LTPP Solar PV Performance and Cost Estimates Ryan Pletka, PE June 18, 2010

- 2 PV Performance Estimates

Black & Veatch - 3 Definitions DC Capacity Sum of module nameplate rating. Modeled as 20 percent higher than inverter nameplate rating. AC Capacity – Sum of inverter nameplate rating. AC Capacity DC Capacity Inverter M AC Output

Black & Veatch - 4 Definitions Continued AC Annual Energy Output The total energy output on an annual basis at the energy meter. This number includes dc and ac wiring losses. AC Capacity Factor = AC Annual Energy Output [kWh] / (AC Capacity [kW] * 8760 [hours per year]) DC Capacity Factor = AC Annual Energy Output [kWh] / (DC Capacity [kWp] * 8760 [hours per year])

Black & Veatch - 5 Overview of Performance Estimate Methodology System definition Chose 4 representative locations and 4 system configurations Developed high-level conceptual design Gathered solar resource data (TMY – Typical Meteorological Year) Modeled system performance in PVSYST Similar methodology as that used by B&V’s Solar Performance Group for more than 100 operating and proposed PV projects, most in California

Black & Veatch - 6 System Definitions Size and Locations 0.5 – 2 MW Rooftop 0.5 – 2 MW Ground 2 – 5 MW Ground 5 – 20 MW Ground North Coast Central Valley South Coast Desert

Black & Veatch - 7 Conceptual Design Basis System TypeRooftopSingle Axis TrackingFixed Tilt Size1 MW 5 MW, 20 MW Module Tilt15 degrees East-West (+/- 45 degrees) 25 degrees Module Technology Polycrystalline Cadmium Telluride Inverter CEC Efficiency 95 percent

Black & Veatch - 8 Modeling Locations & Resource Data Defined regions that team felt had significantly different PV deployment potential and performance characteristics Identified representative locations in each region for which to pull solar data NREL’s TMY3 data chosen to be representative of each location TMY3 data of Class II or higher chosen Central Valley North Coast South Coast Desert

Black & Veatch - 9 Modeling and Loss Assumptions System Voltage: 600 Vdc Module Quality: According to typical warranties Row Spacing: Assumed to be sufficient enough for no inter-row shading between 9 am and 3 pm all year. Thermal Parameters: According to mounting (roof mount, ground mount) Soiling: Weather conditions in each region assessed to develop soiling loss assumptions. AC wiring loss: 0.5 percent DC wiring loss: 1.5 percent Transformer loss: 1 percent Availability loss: 1 percent, that is, system assumed to be available (operating) 99 percent of the time

Black & Veatch - 10 Performance Estimates TypeLocationDC Capacity Factor 1 MW Rooftop Daggett18.3% Fresno16.3% Oakland15.6% Riverside16.9% 1 MW Tracking Daggett23.5% Fresno20.5% Oakland19.0% Riverside20.9% 5 MW, 20 MW Fixed Tilt Daggett21.3% Fresno18.8% Oakland17.5% Riverside19.5%

- 11 PV Cost Estimates

Black & Veatch - 12 Cost Development Approach Detailed bottoms-up estimate based on B&V’s engineer- procure-construct (EPC) methodology Market Pricing Costs for projects where Black & Veatch is owner’s engineer or lender’s engineer Financial statements for module suppliers (e.g., First Solar) Recent project announcements/public/private data (other market information) Hundreds of recent PPA bids All capital costs are on a $/Wdc basis and assume 2010 commercial operation date

Black & Veatch - 13 Cost Uncertainty/Variability Estimates are inherently uncertain and costs in the PV industry are “dynamic” The costs of CSI projects and other market sources appear to vary by approximately +/- 25 percent This variation is due to site specific factors, owner specific factors, and component/system pricing Site specific factors are usually captured in the BOS and might include rooftop mounting issues, grading or foundation issues, shading considerations, interconnection issues or similar Owner specific factors are captured in the Owner’s costs and might include permitting issues, legal issues, land acquisition, or owner’s management issues Component/system pricing variation is related to market supply vs. demand, market efficiency, timing, etc.

