Levelized Cost of Electricity

8/3/2019 Levelized Cost of Electricity

1/27

T h e D r i v e r s o f T h e

Llzd Ct e ctcty Utlt-SclPtltcs

s u npower CorporaT on

14 August 2008


2/27

s un powe r 2

The DriverS of The LeveLizeD CoST of eLeCTriCiTy for UTiLiTy-SCaLe PhoTovoLTaiCS

eXeCuTive suMMarY

For more than three decades, utilit-scale solar generated electricit has been dismissed as too costl.

But the cost of solar generated electricit is consistentl coming down, while the cost of conventionalelectricit is increasing. Advances in solar cell technolog, conversion efcienc and sstem

installation have allowed utilit-scale photovoltaic (PV) to achieve cost structures that are competitive

with other peaking power sources. The calculation of the levelized cost of electricit (LCOE) provides

a common wa to compare the cost of energ across technologies because it takes into account

the installed sstem price and associated costs such as nancing, land, insurance, transmission,

operation and maintenance, and depreciation, among other epenses. Carbon emission costs and

solar panel efcienc can also be taken into account. The LCOE is a true apples-to-apples comparison.

Around the globe, the solar industr has installed approimatel 10 gigawatts of solar PV sstems.Pacic Gas & Electric Co. has announced more than 2 gigawatts of agreements involving both solar

thermal and PV technologies, including 800 megawatts of photovoltaic power the largest utilit-scale

contracts for PV in the world. SunPowers 250 megawatt central station, high-efcienc, PV power

plant in California Valle will be the rst to deliver utilit-scale PV power to PG&E. These solar power

plants are vivid eamples of how the electricit production landscape is changing rapidl to embrace

a much broader portfolio of renewable resources. The LCOE equation sorts through the relative costs

of such sstems and pinpoints the increasingl positive economics for harvesting the worlds most

abundant energ resource sunshine.

The economies of scale inherent in utilit-scale solar sstems are similar to those found with other

power options, but PV has the benet of being completel modular PV works at a 2 kilowatt

residential scale, at a 2 megawatt commercial scale or at a 250 megawatt utilit scale. PV has the

unique advantage among renewable resources of being able to produce power anwhere: deserts,

cities or suburbs. Smaller scale PV costs more on an LCOE basis, but it can be selectivel deploed on

the grid wherever and whenever needed to reduce distribution capacit constraints and transmission

congestion while producing pollution-free power. All PV can be constructed quickl and even utilit-

scale power plants can begin delivering power within a few quarters of contract signing a major

advantage when compared to conventional power plants. At SunPower, we serve customers across thespectrum, from small-scale to utilit-scale solar, because each application has distinctive advantages

and will contribute to driving solar power to become a major source of carbon-free power.

T LCoe eqt Ky t eltg emgg egy Tclg

The LCOE equation is an evaluation of the life-ccle energ cost and life-ccle energ production. It

allows alternative technologies to be compared when different scales of operation, investment or

operating time periods eist. For eample, the LCOE could be used to compare the cost of energ


3/27

s un powe r 3


generated b a PV power plant with that of a fossil fuel-generating unit or another renewable

technolog.1 It captures capital costs, ongoing sstem-related costs and fuel costs along with the

amount of electricit produced and converts them into a common metric: $/kWh.

wt d LCoe dct pv?

Both capital costs and operating and maintenance costs are driven b the choice of technolog and

the area of the solar sstem. We outline in this paper how the following ke factors drive the LCOE for

solar PV power plants.

1) Panel Efcienc: SunPowers high-efcienc solar panels generate up to 50 percent more power

than conventional technolog and up to four times as much power as thin lm technologies,

thereb lowering area-related costs.

2) Capacit Factor: SunPowers tracker technolog can increase energ production from solar panels

b up to 30 percent, further reducing area-related costs and contributing more high-value energ

during afternoon hours than ed-tilt sstems.

3) Reliable Sstem Performance and Lifetime: SunPowers established crstalline silicon technolog,

with its histor of consistent, predictable performance, reduces power plant nancing costs

lowering the LCOE.

Sunlight is a diffuse energ resource.

Maimizing energ production per panel area

is critical to achieve the best LCOE in a utilit-

scale PV power plant. As shown in Figure 1, if

a PV power plant with 1 terawatt hour (TWh)

of annual energ production is built with

SunPower high-efcienc PV panels mounted

on solar trackers, up to 75 percent less panel

area is required when compared with thin lm

technolog mounted in a ed tilt conguration.

This energ production densit leverages almost all PV power plant ed plant and operation and

maintenance (O&M) costs, directl reducing the sstem LCOE.

Based on the LCOE, SunPowers high-efcienc power plants generate energ at a price competitive

with other peak power resources. Given our technolog roadmap and LCOE forward cost curve, we

epect our high-efcienc silicon PV technolog to maintain that competitive position.

