Utility-Scale Solar 2012

An Empirical Analysis of Project Cost, Performance, and Pricing Trends in the United States

Mark Bolinger & Samantha Weaver

Lawrence Berkeley National Laboratory

September 2013

This research was supported by funding from the U.S. Department of Energy’s SunShot Initiative. 1 Presentation Outline

• Recent market growth has resulted in a critical mass of project-level data ripe for analysis • Key findings from this inaugural edition  Installed Costs/Prices  Operating (O&M) Costs  Performance (Capacity Factors)  Power Purchase Agreement (“PPA”) Prices

A few background notes about this first edition: • Certain data (e.g., O&M costs) were still rather limited for this first edition, but are expected to become more widely available in future years • For this first edition, we define “utility-scale” as any ground-mounted project that is larger

than 2 MWAC (we may raise this threshold for next year’s second edition)

2 Utility-Scale Solar: Young But Growing Like a Weed

Source: GTM/SEIA’s 2012 U.S. Solar Market Insight • In the United States, utility-scale PV increased from just 5% of total annual PV installations in 43% 2008 to 54% in 2012 (chart left) Utility-scale PV was 13% in 2009 and just 51% • 2012 was the first year in which utility-scale 5% in 2008 51% PV made up the largest segment of the U.S. 34% 55% 54% PV market – a distinction that it is projected to

40% retain through at least 2016 23% 31%

20 Utility-Scale CSP & CPV (GW-AC) 18 (GW) Utility-Scale PV (GW-DC) 16

• Including CSP/CPV, cumulative installed Solar 14 utility-scale solar capacity could double 12 Scale in 2013, and again by 2015 (chart right) - 10 • More than 18 GW of utility-scale solar 8 projected to be online by the end of 2016 6 Mostly SEGS trough 4 projects from the 1980s  2013-2016 GTM/SEIA utility-scale solar 2

projections account for 35%-40% of the total Utility Cumulative announced and contracted pipeline that they 0 track (or 118% of just the contracted pipeline) 2006 2007 2008 2009 2010 2011 2012 2013e 2014e 2015e 2016e

Source: GTM/SEIA’s U.S. Solar Market Insight reports 3 Installed Prices Have Fallen, But Pace of Decline Slowed in 2012 $10 Capacity-Weighted Averages $9 Individual Projects - $/W-DC Individual Projects - $/W-AC $8 $7 $6 $5 $4 $3 $2

Installed Price (2012 $/W) (2012 Price Installed $1 $0 2007-2009 2010 2011 2012 n=10 (81 MW-AC) n=17 (196 MW-AC) n=54 (429 MW-AC) n=113 (838 MW-AC) Installation Year • Installed prices are shown here in both DC and AC terms, but because AC is more relevant to the utility sector, all metrics used in the rest of the report (and this slide deck) are expressed solely in AC terms • Anecdotal evidence of further installed price declines in 2013/2014

 Example: PNM recently filed for regulatory approval of 23 MWAC of thin-film PV projects to be built in 2014 at a contracted price of just $2.03/WAC, compared to $2.29/WAC for 20 MWAC of PV currently under construction in 2013 and $4.15/WAC for 22.5 MWAC built in 2011 4 Installed Price Decline Led By c-Si, While Thin-Film Prices Held Steady 10 c-Si, Fixed-Tilt c-Si, Tracking Thin-Film, Fixed-Tilt Thin-Film, Tracking 9

8 ) AC 7

6

5

4

3 Installed Price $/W Price (2012 Installed 2 Individual Project

1 Capacity-Weighted Average

0 Year: 2007-09 2010 2011 2012 2007-09 2010 2011 2012 2007-09 2010 2011 2012 2007-09 2010 2011 2012 Projects: 3 3 27 83 3 7 14 22 4 7 12 7 - - 1 1 MW: 7 20 146 386 39 54 155 241 35 123 108 186 - - 20 25 c-Si, Fixed-Tilt c-Si, Tracking Thin-Film, Fixed-Tilt Thin-Film, Tracking • An explosion of c-Si projects (particularly fixed-tilt) as prices have converged • Not much installed price difference between fixed-tilt and tracking  Could relate to DC/AC nameplate ratio – DC often oversized for fixed-tilt systems (raises price in AC terms) • Circled high-priced c-Si/fixed-tilt outliers in 2012 are often small, behind-the-meter, and/or installed on top of capped landfills (i.e., high-value and/or customized installations) 5 Economies of Scale Most Evident At Low End of System Size Range $9

) c-Si, Fixed-Tilt

AC $8 c-Si, Tracking $7 Thin-Film, Fixed-Tilt $6 Thin-Film, Tracking

(2012$/W $5 Trendline (All Systems) $4 Price $3 $2 Figure only includes PV projects installed in 2012 Installed $1 $0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

