Is Overproduction Costing You? A review of the impacts of solar resource data on financing and revenues for PV plants Presented by: Marie Schnitzer Vice President of Consulting Services Paul Thienpont Meteorologist
Denver Colorado Albany New York (HQ) Barcelona Spain Warsaw Poland (*) Istanbul Turkey (*) Bangalore India Buenos Aires Argentina (*) Rio de Janeiro Brazil Company Snapshot Renewable energy professionals providing technical consulting, software, data, and advisory services for the wind and solar energy markets Established in 1983; over 30 years of renewable energy industry experience Assessed over 100,000 MW Project roles in over 80 countries Offices around the globe and *with partnerships in Argentina, Poland and Turkey
Key Messages Obtaining an accurate, reliable solar resource assessment Overproduction from unreliable data sources can result in lost revenue Overproduction can result in lost returns
Financial Impacts Power Purchase Agreements (PPA) Over/Under production penalties PPA bid process Financial Models Tax Equity Financing Debt to Equity ratios Debt Service Coverage Ratio (DSCR) Cost of Capital and Rate of Return
Investor Interests Equity Investors Long-Term Productions Estimates Tax Equity Investors Interests are greatest in the beginning of the project life-cycle Debt Lenders Interested only in the minimum ability to pay back loan P50 Estimates P50 Estimates P90/99 Estimates
How do you quantify these risks? And how the variables impact your bottom line.
Study Approach Solar Resource GHI, DNI, DHI POA Energy Revenue Financial Model
Resource Assessment Methods Irradiance Data NASA SSE National Solar Radiation Database (NSRDB) -TMY 3 Satellite Derived Datasets Publicly available Commercially available (for purchase) Site Specific Typical Meteorological Year AWS Truepower Analysis Onsite measurements Satellite derived datasets Measured reference networks Conversion to Plane of Array (POA) Measured Reference Networks USCRN NREL On-Site Measurements
Study Approach Solar Resource Energy Losses Uncertainty Revenue Financial Model
Energy Modeling Satellite Reference Station Site Specific (Advanced) TMY Onsite Measurement Plant Specifications Energy Simulation Model Environmental & System Losses Uncertainty from all inputs Annual Energy Prediction P50, P90, P99
Sources of Energy Uncertainty Annual Degradation (0.5 1 %) Transposition To Plane of Array (0.5 2%) Energy Simulation, Plant Losses (3 5 %) Solar Resource Uncertainty (5 17%)
Energy Estimates and Probability of Exceedance Probability of exceedance: the level of confidence that a plant s actual energy production will be at least a certain value P90 Based on project uncertainty High Quality Resource Data P90 Reduced uncertainty Increased value of P90(99)
Study Approach Solar Resource Energy Revenue PPA Over/Under Financial Model
PPA Pricing Time of Day Pricing Base Price based on Locational Marginal Pricing (LMP) Time of Day Pricing (TOD) multipliers are used to adjust base price based on time of day and seasonal adjustments Production Thresholds PPA s typically include maximum annual production restrictions Restrictions are commonly applied at 110% of the estimated annual P50 plant production Net metering caps
Study Approach Solar Resource Energy Revenue Financial Model CapEx OpEx Debt Sizing
Financial Model Capital Expenses: EPC (full wrap) Permitting Developer fees Land lease Other Production Estimates: P50, P90, P99 Energy estimates PPA pricing Operational Expenses: Operating costs General administration Project Finance Structures Debt finance Debt Service Coverage Ratio Tax equity or cash equity finance Internal Rate of Return
Quantifying financial risk Case study results for three project locations
Case Studies: Methodology Three sites collocated with United States Climate Reference Network (USCRN) measured reference stations were selected for evaluation based on geographic location and consistency of the observations. Three resource methods were examined for each location: Actual resource data (measured) Satellite Derived TMY NSRDB TMY3 Uncertainty and Probability of Exceedance (P-tables) were assessed for each method Energy was simulated in PVsyst using the AWS Truepower standard approach for each location
Case Study: Methodology Plant Design: 12.5 MW DC 10 MW AC DC/AC Ratio: 1.25 Generic 300 W polycrystalline module Generic 500 kw inverter Row tilt optimized for each location using PVsyst Modeled without near shading Plant loss assumptions were applied consistently for each project location and resource analysis
Solar and Met Resource Variation Difference From Actual (%) 20% 15% 10% 5% 0% -5% -10% -15% -20% Merced, CA Satellite TMY3 GHI DNI DHI POA Temp Difference From Actual (%) 20% 15% 10% 5% 0% -5% -10% -15% -20% Millbrook, NY Satellite TMY3 Difference From Actual (%) 30% 25% 20% 15% 10% 5% 0% -5% -10% -15% -20% -25% -30% Tucson, AZ Satellite TMY3 GHI DNI DHI POA Temp GHI DNI DHI POA Temp
Energy Variation 10% 8% 6% Difference From Actual (%) 4% 2% 0% -2% -4% -6% -8% -10% Gross Energy Long-Term Net Energy Merced, CA Satellite Merced, CA TMY3 Tucson, AZ Satellite Tucson, AZ TMY3 Millbrook, NY Satellite Millbrook, NY TMY3
Case Study: PPA Revenue Assumptions PPA Pricing Assumptions: Base price: $66/MWh TOD and seasonal multipliers PPA overproduction: Assumed 75% decreased payment structure beyond 110% of the guaranteed energy production level (P50 Production) Energy > Annual P50 Production = 0.75 x (Base Price x TOD Multiplier) Financial model used LMP data to calculate baseline PPA pricing and TOD based off of the Day Ahead Market (DAM) prices
Case Study: PPA Revenue 4.0% Financed With Satellite TMY Financed with NSRDB TMY3 Difference in Revenue From Actual(%) 3.0% 2.0% 1.0% 0.0% -1.0% -2.0% -3.0% -4.0% TOD Revenue Lost Revenue to Cap TOD revenue is based of long-term energy 8760 results for each resource file utilizing the TOD and seasonal multipliers to the baseline PPA price Lost revenue is based off of using 110% of the estimated long-term net energy P50
Case Study: Debt Finance Structure Model Assumptions Financial Model Assumptions: CAPEX: $2.2/W AC installed capacity OPEX: $25/kW AC /year installed capacity Debt sizing: Debt ratio sized using a DSCR of 1.0 and P99 production estimates Financial Models: Three independent models were created: 1. Actual resource data 2. Satellite derived TMY 3. NSRDB TMY3 Analysis: Satellite and TMY3 model financing assumptions were stressed using the actual resource data
Case Study: Debt Finance Structure Key Metrics Debt Ratio: The amount of acquired debt Projected IRR: The estimated IRR for the cash equity investor when financing with each resource dataset Actual IRR: The realized IRR for the cash equity investor when running the actual resource data file through the financial model for the Satellite TMY and NSRDB TMY3
Case Study Results: Debt Financing Merced, CA 1.0% Financed With Satellite TMY Financed With NSRDB TMY3 Difference From Ground Corrected (%) 0.0% -1.0% -2.0% -3.0% -4.0% -5.0% -6.0% -7.0% Debt Ratio Projected IRR Actual IRR The project should have financed using 3% more debt The forecasted IRR was 2% lower than optimal results As a result the actual IRR was 0.5% lower than optimal results
Case Study: Tax Equity Financial Model Assumptions Financial Model Assumptions: CAPEX: $2.2/W AC installed capacity OPEX: $25/kW AC /year installed capacity Tax Equity Investment: Tax equity investment sized assuming an 8.0% IRR Financial Models: Three independent models were created: 1. Actual resource data 2. Satellite derived TMY 3. NSRDB TMY3 Analysis: Satellite and TMY3 model financing assumptions were stressed using the actual resource data.
Case Study: Tax Equity Finance Structure Key Metrics Tax Equity Contribution: The difference in tax equity contribution Projected IRR: The estimated IRR for the cash equity investor when financing with each resource dataset Actual IRR: The realized IRR for the cash equity investor when running the actual resource data file through the financial model for the Satellite TMY and NSRDB TMY3
Case Study Results: Tax Equity Financing Merced, CA 1.0% 0.5% Financed With Satellite TMY Financed With NSRDB TMY3 Difference From Optimal (%) 0.0% -0.5% -1.0% -1.5% -2.0% -2.5% -3.0% Tax Equity Contribution Projected IRR Actual IRR The project should have financed using 0.3% more Tax Equity Contribution The forecasted IRR was 0.9% lower than optimal results As a result the actual IRR was 0.5% lower than optimal results
Conclusions Site Specific Resource yields more accurate P50 analysis and improved uncertainty bounds Many sources of unreliable data resulting in higher uncertainty Overproduction results in lost revenue Overproduction results in lost returns for both debt, tax equity and cash equity finance
Thank You +1 518-213-0044 contactus@awstruepower.com awstruepower.com Marie Schnitzer Vice President of Consulting Services Paul Thienpont Meteorologist