1 W. Short, D. Packe, T. Holt, A Manual for the Economic Evaluation of Energ E fficienc and Renewable Energ Technologies,National Renewable Energ Laborator March 1995

0.0

0.5

1.0

1.5

2.0

2.5

SunPower T20 Tracker 11% CIGS/CdTe Fixed

Tilt

6% aSi Fixed Tilt

Square M iles of PV Panels per 1 TWh

Sq. Miles of

PV

fg 1 - PV Panel Area Required for 1 TWh Annual Production


4/27

s un powe r 4


Tbl Cntnts

I. The LCOE Equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5A. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

B. Major LCOE Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

1. Initial Investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2. Depreciation Ta Benet . . . . . . . . . . . . . . . . . . . . . . 6

3. Annual Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

4. Sstem Residual Value . . . . . . . . . . . . . . . . . . . . . . . 7

5. Sstem Energ Production . . . . . . . . . . . . . . . . . . . . 7

C. The LCOE Model Sensitivit. . . . . . . . . . . . . . . . . . . . . . . 8

II. LCOE Variables for Utilit-Scale PV . . . . . . . . . . . . . . . . . . . 9

A. PV Power Plant Performance . . . . . . . . . . . . . . . . . . . . . 9

1. Sstem Capacit Factor . . . . . . . . . . . . . . . . . . . . . . . 9

2. PV Panel Performance and Lifetime . . . . . . . . . . . 12

3. Predicting Sstem Performance . . . . . . . . . . . . . . 14

B. Initial PV Power Plant Investment . . . . . . . . . . . . . . . . 15

1. PV Panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2. Area-Related Epenses. . . . . . . . . . . . . . . . . . . . . . 173. Land Use and Epense . . . . . . . . . . . . . . . . . . . . . . 19

4. Grid Interconnection Costs . . . . . . . . . . . . . . . . . . . 21

C. PV Power Plant Operating Epenses . . . . . . . . . . . . . . 21

D. Sstem Residual Value . . . . . . . . . . . . . . . . . . . . . . . . . 24

E. SunPowers LCOE Forecasting Tool . . . . . . . . . . . . . . . 24

III. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26


5/27

s un powe r 5

-

i. T LCoeequtn

The recent announcements of a wide variet of utilit-scale solar

photovoltaic (PV) power projects provide evidence that utilit-scale

PV is now reaching levels that are price competitive on a levelized

cost of electricit (LCOE) basis with other peak power sources.

Deriving a utilit-scale PV LCOE requires a range of inputs which

this paper discusses in detail. It also will review the LCOE benets

of high-efcienc silicon PV technolog for utilit-scale solar power

plants.

a. i tdctThe LCOE equation is one analtical tool that can be used to compare

alternative technologies when different scales of operation, invest-

ment or operating time periods eist. For eample, the LCOE could

be used to compare the cost of energ generated b a PV power

plant with that of a fossil fuel generating unit or another renewable

technolog.

The calculation for the LCOE is the net present value of total life

ccle costs of the project divided b the quantit of energ produced

over the sstem life.

The above LCOE equation can be disaggregated for solar generation

as follows:

W. Short, D. Packe, T. Holt, A Manual for the Economic Evaluation of Energ Efcienc andenewable Energ Technologies, National Renewable Energ Laborator March 1995


6/27

s un powe r 6

-

When evaluating the LCOE and comparing other commonl known

$/kWh benchmarks it is important to remember that the LCOE is an

evaluation of levelized life ccle energ costs. The price of energ

established under Power Purchase Agreements (PPAs) or b feed-

in-tariffs (FITs) ma differ substantiall from the LCOE of a given

PV technolog as the ma represent different contract or incentive

durations, inclusion of incentives such as ta benets or accelerated

depreciation, nancing structures, and in some cases, the value of

time of da production tariffs.

B. Mj LCoe i t

1. INITIAL INVESTMENT

The initial investment in a PV sstem is the total cost of the projectplus the cost of construction nancing. The capital cost is driven b:

a-ltd ct which scale with the phsical size of

the sstem namel the panel, mounting sstem, land, site

preparation, eld wiring and sstem protection.

Gd tcct ct which scale with the peak power

capacit of the sstem including electrical infrastructure such

as inverters, switchgear, transformers, interconnection relas

and transmission upgrades.

pjct-ltd ct such as general overhead, sales and

marketing, and site design which are generall ed for similarl

sized projects.

2. DEPRECIATION TAx BENEFIT

The depreciation ta benet is the present value of the depreciation

ta benet over the nanced life of the project asset. Public policwhich enables accelerated depreciation directl benets the sstem

LCOE because faster depreciation translates to faster recognition of

the depreciation benet.


7/27

s un powe r 7

-

3. ANNUAL COSTS

In the LCOE calculation the present value of the annual sstem

operating and maintenance costs is added to the total life ccle

cost. These costs include inverter maintenance, panel cleaning,

site monitoring, insurance, land leases, nancial reporting,

general overhead and eld repairs, among other items.

