat land up north, much of it is either farm
> or forest. Good luck stripping that for large solar farms.

Remember, we're only looking for 0.15% of land area in all.  Farms have a
good amount of area already stripped (and incidentally already have barns
or silos with large roofs good for mounting large solar panels).  If people
live in forests, that land is already similarly stripped; if people don't
live there, I'm not sure why we'd be trying to put solar panels there
instead of in other places (like deserts and plains).  So overall, I don't
foresee a problem finding 0.15% of land in the North which suits.


> > ranges (California to Canada)...  Do you know of an analysis showing that
> > northern states have "much less flat land"?
> 
> Look at a map. Like this one:
> http:%2F%2Fwww.earthscienceeducation.org%2FDi-BiosphereNotPost%2FEPA-EcoregionsLev01x600.jpg&sp=9f6cecc9d6e3992197bd8e271bb214fc

OK, I'm looking at that map (http://www.earthscienceeducation.org/Di-BiosphereNotPost/EPA-EcoregionsLev01x600.jpg)
and I see "Ecological Regions" drawn which have little variation as you
go north-south.  That seems to show that northern states have the same
composition as the southern states, in contrast to the claim.


> The eastern half of the country is off the table because it's forests
> and I see clear-cutting to make room for solar farms to be a hard sell.

Well, it marks nearly everything east of the Mississippi as "forest", and
I'm pretty sure there are some fairly flat, treeless large cities in there
somewhere.  I'm beginning to think this particular map might not be
depicting the information salient to this discussion.  And remember, we're
only looking for 0.15% of land area - we don't need the eastern half of the
country to be 100% flat and clear, we only need it to be 0.15% flat and
clear (or already covered with e.g. roofs).  Categorically dismissing half
the country misses a lot of opportunities.

(Also note that "forest" on the map doesn't mean "not flat", in reference
to the question this was presented as an answer to.)


> The Great Plains are partially off the table as well because that's
> where the bulk of our livestock production comes from and that's our
> number two in agriculture after corn (maize).
...
> The Pacific Northwest is out due to forests and mountainous terrain much
> like the Atlantic Northeast.

Again, looking for 0.15% of land.  I wouldn't write off the entirety of the
Great Plains and say there's no room for any solar there just because some
of that land is used for cows and corn.


> The Nevada and southern California deserts might be good choices.

Agreed!  And Utah, and Arizona, and New Mexico, and large parts of
Colorado, Wyoming, Idaho, Oregon, and Washington by your map.  And don't
forget Great Plains states like Texas, Montana, North Dakota, South
Dakota... hey, I think we're over 0.15%!


> > Oh, maybe you mean that a panel on the equator is in the dark half the day,
> > and you're arranging to put your solar panels in a sun-synchronous orbit so
> > they're never behind the Earth?  OK, that would make for a factor of pi
> > difference, but that's only halfway to the "at least an order of magnitude".
> > What accounts for the rest?
> 
> Partially. An equatorial GBSP station is actually only good for about
> 7-8 hours a day, not 12. Efficiency drops off as the angle increase and
> more atmosphere needs to be penetrated (problem shared by northern
> ground stations).

Right - as in my prior email, when you do the math it comes out to a factor
of pi (and 24/pi is 7.64 hours, within the range you give).


> Combine this with PV efficiencies that are much
> greater than equatorial PV (the figures I've seen say at least 140%
> greater). You can do the math but I think that's close enough to a order
> of magnitude for a first order approximation.

Again, if by "efficiencies" you mean total solar influx, it's 140% *OF* the
ground value, not 140% GREATER.  It's only 40% greater.  1370 W/m^2 in space,
~1000 W/m^2 on the ground.  But since we all seem to agree that space-based
solar power really isn't practical now, I think we can drop this point.


Overall, if the only remaining point of disagreement about the feasibility
of powering the US's electrical needs with 10,000 km^2 of ground-based
solar + batteries is that you look at the Great Plains, and Spokane, and
Eugene, and Albany, and Bangor, and heck, Acton and Hopkinton, and you
can't find anywhere you could put a solar panel, then I think we've
concluded the technical part of this feasibility discussion.

FWIW, on that last non-technical bit, I and I wager many others on this
mailing list see very many places in all the named locales which have good
potential for solar.  And that's one of the great things about solar power:
you don't need enough contiguous space to build a half dozen cooling
towers, you just need a few m^2 of suitable space to make it worth putting
some panels up and making progress towards the overall 10,000 km^2.  (You
weren't looking for a single contiguous clear & flat square of 60 miles X
60 miles to put up a single 10,000 km^2 solar farm, were you?)

--grg