NREL applied the Distributed Generation Market Demand (dGen) model to the following projects, demonstrating the breadth of analysis that is possible when using it.
National Policy Analysis
NREL's dSolar model was used to explore the sensitivity of national distributed generation photovoltaics (DGPV) deployment to three uncertain market factors—technology cost, future net metering policy, and a hypothetical carbon fee. The modeling results show that a high carbon fee ($25/ton with 5% annual escalation, representing either a direct tax or an effective fee through other mechanisms) increased the average annually installed capacity of DGPV by approximately 30% over the reference scenario (which did not have a fee on carbon or assume implementation of the Clean Power Plan). As modeled, the fee spurs de-carbonization in the bulk power system, which dampens the impact of the fee on DGPV deployment. The $25/ton carbon fee had roughly the same impact on deployment as a 10-year extension of net metering policies (beyond when those caps were projected to expire). Because of the long-term focus of the model, the study also found that whether projects installed in 2010s are rebuilt at the end of their lifetime will substantially impact annual market sales starting in 2030.
To learn more, read Distributed PV Adoption—Sensitivity to Market Factors.
State or Utility-level Projections
NREL staff used the dSolar model in 2015 to advise the Maine State Legislature on state policy by developing a range of forecasted distributed solar deployment by 2021. The NREL team used a scenario-based method that first identified the primary factors impacting distributed solar demand in Maine technology costs, retail electricity rates, load growth, and extension of the Federal Investment Tax Credit. They varied these factors to develop high, medium, and low PV forecasts. The Maine legislature used NREL's analysis to develop a five-year PV procurement strategy.
To learn more, read Maine’s Overview of Distributed Generation Legislation.
All technologies modeled within the dGen framework leverage a database of highly resolved geospatial information. Because each agent is assigned an actual location, each potential agent can be associated with a number of attributes from data sets in the underlying geospatial database using a simple spatial overlay. The spatial layers include but are not limited to energy consumption profiles, census demographics, zoning and building counts, and renewable resource assessment. This process of spatial overlay results in a nearly complete characterization of the attributes at each 200m by 200m potential customer location, where the attributes assigned to an agent can be used to develop an adoption likelihood score. Finally, the spatial nature of the model and agent attribution means the model can easily be trained on new sources of data as well as be applied at multiple spatial resolutions—national, state, and even distribution feeder level.