. SySTEM RESIDUAL VALUE

The present value of the end of life asset value is deducted from

the total life ccle cost in the LCOE calculation. Silicon solar panels

carr performance warranties for 25 ears and have a useful life

that is signicantl longer. Therefore if a project is nanced for a 10-

or 15-ear term the project residual value can be signicant.

5. SySTEM ENERGy PRODUCTION

The value of the electricit produced over the total life ccle of the

sstem is calculated b determining the annual production over the

ife of the production which is then discounted based on a derived

discount rate. The rst-ear energ production of the sstem is

epressed in kilowatt hours generated per rated kilowatt peak ofcapacit per ear (kWh/kWp). The kWh/kWp is a function of:

The amount of sunshine the project site receives in a ear

How the sstem is mounted and oriented (i.e. at, ed tilt,

tracking, etc.)

The spacing between PV panels as epressed in terms of

sstem ground coverage ratio (GCR)


8/27

s un powe r 8


The energ harvest of the PV panel (i.e. performance sensitivit

to high temperatures, sensitivit to low or diffuse light, etc.)

Sstem losses from soiling, transformers, inverters and wiring

inefciencies

Sstem availabilit largel driven b inverter downtime

To calculate the quantit of energ produced in future ears, a

sstem degradation rate is applied to initial sstem performance

to reect the wear of sstem components. The sstem degradation

(largel a function of PV panel tpe and manufacturing qualit)

and its predictabilit is an important factor in life ccle costs as it

determines the probable level of future cash ows.

Finall, the sstems nancing term will determine the duration ofcash ows and impact the assessment of the sstem residual value.

C. T LCoe Mdl stty

The LCOE is highl sensitive to small changes in input variables

and underpinning assumptions. For this reason, it is important to

carefull assess and validate the assumptions used for different

technologies when comparing the LCOE.

Figure 2 illustrates the models sensitivit to input assumptions. We

provide three scenarios that all start with the same PV sstem price

and predicted energ output using a tracker in a high insolation3

location. We then modif 1) the annual degradation rate, 2) the

forecasted economic life, 3) the annual O&M epense, and 4) the

discount rate. The resulting LCOE for the three scenarios range from

$0.09 / kWh to $0.23 / kWh, illustrating that for the same sstem

capital cost and initial energ output the range of energ prices can

var b a factor of two or more. Comparing LCOE calculations and

power plant energ pricing requires aligning assumptions across

eamples and calibrating against empirical data to generate a moreaccurate LCOE forecast.

3 Insolation is the level of solar radiation received at a given location


9/27

s un powe r 9


One use for LCOE calculations is to compare costs without

incentives. If incentives such as the U.S. Investment Ta Credit (ITC)

are assumed in an LCOE calculation the should be specicallreferenced to make clear the basis for comparison between

technologies.

Given the high sensitivit of the LCOE to input variables, it is

important to understand the validit of performance output over

a sstems lifetime. Silicon PV sstems have been operating

outdoors for more than 20 ears4 and therefore the performance

and degradation mechanisms are well understood. For silicon-

based PV sstems it is possible to accuratel forecast future output

allowing one to populate the LCOE equation variables with a high

level of condence.

To understand the LCOE outlook for utilit-scale PV it is important to

understand the lifetime sstem performance and cost. The following

sections summarize ke cost and performance drivers for a utilit-

scale PV power plant.

a. pv p plt pmcThe lifetime energ generated from a PV power plant is a product

of the plant location, annual performance for a given capacit,

component degradation and sstem lifetime.

fg 2

Solar PV LCOESensitivit toVariable Changes

C 1 C 2 C 3

Sstem Price 100% 100% 100%

kWh/kWp 100% 100% 100%Annual Degradation 1.0% 0.5% 0.3%

Sstem Life 15 25 40

Annual O&M $/kWh $0.030 $0.010 $0.005

Discount Rate 9% 7% 5%

LCOE $/kWh $0.23 $0.13 $0.09

ii. LCoevbls

Utlt-Scl Pv

4 E. Dunlop, D. Halton, H. Ossenbrink, 20 years of Life and More: Where is the End of Life ofa PV Module? IEEE Proceedings 2005, p.1595


10/27

s un powe r 10

-

CapacityFactor(AC)

Sstem Price $ / Wp (DC)

20%

22%

24%

26%

28%

30%

32%

34%

36%

38%

40%

$2.00 $2.50 $3.00 $3.50 $4.00

ssmlml rrg g eeqlt LCqlt LCoeoe vv ll

1. SySTEM CAPACITy FACTOR

The capacit factor is a ke driver of a solar projects economics.

With the majorit of the epense of a PV power plant being ed,

capital cost LCOE is strongl correlated to the power plants

tilization. The annual capacit factor for a PV power plant is

calculated as:

A PV power plants capacit factor is a function of the insolation at

the project location, the performance of the PV panel (primaril as it

relates to high temperature performance), the orientation of the PVpanel to the sun, sstem electrical efciencies and the availabilit of

the power plant to produce power.

