1. How Much Solar Can We
Use, How Fast, and at What
Cost?
Ken Zweibel
The George Washington University
Institute for Analysis of Solar Energy
National Academies of Science, Engineering, and Medicine
July 29, 2008
2. Some Background
• There’s more solar energy than we’re ever likely to need for
all our energy demands combined
• Making intermittent solar during the day is easy and
relatively cheap
– In the best solar locations, about 15 ¢/kWh for CdTe PV
(First Solar) and claimed for CSP
• We also must take into account solar’s daily and
seasonal variations
• Other important details include
– PV output varies proportionally with local sunlight, which
within the US varies by about a factor of almost two
– CSP is uneconomical in cloudier regions due to its
dependence on direct sun
– HV DC transmission has losses and costs
3. General Comments
• The following is a toolbox of approaches that harnesses solar
in the US at the multi-TW level and aims at minimizing total
cost in terms of solar variation by
– Blending solar to reduce as many natural variations as
possible
– Using existing fossil fuels and nuclear as a back-up (while
reducing fuel use by a large fraction), but never building a
new conventional plant
– Using solar mostly in the daytime and electric storage
(compressed air) only when we must
• Timeliness:
– Uses today’s best solar prices, and prices achievable with
a high degree of confidence during the next 10 years
– Aims at harnessing today’s solar ASAP, not waiting
4. The Opportunity
1 day of unconverted US
solar energy: 48,000 TWh
1 year of US
electricity: 4000 TWh
Imported oil is ¾
of this, if electric
5. Approach
• Conversion of vehicles to plug-in hybrids
• Solar and wind mostly converted in their best resource locations
(Southwest & Midwest), but spread out within those regions to de-
couple weather
• Low-loss transmission lines (HV DC) from location of large fields to
demand
• Wind and solar combined along transmission lines to make smoothly
varying, 24/7 output
• Short-term solar thermal heat storage used to its economic limit
• Compresses air energy storage (CAES) use only for evening peaks (no
overnight or seasonal storage)
• To minimize impact of transmission losses, build large solar farms
wherever there is decent sunlight closer to demand
– Large arrays in sunny Idaho, Florida, Eastern Colorado, Texas,
Utah, rest of CA, Northern Mexico, Eastern Oregon, etc.
– Consider a “beltway” HV DC linking these to reduce weather,
climate, and seasonal impacts
6. How to get the solar and wind electricity
(the first order approach)
Wind
Midwest
Courtesy UniSolar
National Electricity
Transmission
Sola
System to Export
r
Sout
hwe st
Solar Electricity
from Southwest and
Wind Electricity from
Courtesy SunEdison
the Midwest.
15 ¢/kWh solar electric from the Southwest can
be sent nationwide with about 11% losses. We
can get solar electricity in Maine for about
20 ¢/kWh
7. PV Geographic Smoothing:
This is what we want as output nationwide
“Capacity Valuation Methods,” SEPA 02-08, Hoff, Perez, Ross, Taylor, 2008
9. Transmission corridors will pick up complementary wind
along the way to demand
• Wind blows at
night and
winter
(opposite of
solar)
– Of 6000 MW
ERCOT
(Texas) wind,
only 10%
available at
daytime peak
• 24/7 waves of
wind/solar
power
10. Wind and Sun Are Complimentary
High Plains Express Feasibility Study, June 2008, p. 35
11. Increased Capacity Use with Wind and Solar
Lowers Transmission Cost
“We found that by
blending wind and
solar for
geographically
diverse sites, we
can achieve a more
consistent product
for delivery,
thereby offering the
potential for
reducing
integration costs
and improving the
economics and
acceptability of
renewables.” Jerry
Vainineti (co-
author)
From High Plains Express Feasibility Study, HV AC, for wind and solar combination
12. First Solar has a contract to install a thin film CdTe PV
System in Blythe, CA and sell its electricity at
12 ¢/kWh (after incentives)
JUWI Group is
installing 40 MW of
First Solar modules in
Waldpolenz,
Germany. At the time
of the announcement,
it was both the largest
and lowest-priced PV
system in the world at
€3.25/W, which was
then equal to $4.2/W.
A program for 250-
MW of rooftop
systems by Southern
CA Edison has been
signed for $3.5/W.
http://www.juwi.de/international/information/press/PR_Solar_Power_Plant_Brandis_2007_02_eng.pdf
14. Solar Thermal: BrightSource/LUZ II has a contract to install a 400 Megawatts
(expandable to nearly 1 GW) for Southern California Edison
Claimed to be in the same 15 ¢/kWh
range as the best PV (prior to incentives)
15. Daytime Solar Costs (¢/kWh)
Timeframe Intermittent Estimated Cost to Transmission Total at
Cost in Best Make Fully Usable Cost (¢/kWh 2000 miles
Locations (daytime only) per 2000 (East
miles) Coast)
15 1 4.5 (solar) or 19 - 21
2008
3.2 (solar +
wind)
8 0.5 (falls 3.8 (solar) or 11 - 12
2015
proportionally to cost 2.6 (solar +
of solar) wind)