System Size (MWAC) • For many projects, the basic modular unit is a 1-1.5 MW “power block”  A pre-fab package of components (modules, trackers, inverters, controllers, SCADA) that is easily scalable  e.g., SunPower’s 1.5 MW “Oasis” power block – a complete solar project “in a box” • Once you move beyond installing a few power blocks, economies of scale appear to diminish (or perhaps be offset by higher costs elsewhere)

6 Installed Prices By Region: Convergence With the West

$2/WAC diff in 2010 declined to $0.8/WAC in 2011 and $0.2-$0.7/WAC in 2012  Southeast and Northeast have had more of a learning curve to travel down  Bigger projects in the West perhaps more likely to get bogged down in costly permitting and environmental issues

7 O&M Cost Data Still Very Thin, But Largely Consistent With Early Years of Cost Projections 160 Genesis (CSP trough) yr) • Due to limited empirical data at - 140 AC this early stage in the market, 120 Total OpEx 100 this first edition compares Just 80 O&M projected project-level O&M 60 costs pulled from bond rating Solar Star (tracking c-Si) 40 agency research (top graph) to

20 Topaz (fixed-tilt thin-film) what limited empirical data are Operating $/kW (2012 Costs Operating 0 available so far (bottom graph) 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 70 • Results suggest that actual

year) 60 operating costs (from a VERY - AC 50 limited sample) are consistent with early year projections: 40 $20-$40/kWAC-year for PV, and 30 5 projects Generation-Weighted Average (PV only) overlap ~$60/kWAC-year for the lone 20 Tracking c-Si (1 project, 25 MW) CSP parabolic trough project in

O&M Costs (2012 $/kW (2012 Costs O&M Fixed-Tilt c-Si (2 projects, 15 MW) 10 Fixed-Tilt Thin-Film (5 projects, 22 MW) the sample CSP (1 project, 75 MW) 0 2010 2011 2012 8 Oversizing DC Array Boosts PV Capacity Factor

35%

30% R² = 0.0058 R² = 0.6036 25%

20%

15%

10% Net Capacity Factor Net Capacity Fixed-Tilt 5% Tracking 73 pre-2012 projects totaling 743 MWAC 0% 100% 105% 110% 115% 120% 125% 130% 135% 140% 145% DC / AC Nameplate Ratio • With the decline in module prices, some projects have oversized the PV array relative to inverter capacity as a way to broaden/flatten the generation profile throughout the day, thereby also boosting capacity factor (in AC terms) • This surrogate/synthetic form of “tracking” makes less sense (both cost- and performance-wise) for projects already using tracking systems

9 More on PV and CSP Capacity Factors

35% Top Graph (PV): NCF’s highest in Generation-Weighted Average 30% Individual Project the West; tracking boosts NCF 25% by ~20%; thin-film appears to 20% beat c-Si in the West (higher 15% temps), but not elsewhere

Net Capacity Factor Capacity Net 10% Bottom Graph (CSP): 5% • SEGS plants from the 1980s still 5 projects 5 MW57 16 projects 16 MW187 project 1 MW20 3 projects 3 MW15 1 project 1 MW10 1 project 1 MW13 40 MW40 projects 19 7 projects 7 MW238 3 projects 3 MW36 0% projects 15 MW115 c-Si Film c-Si Film c-Si Film c-Si Film c-Si Film c-Si Film going strong (on par with 2007’s Fixed-Tilt Tracking Fixed-Tilt Tracking Fixed-Tilt Tracking Nevada Solar One) Northeast Southeast West • SEGS III-IX performance has 30% SEGS III-IX and Nevada Solar One (dashed) converged in recent years due to 25% 1603-funded upgrades at the two lowest-performing projects 20% • SEGS I and II also upgraded, but 15% have lower capacity factors due SEGS I & II 10% to smaller collector fields

5% • Sierra SunTower pilot-scale Sierra SunTower Net Capacity Factor (solar portion only) portion (solar Factor Net Capacity “power tower” has not met 0% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 performance expectations 10 Levelized PPA Prices Have Fallen By More Than Two-Thirds in the Past Five Years $300 32 MW (New York) • PPA prices are levelized $250 550 MW over the full term of the $200 contract, after accounting