7

Annual kilowatt -hours generated for eac h kilowatt AC of peak ca pacity (kWh/kW p)

8760 hours in a year

fg 3

Associated Capacit Factorsand Sstem Prices Producing

an Identical LCOE5

Capacit factor is generall epressed as a function of the AC rating of a plant so the abovekWh/kWp calculation is based on the kWh per AC watt peak as opposed to the DC watt peak


11/27

s un powe r 11


The economic impact of the capacit factor is substantial. Figure 3

illustrates a range of identical LCOE values, epressed in $/kWh,

for a given PV power plant sstem price as epressed in $/Wp and

the associated capacit factor. As the capacit factor declines, the

required installed sstem price must also substantiall decline to

maintain sstem economics. For eample, a $2.50/Wp sstem with

a 24 percent capacit factor (such as with a ed tilt conguration)

delivers the same LCOE as a $3.50/Wp sstem with a 34 percent

capacit factor (such as with a tracker). The highest capacit factorsare generated with trackers which follow the sun throughout the

da to keep the panel optimall oriented towards the sun. This

tracking also has the benet of generating more energ in the

peak electricit demand periods of the afternoon. SunPower has

developed two patented tracking sstems to optimize the capacit

factor of a PV power plant: the T0 Tracker optimized for space-

constrained sites and the T20 Tracker optimized for maimum

energ production.

fg 4 - SunPws T0 Tck fg 5 - SunPws T20 Tck


12/27

s un powe r 12

-

The LCOE model assigns an equal value to electricit generated

throughout the ear. However, electricit generated at peak

periods is more valuable to the utilit. The use of tracking with a

solar sstem can increase the output of a plant after 4 p.m. in the

summer b more than 40 percent, which is often a period of peak

demand on the sstem when energ is highl valued. Figure 6 gives

a comparison for the summer energ output of a ed and tracking

PV power plant as compared with the California ISO load. A tracker

enables higher output during the peak afternoon period for a given

plant capacit.

2. PV PANEL PERFORMANCE AND LIFETIME

Successful prediction of PV panel performance over time is critical

to project investors. Furthermore, demonstrating the historical

performance of a compans panel technolog is critical to

determine nancing parameters which underpin the LCOE

calculation.

Silicon PV has the longest operating histor of an solar cell

technolog. The photograph in Figure 7 shows a monocrstalline

silicon panel after 20 ears of outdoor eposure with no major

visual degradation. Studies on the performance of silicon PV panels

show onl four percent total degradation after 23 ears of outdoor

eposure.6 This eperience provides a high level of condence in

fg 6

Comparison of California SummerLoad Requirements with Fied and

Tracking PV Sstems

Summer CA

Peak Load

!

"

#!

#"

$!

$"

%!

%"

&!

&"

!

$!

&!

'!

(!

#!!

#$!

#&!

#'!

#(!

0

0

0

0

0

0

T20 Tracker

Fixed Tilt

CAISO load

sspp ppdct . Lddct . Ld ppll

SummertimeMWh

perMWpNameplate

C

SO

oad(G

)


13/27

s un powe r 13

-

making future performance predictions. Note that most investors

nance a solar sstem based on an assumed panel degradation rate

of 0.5 to 1.0 percent per ear, a faster rate than this historical data

for silicon PV might indicate.

Research on silicon PV historical performance suggests that panel

ife ma etend much further than the 25-ear design life.7 This

demonstrates that long-term performance ma enable longer

nanceable sstem lives in the future. Figure 8 illustrates the LCOE

model sensitivit to nanced sstem life based on a seven percent

discount rate. As indicated in the gure, etending the nanced term

of the project beond todas 20- to 25-ear values could have a

material impact to the LCOE.

fg 7

Monocrstalline Silicon PV Panelafter 20 ears of Outdoor Eposure

fg 8

LCOE Sensitivit toFinanceable Sstem Life

CCoeoe vs.vs. sYsY ssTT MM LifeLife

60%

65%

70%

75%

80%

85%

90%

95%

100%

!"# !!# !$# !%# ! '"# '!# '$# '%# ' $"#

Financeable Sstem Life (years)

6 F. De Lia, S. Castello, L. Abenante, Efcienc Degradation of C-Silicon Photovoltaic Mod-ules After 22-year Continuous Field Eposure, Proc. 3rd World Conf. on PV Energ Conver-sion, Ma 2003, Osaka, Japan.

7 E. Dunlop, D. Halton, H. Ossenbrink, 20 years of Life and More: Where is the End of Life ofPV Module? IEEE Proceedings 2005, p.1595


14/27

s un powe r 14


13. PREDICTING SySTEM PERFORMANCE

In addition to calculating PV panel output, an estimate of the

sstems overall performance must be made to nance a project.

The ke variables in a PV power plants performance are plant

uptime, weather-based performance (insolation, ambient temperature,

soiling, etc.), inverter and power sstem efcienc, and sstem

component degradation (largel from the panel).