Black & Veatch - 14 Black & Veatch Cost Estimates for PV Systems 1 MW rooftop$5.00/watt dc 1 MW tracking$4.75/watt dc 5 MW ground$3.90/watt dc 20 MW ground$3.70/watt dc Based on configurations identified earlier As stated previously, typical cost uncertainty is +/- 25%

Black & Veatch - 15 Comparison of PV Costs, Plus Large Central Station Costs

Black & Veatch - 16 Example Breakdown of Component Costs 20 MW5 MW Module 1.65 $/watt1.65 $/watt Support structure 0.82 $/watt0.84 $/watt Inverter 0.24 $/watt0.25 $/watt BOS0.31 $/watt0.40 $/watt Owner’s Costs 0.68 $/watt0.76 $/watt Total Capital Req. 3.70$/watt3.90 $/watt

- 17 Appendix Material

Black & Veatch - 18 Details on Component Costs

Black & Veatch - 19 Module cost (5 MW and 20 MW) Black & Veatch has seen estimates between 1.3 and 2.25$/watt From Review of First Solar Financial Statements we derive 1.63$/watt From April 2010 SEIA report, SEIA indicates average 2009 costs at 1.85 to 2.25 $/watt. We have chosen 1.65 $/watt, which is near the First Solar data point and slightly below the midpoint (1.77) of the Black & Veatch low observation of 1.3$/watt and high SEIA point of 2.25$/watt

Black & Veatch - 20 Structure Cost (5 MW and 20 MW) We have chosen $/watt which is the result of reviewing quotes from three manufacturers.

Black & Veatch - 21 Inverter Costs (5 MW and 20 MW) We have chosen $/watt which is based on our experience, and verified with multiple vendors

Black & Veatch - 22 BOS Cost (5 MW and 20 MW) BOS includes commodities like trenching, wire, electrical connections and grounding rods, step-up transformer, and similar items We have chosen $/watt after completing a bottoms up estimate.

Black & Veatch - 23 Owner’s Costs (5 MW and 20 MW) Owner’s costs include spare parts, water supplies, project development, owner’s project management, sales and other taxes, insurance, advisory fees, legal, financing (including interest during construction), land/roof acquisition (as applicable), startup and construction support. There is a high degree of variation in this item due to unique customer approaches. We have chosen $/watt after reviewing actual owner’s costs for three development projects and reviewing allocation percentages made for previous studies (up to 25 percent)

Black & Veatch - 24 Analysis of Data from California Solar Initiative Reported system cost data for systems installed under CSI Generally <1 MW Data is known to contain errors and other inconsistencies Data source: CSI PowerClerk, June 9, 2010 Ratings are reported nameplate rating (kW dc)

Black & Veatch - 25 Economies of Scale – Data from Installed Systems >100 kW Data source: CSI PowerClerk, June 9, 2010

Black & Veatch - 26 Falling System Costs – Installed Systems Only Data source: CSI PowerClerk, June 9, 2010

Black & Veatch - 27 Data from Pending Applications Indicates Potentially Lower Costs, But Should be Treated Cautiously (Systems >750 kW Only) CSI Status Data source: CSI PowerClerk, June 9, 2010

Black & Veatch - 28 Most Recent CSI Pending Applications, >300kW Data source: CSI PowerClerk, June 9, 2010

Black & Veatch - 29 Most Recent CSI Pending Applications, >300kW Reporting errors? Average $/kW = 5,050/kW Data source: CSI PowerClerk, June 9, 2010

Black & Veatch - 30 Performance Estimates for Large 150 MW Sites DC Capacity Factor Desert Thin Film: 21.3% Tracker: 23.2% Central Valley Thin Film: 18.8% Tracker: 21.3%

31 LTPP Solar PV Potential and Levelized Cost of Energy (LCOE) June 18, 2010

32 Overview of Presentation DG PV Potential estimates by size and location  Methodology  Results Levelized cost of energy estimates by size and location  LCOE Tool  Results Appendix  Potential Estimates  LCOE Input Assumptions (also available in tool on website)

33 Goals of Potential Analysis Develop PV Potential estimates  Identify ‘Easy to connect’ and ‘harder to connect’  4 size and configuration categories 0.5 – 2 MW Roof, 0.5 – 2 MW Ground, 2 – 5 MW Ground, 5 – 20 MW Ground  4 locations across California Desert, Central Valley, North Coast, South Coast