$150 for any escalation rates 210 $100 PV (3,822 MW, 51 contracts) MW and/or time-of-delivery CPV (35 MW, 2 contracts) (“TOD”) factors $50 Mix (7 MW, 1 contract) CSP (424 MW, 3 contracts) Levelized PPA PPA Levelized $/MWh) (2012 Price • Strong downward price $0 trend since 2007 Sep-02 Sep-03 Sep-04 Sep-05 Sep-06 Sep-07 Sep-08 Sep-09 Sep-10 Sep-11 Sep-12 Sep-13 • Smaller projects (e.g., 20 PPA Execution Date $300 MW) no less competitive 32 MW (New York) $250 • CPV and CSP largely 550 MW $200 competitive at the time, but little visibility recently $150 210 • Only about 60% of the MW $100 sample is currently Operating (2,605 MW, 42 contracts) Levelized PPA PPA Levelized $/MWh) (2012 Price $50 operational (or at least Planned (1,684 MW, 15 contracts) partly operational) – $0 bottom graph Sep-02 Sep-03 Sep-04 Sep-05 Sep-06 Sep-07 Sep-08 Sep-09 Sep-10 Sep-11 Sep-12 Sep-13 PPA Execution Date 11 Levelized PPA Prices Have Fallen By Roughly $25/MWh per Year $250 2006 (7 MW, 1 contract) • 60% of sample has flat $200 annual PPA pricing (in 2009 (1,066 MW, 14 contracts) 2008 (770 MW, 3 contracts) nominal dollars), while the $150 rest escalate mostly at low 2010 (1,193 MW, 21 contracts) $100 rates intended to keep 2011 (463 MW, 7 contracts) pace with inflation – this 2012 (590 MW, 5 contracts) $50 means that average Weighted Average PPA PPA $/MWh) Weighted Average(2012 Price - 2013 (135 MW, 5 contracts) sample PPA prices Gen $0 decline over time in real 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 2032 2034 2036 2038 2040 2042 2044 2046 2048 2050 dollars (top graph) $250 • Top graph also shows the $200 steady march downward by PPA vintage $150 • Bottom graph simply $100 levelizes each line in the top graph, to provide a $50 Weighted Average PPA Price (2012 $/MWh) Weighted Average (2012 Price PPA - clearer picture of the time Gen $0 trend PPA Year: 2006 2007 2008 2009 2010 2011 2012 2013 Contracts: 1 0 3 14 21 7 5 5 Levelized MW: 7 0 770 1,066 1,193 463 590 135 12 Solar and Wind: Neck-and-Neck in the Southwest

$250 Graph only includes projects located in CA, NV, AZ, NM 550 MW • 96% of our solar PPA $200 sample (in MW) is in $150 Wind (4,666 MW, 43 contracts) CA, NV, AZ, and NM 39 MW Solar (4,125 MW, 47 contracts) $100 • These two graphs

$50 compare levelized 150 MW 50 MW PPA prices from solar Levelized PPA Price (2012 $/MWh) (2012 PPA Price Levelized $0 and wind projects in Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 these four states PPA Execution Date 180 • In 2012 and 2013, 160 Wind (in CA, NV, AZ, NM) solar has given wind a 140 Solar (in CA, NV, AZ, NM)

2012 $/MWh)2012 run for its money 120 100 • This is particularly true 80 given solar’s greater 60 Weighted Average, Weighted - time-of-delivery value Levelized PPA Price PPA Levelized 40 20 to utilities (see next 6 contracts MW840 11 contracts MW 1,004 16 contracts MW 1,099 3 contracts MW770 4 contracts MW572 6 contracts MW463 1 contract MW50 5 contracts MW590 2 contracts MW141 5 contracts MW135 3 contracts MW264 16 contracts MW 1,584

(Generation 0 slide) 2008 2009 2010 2011 2012 2013 PPA Execution Year 13 Time of Delivery (“TOD”) Factors Favor Solar Over Wind

250 WIND Super-Peak TOD Factor SOLAR (PV) 100% Super-Peak Generation (%) Shoulder TOD Factor 90% Night TOD Factor 200 Solar Post-TOD Revenue 80% 70% 150 Shoulder Generation (%) Wind Post-TOD Revenue 60% 50% $/MWh 100 40% 30% 50 20% Night Generation (%) 10% 0 0% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

• Graphs show PG&E’s 2011 TOD factors applied to a hypothetical $100/MWh base PPA price for both wind (left graph) and solar PV (right graph) • Wind & solar generation profiles are from real projects selling to PG&E under TOD prices • Over the course of a year, solar earns ~$25/MWh more post-TOD revenue than wind  The vast majority of solar’s TOD advantage over wind comes from differences in diurnal generation profiles (e.g., no night generation for solar), but seasonal profiles do help a little  This is a PG&E example, but results are similar for other investor-owned utilities in California • As solar penetration increases, causing “net peak load” to shift later into the afternoon or evening, solar’s (or at least PV’s) TOD advantage will likely diminish

14 Questions?

Report and slide deck available at: http://emp.lbl.gov/reports/re

Mark Bolinger: [email protected] Samantha Weaver: [email protected]

This research was supported by funding from the U.S. Department of Energy’s SunShot Initiative.

15