SunPower has developed an analtical model, PV Grid, which

accounts for the above variables and makes future performance

predictions based on SunPowers eperience with more than 450

installed commercial rooftop and power plant sstems. With this

tool SunPower provides project investors with a well-demonstratedmeans of estimating project cash ows.

Figure 9 illustrates the actual versus epected performance for a

10 megawatt SunPower tracking power plant sstem in German,Bavaria Solar I. During the rst three ears of operation, the sstem

performance has eceeded the performance estimates under which

the project was nanced.

fg 9

Epected and ActualEnerg Production for10MW Bavaria Solar8

B slk pmc

8 A. Kimber, Long Term Performance of 60MW of Installed Sstems in the US, Europe, andAsia, Proceedings of the 22nd Annual Photovoltaic Solar Energ Conference, September 2007


15/27

s un powe r 15


This correlation between empirical data and future predictions is

critical in reducing investor risk and the related cost and terms

of capital investments. An important path to utilit-scale LCOE

reduction is to demonstrate to investors the predictable output,

degradation and sstem life which would support a lower cost of

project capital. As more PV data is generated and investors become

more familiar with the technolog, this ma become possible.

B. itl pv p plt itmt

1. PV Panel

When discussing the potential for photovoltaic solar cost reduction

the focus is understandabl placed on the panel. Over the past

several ears, solar panel prices have represented approimatel

$4/Wp of total PV sstem installed prices of $6-$9/Wp9 dependingon the market and application tpe.

Until 2004, PV cell and panel production costs were steadil

declining following classic learning curve behavior as the solar

fg 10

Representative Eperienceof SunPower PV Power

Plant Technolog

9 Within the PV industr sstem prices and sizes are often referred to in terms of the DC Wp ofthe sstem such as here. In other instances AC Wp prices and sizes are published. AC Wp pricesare higher than DC values because of the losses in transforming power from DC to AC i.e. a1 megawatt DC sstem at $7.00/Wp might be rated as 0.8 megawatts AC and $8.75/AC Wp.


16/27

s un powe r 16


industr grew. In 2004 and through 2008 however, the rapid growth

in PV demand led to a global shortage of solar-grade polsilicon,

the ke raw material used in conventional silicon solar cells. The

spot market price of polsilicon during this period rose from $25/

kg to greater than $500/kg for some reported transactions. The

cost of polsilicon became the driving cost of a conventional solar

panel, increasing production costs to articiall high levels relative

to the historic learning curve. As a secondar effect, solar cell

manufacturing costs also suffered as the result of underutilized,

silicon-constrained factories.

In 2007, some solar manufacturers entered into new intermediate

and long-term contracts that will continue through the rest of thedecade, lowering feedstock costs for those that have contracted

for that silicon. The polsilicon industr should also benet from

an improved cost structure as compared with pre-shortage

levels due to the scale economies of the new factories being built

and new silicon purication process technolog. In the rst half of

2008, SunPower saw its rst material silicon cost reductions as we

beneted from the deliver of substantial volumes of polsilicon

under suppl contracts from new production facilities.

One benet of the silicon shortage was that the cost and scarcit

of silicon prompted a signicant improvement in silicon utilization

b solar cell manufacturers. In SunPowers case, the grams of

polsilicon consumed to manufacture a watt at the solar cell level

declined from 13 g/W in 2004 to 6.3 g/W in 2008 and is planned to

decline to an estimated 5 g/W with SunPowers Gen 3 technolog

now under development. B 2011 this approimatel 60 percent

reduction in the use of silicon, coupled with an approimatel 50

percent decline in the price of polsilicon, will independentl drive

large cost reductions for PV panels.


17/27

s un powe r 17


Cell and panel conversion costs are driven b ield, depreciation,

labor, chemical consumption, electricit cost and materials.

Conversion costs can be improved b shorter and more efcient

processes, higher throughput production lines, larger plant sizes

driving scale economies, and greater automation, among other

factors. All of these costs are also leveraged b the efcienc

of the solar cell. SunPowers cost structure and cell efcienc

advantage demonstrate that higher efcienc cells can absorb the

increased manufacturing costs to make each cell due to the higher

watts per cell. Efcienc advantages continue downstream into

panel assembl, sales, marketing and installation. For eample,

holding all other costs constant, an increase in cell efcienc of

one percentage point will equate to approimatel a ve percentdecrease in installed sstem costs. Figure 11 illustrates the solar

conversion of efcienc of SunPowers solar cells relative to

conventional silicon and thin lm PV technologies.