34 PV Potential Estimation Adjusted the 33% RPS Implementation Analysis potential study approach  Same underlying proprietary utility substation loadings and locations as used previously  Same large rooftop potential with satellite imagery Key changes  Added small roofs in rural areas  “Set aside” potential for current programs

35 Screening Assumptions ‘Easy’ Interconnection  Nameplate PV system is less than or equal to 30% of peak load at point of interconnection to avoid reverse flow Participation  33% of large roof owners will participate Penetration  33% of feeders accommodate ground-mounted systems up to the ‘easy’ interconnection limits  33% of RETI identified large PV sites can be interconnected with a moderate transmission interconnection cost  10% of rural ‘easy’ interconnection potential in small roofs

36 PV Potential Screening Method Peak Loading on Each Substation RETI Identified 20MW Projects Urban Location Large Roof Potential Ground Mounted ‘Easy’ Interconnect Large RooftopSmall Rooftop Ground Mounted ‘Hard’ Interconnect 33% Participation of Roofs 30% of Peak Load Screen Rural Location 30% ‘Easy’ Interconnection 90% to Ground Mounted 10% to Small Roofs 33% Penetration at Moderate Cost 33% Penetration 2/3 Remaining Potential

37 Screening Steps Raw Potential (MWs): After Screening (MWs): After Removing Existing Programs (MWs): RETI Identified Sites 27,500 Substation Load Total 39,323

38 Modeled PV Potential (MW)

39 Goals of PV LCOE Analysis Create a publicly available pro-forma tool that calculates a levelized cost (LCOE) Develop model inputs  Capital Costs and Operating Costs  Performance parameters  Financing assumptions Calculate levelized cost of solar PV Standardize the LCOE presentation

40 PV Financial Pro Forma Tool Balance complexity vs. applicability for a broad range of projects Some of the features:  Debt Service Coverage Ratio (DSCR) limit  Inverter replacement fund  Debt service reserve fund Available on E3 website for download; 

41 Example – Model Inputs System Cost & Performance Inputs:Financing Inputs: Location: Desert Technology: 5-20 MW Ground Mounted

42 Example – Cash flow Screenshot of cash flow: Location: Desert Technology: 5-20 MW Ground Mounted

43 PV Bid Pricing vs. LCOE The same $/kWh price can be presented in several different ways PV bids typically reflect the price before Time of Day (TOD) factors are applied Developers see the post-TOD value, which is the true cost of the PV system Escalators can skew costs when compared to flat levelized costs Results herein are post-TOD, flat nominal levelized

44 Comparison: PV LCOE metrics Post-TOD flat nominal levelized used to show results Year LCOE Post Time-Of- Delivery (TOD) Flat nominal levelized: $0.1678/kWh Year LCOE Post-TOD Year-1 cost with escalator: $0.1441/kWh Year LCOE Pre-TOD Flat nominal levelized*: $0.1266/kWh Year LCOE Pre-TOD Year- 1 cost with escalator* : $0.1087/kWh Note: Costs shown correspond to a project in the 5-20MW ground mounted category in the desert. *Using a TOD factor of (SCE TOD schedule using TMY3 output data from Daggett with a ground mounted 25°fixed tilt system)

45 Levelized Cost of Energy from PV

46 Results – Post-TOD Nominal LCOE (Nominal $/kWh) MW Rooftop / Fixed Tilt MW Ground / Tracker MW Ground / Fixed-Tilt 5-20 MW Ground / Fixed-Tilt 150 MW Utility- Scale / Tracker 150 MW Utility- Scale / Fixed-Tilt Mojave Desert (Daggett) $0.2483$0.1852$0.1748$0.1678$0.1482$ South Coast (Riverside) $0.2683$0.2085$0.1916$0.1840N/A Central Valley (Fresno) $0.2788$0.2127$0.1979$0.1900$0.1612$ North Coast (Oakland) $0.2904$0.2294$0.2132$0.2048N/A

47 Appendix

48 Diagram of Interconnection Points Direction of electricity flow RETI PV Projects Assumed To flow in Opposite direction

49 Distributed Solar PV 20 MW sites near non- urban 69 kV substations Smaller projects on rooftops, large commercial rooftops with 0.25 MW potential Limited by 30% peak load at a given substation 20 MW near substations Large commercial rooftops Residential rooftops Illustrative Example of Distributed Solar PV