16. Low CF High CF
capacity factor 26.7% 45.0%
distance 2000 2000
losses per 1500
miles 6% 6%
capital cost of
transmission
wire alone 1.293333 0.767377778
loss at DC-AC 1% 1%
3 stations
per 1000 Capital cost of
miles DC-AC eqpt 1.733333 1.028444444
feedstock elec 8 8
% losses total 9% 9%
losses cost 0.783382 0.783381522
3 stations
per 1000
miles Capital cost 3.026667 1.795822222
total added
costs 3.810048 2.579203745
17. How Fast Can Solar Be Scaled Up?
• Recent solar PV growth rate has been
around 50% per year, to about 5 GW
– There have been some undesirable
bottlenecks and price increases
– Having an agreed-upon timeline for expansion
could avoid future bottlenecks
• Solar thermal appears to be built with
basic materials and components,
suggesting rapid growth would be possible
18. Possible Growth of PV from 5 GW (world, 2008)
Assumes 50%/yr Growth, 1/3 in US
4000
US electricity today
3500 GW World Annual PV
GW US Annual Installed PV
3000 GW Cumulative US Installed
US PV Installed
PV
TWh/yr in US SW
2500
2000
1500
1000
500
0
09
11
12
14
17
19
20
22
08
10
13
15
16
18
21
23
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
19. What about CSP and Wind?
• CSP – assume similar level – 3700 TWh by
2023
• Add in 1000 TWh of wind (DOE & various
estimates)
• Deploy up to 8400 TWh/yr non-CO2 renewables
by about 2023
– Not quite as fast as Gore’s Plan
• Perspective: US electricity 4000 TWh; imported
oil if displaced by electricity, 3000 TWh
20. Using Daytime Solar Well
• Now – can add solar • Later – solar
and move daytime overwhelms
fossil fuels to meet traditional midday
solar/wind evening electricity demand
shortfalls and post-2020; must be
nighttime plug-in used for midday plug-
charging* in charging, with
some solar and
• Wind meets a large
nuclear moved to
part of nighttime
nighttime by CAES
charging
*Better than Pickens Plan of using natural gas for cars, since this displaces the natural gas with solar
21. Economic Perspective
• Cost of smoothly varying daytime solar, nationwide (lower in
proximity to US SW)* :
– Up to ~20 ¢/kWh now (less in CA, Midwest, etc.)
– Up to ~12 ¢/kWh 2015 (less in CA, Midwest, etc.)
• First Solar has $2/W (8 ¢/kWh) installed systems as their
stated goal in the next few years; and wafer silicon says that
once silicon prices subside, something similar is possible
• Cheaper transportation, more costly electricity at first, permanent
solution that is much less expensive than business as usual
– Saves society money with switch to electric plug-in hybrids
– Removes the terrifying threat of continually escalating fossil fuel
prices and obligations
– Grows jobs and dollars domestically
• First decade versus coal needs ITC as CO2 offset
*Does not include additional charge by local utility for AC distribution
22. Footnote to Economics
• Nothing in economics takes into account the long life of
PV
• Once paid off in 20-30 years, PV has another 30 plus
years of life
• No one will plow under a PV system that is at 85% of full
output in 30 years and costs practically nothing to
maintain
– Meanwhile, who knows where fossil fuel prices will be?
• This is not captured anywhere in above, because the
above is “levelized cost of energy” over life of the loan
– This will be a great gift to future generations, as the Hoover dam
and TVA was to ours
23. Impact
• Eliminate almost the equivalent of all of our current
energy use (100 Quads) in 15 years
– Up to 8400 TWh/yr (2000 of PV, 2000 of CSP,
1000 of wind)
• ~10,000 TWh used electrically can displace
about 100 Quads primary energy
• Reduce CO2 by almost 5 gigatons (almost all
today’s energy emissions) versus 2023
“business as usual” level
– Actually reverse carbon dioxide buildup
• Achieve energy self-sufficiency and strong economy
24. Action Items
• Build solar and wind as fast as humanly possible, mostly in
large fields in high resource locations
• Build HV DC to send to demand, include source and
geographic diversity on each HV DC line to smooth output
• Shift vehicle fleet to plug-ins, use mostly wind and shifted
natural gas and coal at night to charge
• Replace natural gas and coal during day with solar
• Use existing conventional capacity to fill in daytime gaps and
meet any missed evening peaks
• Require large proportion of midday charging as solar output
outstrips daytime demand post 2020
• Add CAES to meet evening peaks, starting with storing
daytime nuclear electricity
25. Imagined Scene
• NY Mayor: “Should we build our electricity
aqueduct to Kansas and use their wind and
solar, or extend it all the way to Nevada?”
• Official: “Well, the sunlight in Nevada is 25%
better, and the time zone difference means we
can meet our evening peak without storage. But
Florida is offering us a deal.”
• Mayor: “What does New Jersey want, or do we
go through Pennsylvania?”
• Etc.
26. Thanks to James Mason
(ASAP.org), Vasilis Fthenakis
(BNL & Columbia), Tom Hansen
(Tucson Electric), Bill Bailey, and
many others
K. Zweibel
zweibel@gwu.edu
202-994-8433
Institute for Analysis of Solar Energy
The George Washington University