2. AREA-RELATED ExPENSESPV power plant area-related epenses include sstem costs which

directl scale with the area of PV panels used. These ependitures

are the dominant non-panel costs in a PV sstem and include steel,

foundations, mounting hardware, plant installation, shipping and

warehousing, eld wiring, and the electrical components used to

connect the panels. Area-related costs are highl correlated with the

prices of steel, copper and concrete as well as transportation epenses.

fg 11

Relative Solar CellConversion Efciencies

6

8

10

12

14

16

18

20

22

Solar Cell

Efficiency (%)

S u n P o w e r

A - 3 0 0

HITConventionalThin Films Ribbon

= Average produ ction efficiency

S u n P o w e r

G e n 2

rlt sl Cll C ecc


18/27

s un powe r 18


The structural materials necessar for panel installation are driven

b the wind load requirements of the project. These are a function

of the PV panel surface area that is eposed to the wind whether the

sstem is tracking or ed (similar to how the wind force on a sail is

a function of the sail size). As a result, simplied tracking and ed

tilt congurations share similar cost structures with the eception of

the drive and control components.

There is a common misconception that trackers signicantl add

to the cost of a sstem over ed tilt congurations. SunPower

has developed trackers which can move up to 300kWp of panels

with a simple half-horsepower motor which requires little

maintenance. SunPower has determined that the nancial benetof the increased energ production generated b tracking the sun

signicantl outweighs the incremental sstem costs. B the end

of 2008, SunPower and its partners will have deploed more than

250 megawatts of tracking sstems on three continents. With this

eperience, SunPower has determined that tracking sstems have

delivered superior LCOE economics for its customers than ed

congurations.

Area-related installation costs can var substantiall b site and b

countr. For eample, a fence post-like support foundation might

be easil driven into the ground in Bavaria whereas a South Korea

tphoon zone ma require a thick steel beam placed in a hole

drilled into rock and secured with reinforced concrete at four times

the cost. As a result the range of foundation costs for a ed tilt

sstem or single-access tracker could var from $30 - $200 / m2 of

PV depending on the site. Additionall, differences in government

electrical codes can signicantl impact costs; one jurisdiction ma

require epensive steel wire conduit while others allow the direct

burial of cable into the ground.

Once the area-related costs for a sstem are calculated, a simple

transformation to $/Wp can be accomplished b dividing $/m2 b

the Wp/m2 of the panel. In the case of SunPowers high-efcienc

panels, area-related $/Wp costs are approimatel 50 percent

lower than thin lm PV panels. Figure 12 below demonstrates how

area related costs are leveraged through efcienc for a sample


19/27

s un powe r 19


central station PV solar power plant with 1 TWh of annual energ

production. Note that although the material costs are higher for

standard efcienc and thin lm panels, the are largel similar to

what the would be with a ed tilt sstem so tracking still makes

economic sense provided there is available land.

3. LAND USE AND ExPENSE

Land used for solar power plants has been readil available and

inepensive in the past, largel because the land had little economic

value other than in some cases low-ielding agricultural activities.

As solar power plant developers began acquiring land in South

Korea, Southern Europe and the southwest U.S., prices for prime

land conducive to a solar power plant rapidl increased in cost and

general land availabilit became an issue. Korea and Southern

Europe have seen solar-suitable land price increases of more than

300 percent and southwest desert land has sold for prices as high

as a reported $23,000 per acre for at land10 with high insolation

located near electrical transmission lines, a roughl 15,000 percentincrease over historical values for the same parcels.

There are two fundamental drivers for the land consumed b a solar

power plant: solar panel efcienc and sstem ground coverage

ratio (GCR). Sstem GCR is the ratio of solar panel area to land area.

fg 12

Area Related CostComponents for a T20

Tracker Power Plantwith 1 TWh of Annual

Production

PV Panel Used for Project SunPower Standard Efcienc Thin Film

PV Panel Efcienc 20.0% 14.0% 11.0%

Sstem Size (DC kWp) 423,191 439,754 430,108

Land Required (Sq Miles) 5.11 7.58 9.44

Truckloads of Concrete Required 12,718 18,218 22,678

Steel Required (Tons) 35,083 50,258 62,561

Cabling Required (Miles) 410 587 730

Trenching Required (Miles) 32 45 57

Modules to Wash (Sq Feet) 14,579,689 21,643,300 26,941,782

Tracker Installation Labor (hours) 365,450 523,516 651,678

Concrete Required (Tons) 254,353 364,367 453,568

a rltd Ct 1 Tw T20 Tck pjct

10 T. Wood, The Southwest Deserts Real Estate Boom, Fortune Magazine, Jul 11, 2008.


20/27

s un powe r 20


PV panels mounted at use land the most efcientl and have the

maimum GCR but have the lowest capacit factor, meaning lower

utilization of ed plant costs. Conversel a two-ais tracker has

the maimum possible capacit factor but requires up to 10 times

more land than at congurations. To put it simpl, the better the

orientation to the sun (thus capacit factor), the longer the shadows

created and therefore the further apart panels must be placed to

avoid panel to panel shading.

To deliver the best utilit-scale PV LCOE one must balance land use

with the sstem capacit factor. SunPower addresses this optimi-

zation problem b manufacturing the worlds highest efcienc PV

panels along with tracking sstems that efcientl use land whileincreasing energ production. SunPowers tracker offerings include

the T20 Tracker, which maimizes capacit factor in an efcient land

footprint, and the T0 Tracker, which optimizes land use for

constrained sites while still providing a high capacit factor.