50 Ground Mounted PV Initial criteria  near sub stations equal or less than 69 kV  agricultural or barren land  less than 5% slope Environmental screen  Black out areas  Yellow out areas Land parcel  a continuous 160 acre plot (20 MWp)  within 20 miles 69 kV substation Black out area Yellow out area More than 5% slope area Example Map for Solar PV Non-Urban Projects Urban Agricultural or barren land Solar PV plant Substation

51 RETI Results on 20 MW Sites 27,000 MW nameplate PV sites identified ~1300 sites identified Filters Applied  160 acres + for 20 MW  No sites within 2 miles of urban zones  Near substations, most are 2 to 3 miles of the distribution subs with 69kV+ high-side voltage  Land slope < 5% 20 MW on substations with high side voltage of 69kV 40 MW on substations with higher voltage than 69kV Assumed not to be Rule 21 compliant

52 Black and Veatch Rooftop Analysis GIS used to identify large roofs in CA and count available large roof area Criteria  ‘Urban’ areas with little available land  Flat roofs larger than ~1/3 acre  Assumes 65% usable space on roof  Within 3 miles of distribution substation

53 Solar Photovoltaic Rooftop Identification

54 July 31, 2009 Solar Rooftop Identification

55 Solar Rooftop Identification

56 Los Angeles Area “Rooftop Resources” Puente Hills Los Angeles Ontario Anaheim

57 East Bay Area Example Analysis automates the counting of roof space and tallies total acreage of large roof space. Also checks proximity to distribution substation (not shown due to confidentiality).

58 Avoided Capacity Cost Assumption Distribution:$34/kW-yr  Used average of EE avoided costs Subtransmission:$34/kW-yr  Used average of EE avoided costs Transmission:$0/kW-yr  Network is more difficult  Set to zero for 33% RPS analysis Issues Timeframe vs. geographic specificity – must use long time frame for avoided cost value Cost of non-Rule 21 RETI 20MW PV Installations not studied  Network transmission costs of $65/kW-year assumed for these resources See EE avoided costs, R

59 Technical Feasibility of PV Connections that are >15% & <100% of Peak Load Assumption on PV engineering feasibility Caveat These numbers are based on an educated guess not on any engineering analysis % Peak Load 50% of in area PV 30% Peak Load 50% of in area PV 100+% Peak Load RETI projects

60 PG&E Example – Bay Area Clusters of large roofs make it impossible to do every roof and be below the 30% peak load.

61 PV LCOE Input Assumptions Capital Costs Capacity Factors Financing Assumptions Operating Costs

62 Black & Veatch Cost Estimates 1 MW rooftop$5.00/watt dc 1 MW tracking$4.75/watt dc 5 MW ground$3.90/watt dc 20 MW ground$3.70/watt dc Based on configurations identified in B&V presentation As stated previously, typical cost uncertainty is +/- 25%

63 Performance Estimates TypeLocationDC Capacity Factor 1 MW Rooftop Daggett18.3% Fresno16.3% Oakland15.6% Riverside16.9% 1 MW Tracking Daggett23.5% Fresno20.5% Oakland19.0% Riverside20.9% 5 MW, 20 MW Fixed Tilt Daggett21.3% Fresno18.8% Oakland17.5% Riverside19.5%

64 Financing Assumptions The following financing assumptions are used: The model minimizes the % equity constrained to a target average DSCR of This results in ~60% equity which slightly varies by technology and location.

65 Operating Costs Operating costs for the LTPP study are broken down into O&M, insurance and inverter replacement costs: LTPP O&M Costs ($/kWdc) O&M Cost Escalator (%/yr) Inverter replacement cost ($/Wdc) Inverter replacement time (Years) Insurance Expense ($/kWdc) Insurance Escalator (%/yr) Fixed Tilt $ %$ $ % Tracker $ %$ $ % RETI (converted to $/kWdc) O&M Costs ($/kWdc) O&M Cost Escalator (%/yr) Fixed Tilt $32.00% Tracker $44.00% As a reference, operating costs in RETI are presented into a single O&M cost that includes all ongoing capital expenditures: