Solar PV Forecasting Models and Solar PV Forecasting Service


Continuing PV cost reductions, growing popularity of power purchase agreements, tax incentives and other factors are responsible for year-over-year doubling, or more, of PV installations in many states. Recent Austin Energy and NV Energy utility-scale procurements at less than 4 cents/kWh portent a rapidly arriving transformation for residential utility customer installations with significantly increasing solar PV penetrations in nearly all utility service areas. These PV impacts present both utility challenges and benefits that are quantified with the SGRC Solar PV Forecasting Models and Forecasting Service.

While installations are still a relatively small percentage of the total utility customer base, they tend to be geographically highly concentrated with potentially large impacts in certain ZIP areas, substations and along certain individual feeders. Heavy saturations of PV along feeders can create power quality issues, asset life degradation due to excessive regulator and capacitor bank switching, and other problems. Some locations in Hawaii, California, New Jersey, and Arizona are reportedly already facing these challenges. With the substantial increase in solar installations expected in coming years, these issues will become more problematic for many utilities.

For utilities who are promoting solar PV installations, forecasting and program development/analysis presents a special set of challenges reflecting the variety of household, dwelling unit, neighborhood and other variables that impact the decision to purchase solar systems and/or solar PPA services. Tax incentive, utility incentives, declining costs and new solar-based end-user products complicate the forecasting process.

The SGRC PV Forecasting Model and PV Forecasting Service

Recognizing the need to forecast and analyze solar PV utility impacts, the SGRC undertook development of Solar PV Forecasting Models designed to assist utilty distribution planning and to provide a program evaluation tool for utility solar program developers.

The SGRC Solar PV Forecasting Models have been developed by merging the Consortium’s Smart Grid Investment Model (SGIM) financial analysis resources and Jackson Associates (JA) MAISY Utility Customer Databases and Agent-Based Energy and Hourly Load Forecasting Models. The SGIM has been applied for 20 electric utilities since its development in 2010 while MAISY modeling and forecasting analysis have been applied at more than 100 energy-related organizations including utilities, equipment manufacturers, state and federal regulatory agencies and other energy-related organizations. MAISY agent-based modeling has been applied in utility smart grid and other technology impact analysis as well as for forecasting future energy and hourly load impacts for utilities and public utility commissions. The agent-based model structure is perfect for addressing small-geographic area solar residential PV installations and load impacts.

MAISY agent-based models have been applied extensively to analyze and forecast new technology markets, product development other issues for solar and distributed energy technologies companies including Geostellar , Sun Edison, Sungevity, Sharp, Toyota, Ingersol Rand, United Technologies, Bloom Energy, Ice Energy, Aisen and many more.

The agent-based solar PV statistical models are estimated with data on more than 7 million utility customers and nearly 500,000 solar PV installations. Models are applied at feeder/substation and ZIP level and reflect the “clustered” nature of PV installations resulting from geographic patterns of household, dwelling unit, neighborhood and other characteristics that drive PV sales.

The dramatic decline in solar PV prices is driving an industry decentralization transformation that will soon impact every utility at its most publicly vulnerable point: reliable delivery of service to its customers.

The SGRC Solar PV Forecasting Models are available for implementation at individual utilities as a planning and analysis tool while the PV Forecasting service provides comprehensive analysis and reporting for utilities based on a SGRC implementation of the model for each client utility.

Summary of Solar PV Models and Forecasting Service Characteristics.

MAISY Agent-Based Models

Agent-based modeling is used extensively to model and forecast consumer, firm, and government agency decisions where individual agents (households, business and government decision-makers) differ substantially in characteristics that impact a particular decision of interest. In the case of residential solar PV purchases, MAISY agent-based Solar PV Models apply information on more than 7 million utility customers and nearly 500,000 solar PV installations to develop a comprehensive characterization of the diverse set of household, dwelling unit, neighborhood and other factors associated with PV purchases.

The application of MAISY Utility Customer Databases and the MAISY agent based modeling structure in the SGRC Solar PV Models provides a well-vetted agent-based modeling approach that has been extensively applied to evaluate solar and distributed energy technology markets and purchasing decisions for companies like Geostellar, Sun Edison, Sungevity, Sharp, Toyota, Ingersol Rand, United Technologies, Bloom Energy, Ice Energy, Aisen and many more. (Click here for a partial list of MAISY Clients).

MAISY agent-based models have also been applied to forecast utility energy and hourly loads, ESCO and REP market analysis, utility smart grid technology implementations and other applications. Click here to see a general discussion of the MAISY agent-based modeling structure.

The following documents illustrate several MAISY agent-based modeling applications.

  1. Indiana State Utility Forecasting Group (SUFG) application of MAISY agent-based commercial (CEDMS) and residential (REDMS) forecasting models for the Indiana Public Service Commision. Five separate investor-owned utility models are applied for each sector.

    From the report: "Like the residential sector end-use model REDMS, Jerry Jackson and Associates actively supports CEDMS, and it continues to define the state-of-the-art in commercial sector end-use forecasting models." Click here to view SUFG's report documenting analysis and evaluation of the models (see chapters 5 and 6).

  2. This Energy Policy paper illustrates the application of JA agent-based end-use forecasting to evaluate energy efficiency and smart grid program impacts.
    Click here to access this paper.

  3. This Energy Policy paper assesses the impacts of standby rates on the market penetration of combined heat and power systems.
    Click here to access this paper

For more information on Solar PV Forecasting Models and the Solar PV Forecasting Service, contact:

Jerry Jackson, Ph.D.
Leader and Research Director, Smart Grid Research Consortium
37 N. Orange Avenue, Suite 500
Orlando, Fl 32801
979-204-7821 (cell)