Figure 13 illustrates the land consumption versus capacit factor for

a power plant producing 1 TWh / ear in a high insolation location.

One can see in this eample that:

With high-efcienc PV panels, up to 75 percent less land is

required for a given capacit factor conguration.

With high-efcienc PV panels mounted on trackers, up to

30 percent higher capacit factors are attainable while using

a similar or lower amount of land per quantit of energ

produced than low and medium efcienc panels mounted on

ed tilt sstems. This means that lower LCOE congurations

are achievable without prohibitivel increasing the amount of

land required.


21/27

s un powe r 21

-

. GRID INTERCONNECTION COSTS

Grid interconnection costs relate to the inverter, transformer,

switchgear, medium voltage substation and electrical interconnect,

the high-current electrical backbone bringing power in from the

arra and ultimatel the transmission back to the central grid in

the case of a power plant requiring a transmission upgrade for gridintegration. These costs are driven b the price of the manufactured

components, the skilled labor used to install and the price of copper,

which drives much of the inverter and electrical wiring costs. Power

transmission costs are driven down through scale economies, more

intelligent sstem design and through improved plant utilization

such as with solar tracking.

C. pv plt otg ex

The operation and maintenance (O&M) of a PV power plant is

relativel straightforward because there are few moving parts and

no cooling sstems. O&M costs generall scale with three factors 1)

sstem peak power dominated b inverter maintenance, 2) sstem

annual energ production densit, and (3) general site related items.

Improving the capacit factor of a sstem directl reduces O&M

fg 13Land use and CapacitFactors11 for 1 TWh

Production Congurations

!"!!

!#!!

!$!!

!%!!

!&!!

!'(!!

!'#!!

!'$!!

!'%!!

!'&!!

'

'

'

' ' '

' ' ' ' ' ' ' ' ' ''

' ' ' ' '

' ''

#()!*+,-!

'$)!.*/!

'')!012*30456!57!

%)!8*/!57!

LdLd rrqmt qmt hhgg iiltlt pvpv pp ppltlt

t 1 Tt 1 Tww aall ppdctdct

34% T20 Tracker 32% T0 Tracker 26% Fied Tilt

a C C a p a C i T Y f a C To r

SquareMilesRequired

11 Note the listed capacit factors are based on the AC rating of the power plant at the point ofgrid interconnection, the DC nameplate capacit of the PV power plant will be approimatel 20percent higher than the AC rating depending on the PV panel tpe and sstem conguration.


22/27

s un powe r 22

-

costs through higher utilization rates of ed assets. Figure 14

demonstrates this as it relates to the inverter requirements to

generate 1TWh of annual energ in a PV power plant. In this

eample, 1 TWh of energ would require 335 inverters, each 1 MWp,

with a SunPower T20 Tracker versus 442 inverters with a ed tilt

sstem at the same location. The use of a tracking sstem would

therefore signicantl reduce the inverter O&M cost.

Signicant power related maintenance costs also eist with respect

to transformers, switch gear and grid interconnection, and all

benet from a high capacit factor sstem conguration.

Module cleaning, panel repair or replacement, mounting structure

and wiring maintenance, and vegetation control all scale with

the annual energ production densit of the panels. The annual

energ production densit is a criticall important factor for sstem

economics (both O&M and overall LCOE). The annual energ

production densit is the kWh generation per unit area per ear as

measured as:

The impact of the annual energ production densit can be

substantial. Figure 15 shows the area of PV panels required in a high

insolation solar power plant to generate 1 TWh of annual output.

fg 14

Inverters Required for 1 TWhof Energ Production in the

Southwest U.S. Desert

Capacit Factor 34.1% 25.8%

1 MWp Inverters per annual TWh1 MWp Inverters per annual TWh 335335 442442

Inverter O&M Cost 100% 132%

T20 TckT20 Tck ffxd Tltxd Tlt


23/27

s un powe r 23

-

O&M costs which correlate with the area of PV panels used can thus

be reduced using high-efcienc PV panels mounted on tracking

sstems. A simple eample of these O&M savings is with the cost

of cleaning panels. With a high annual energ production densit

panel, washing costs can be reduced b up to 75 percent. This allows

for either a direct reduction of O&M costs or allows for panels to

be washed more frequentl and economicall, increasing sstem

annual energ production. Although often overlooked, washing and

soiling can have a material impact to a PV power plant LCOE.

In a tracking sstem there is the added cost of motor and controller

maintenance. But in SunPowers eperience this cost is relativel

small when compared to the other O&M cost savings the tracker

provides. For eample, the SunPower motor requires onl annual

ubrication and a single motor can control more than 300kWp

of PV. Also, the tracker bearings require no lubrication and are

designed for more than 25 ears of use. The O&M cost of a utilit-

scale tracking sstem would be less than $0.001/kWh over a ed

conguration, which does not include the O&M savings from theincrease in energ production.

Looking to the future, opportunities for O&M cost reduction

include improved inverter reliabilit, scale economies from larger

plant sizes, automated washing and water reccling tools, and

sophisticated remote monitoring.

fg 15

PV Panels Required for 1TWh of Annual Production

!"!#

!"$#

%"!#

%"$#

&"!#

&"$#

SunPower T20 Tracker 11% CIGS/CdTe Fixed Tilt 6% aSi Fixed Tilt

' ' ' ' ' ' ' 'ssq Ml q Ml pvpv pp l 1 Tl 1 Tww


24/27

s un powe r 24


D. sytm rdl vl

Related to the previous section, solar PV nancial models generall

assign zero residual value to the project. The sstem however,

could have a useful life of 50 ears or more ielding a material

residual value to the sstem after the 20- or 25-ear nanced term.

Additionall, the PV power plant could increase in value if fossil-fuel

based energ prices continue to rise.

Due to the time value of mone, the LCOE impact of a sstems

residual value is diluted but could still materiall reduce a PV power

plants LCOE.

It is conceivable that in the future PV sstems will be treated asassets with an active secondar market. In the wind industr,

secondar turbine sale and refurbishment has begun to occur.12

SunPower has seen some value being placed on the future

reclamation of the structural steel used in its power plants, but

placing a value the residual energ of a PV power plant is still

immature in the market.

e. sp LCoe fctg Tl

SunPower has set a compan goal of reducing the LCOE of its

installed sstem cost b at least 50 percent b 2012 based on 2006

costs. Through its vertical integration, SunPower has a unique

window into the detailed costs of a solar sstem from quartz

mining for metallurgical silicon to the construction and maintenance

of a PV power plant.

To plan and track LCOE reductions b market and application around

the world, SunPower has developed a Web-based database that

aggregates hundreds of cost, performance and nancial inputs from

its projects. The project dovetails with SunPowers research anddevelopment work funded b the U.S. Department of Energs Solar

America Initiative (SAI). The SAI sets forth aggressive solar LCOE

reductions through technological and process innovation.13

12 J. Runon, Finding a Second Life For Retired Wind Turbines,hp://www.renewableenergyworld.com, Jul 1, 2008

13www.eere.energy.gov/solar/solar_america/pdfs/solar_market_evoluon.pd f


25/27

s un powe r 25


The LCOE for an incremental PV power plant to be built in the future

is inuenced b a variet of eternal factors including echange

rates, labor prices in respective manufacturing and construction

locations, scarcit of critical raw materials, the cost of capital, land

prices, and man other factors. These risks are minimized b the

use of a high-efcienc solar panel technolog like SunPowers

since the efcienc leverages almost all non-PV plant costs. Once

built, the LCOE of energ coming from a silicon PV power plant is

ver predictable since the LCOE is heavil inuenced b capital cost,

location and sstems technolog choice.

Based on etensive LCOE scenario analsis with a range of cost

and performance structures for incumbent and emerging solar

technologies, SunPower believes that utilit-scale, central station

solar power plants built with high-efcienc silicon PV will deliver a

competitive LCOE now and in the future.

fg 16

SunPowers LCOEForecasting Tool


26/27

s un powe r 26

-

We conclude that on the man dimensions of cost and performance

that underpin the LCOE for a solar power plant, high-efcienc

tracking PV offers a ver compelling solution. To review, the LCOE is

the net present value of total life ccle costs of the project divided b

the quantit of energ produced over the sstem life.

Ke LCOE benets for high-efcienc PV power plants include:

Lt Ttl L Cycl Ct

High-efcienc panels minimize power plant capital coststhrough the reduction in the number of modules and scale of

the mounting sstem and land required to generate a given

amount of energ.

Higher conversion efciencies, more efcient use of silicon

and larger scale manufacturing operations will drive continued

high-efcienc panel cost reductions.

Life ccle O&M costs are substantiall lower for high-efcienc

tracking PV due to up to four times the energ production per

panel per ear.

A higher sstem residual value for a silicon PV plant drives

total life ccle cost reduction.

hgt Ttl Ltm egy dct

Through optimized solar tracking, SunPower PV power plants

maimize the annual energ production of a sstem leading to

high capacit factors and a lower LCOE.

With a more than 20-ear operating histor, monocrstalline

PV modules provide predictable energ production which

reduces investor investment risk and enables longernanceable sstem lives.

LCOE analsis shows how SunPowers high efcienc silicon PV

power plants generate electricit at a price competitive with other

peak power resources. Based on comparison between published

cost predictions for other technologies and our internal cost

reduction roadmap and resultant LCOE forward cost curve, we

epect to maintain this competitive position into the future.

iii. Cnclusns


27/27


For more information about this paper, please contact:

Bb okk

Senior Director, Investor Relations

SunPower Corp.

(408) 240-5447

[email protected]

Lead Author

Mtt Cmbll

Supporting Teampt acb

Jl Bld

ed sml

st wgt

Documents

Levelized Cost of Electricity