We are pleased to announce a new study conducted jointly by Berkeley Lab and the National Renewable Energy Laboratory (NREL), entitled Estimating the Value of Improved Distributed Photovoltaic Adoption Forecasts for Utility Resource Planning. A free webinar will also be held on May 24 at 10 am Pacific / 1 pm Eastern.
The report aims to assist utility resource planners in evaluating potential improvements to techniques for forecasting adoption of distributed photovoltaics (DPV). To do so, the analysis simulates future capital and operating costs for the Western Interconnection under varying assumptions about the accuracy of DPV forecasts used to develop generation-expansion plans. By comparing utility costs across scenarios, the analysis estimates the cost of misforecasting DPV adoption and the potential cost savings from reducing forecast error.
Based on this analysis, several key findings emerge:
The utility-cost impacts of misforecasting DPV adoption can be non-trivial. The cost-impacts of misforecasting DPV adoption depend on the degree of forecast error and on the actual level of DPV growth that occurs, as illustrated in Figure 1. Within our base-case analysis, systematically misforecasting DPV adoption over multiple successive planning cycles increases utility costs by up to $7 million per terawatt-hour (TWh) of electricity sales (the upper left-hand corner of Figure 1). For a large utility with 10 TWh/year of sales, this would translate to a $70 million present-value cost. Naturally, the impacts are smaller in cases with less DPV growth or a smaller degree of misforecasting, but may still be significant enough to warrant investigating improved DPV adoption forecasting methods.
The cost of misforecasting can be asymmetrical. For the particular system modeled in this analysis, the cost impacts of misforecasting DPV adoption are more severe when adoption is underforecasted than when it is overforecasted, as indicated by comparing the left-hand and right-hand sides of the figure above. In general, underforecasting DPV tends to increase capital costs but decrease operating costs (relative to a perfect forecast), while overforecasting DPV does the opposite. Accordingly, the magnitude and direction of any asymmetry depends on the relative degree of sensitivity of capital and operating costs to DPV forecast error. This phenomenon can have practical implications for utility resource
The cost of misforecasting is sensitive to market and planning conditions. For example, as modeled in this analysis, the Western Interconnection is oversupplied with capacity in the initial years of the planning period, which tends to dampen the impact of overforecasting DPV adoption. Other utility systems with lower reserve margins could see greater cost impacts from overforecasting DPV adoption. Our analysis also highlights the importance of renewable energy credit (REC) prices. In a sensitivity case with roughly a $20/MWh increase in REC prices, the cost of severely overforecasting DPV in one adoption scenario rises from $1 million per TWh of retail sales in the base case to roughly $8 million. This is due to the additional cost of having to purchase RECs to cover RPS compliance shortfalls. Conversely, the cost of severely underforecasting DPV falls from $7 million per TWh to $2 million, due to the additional revenues from the sale of surplus RECs.
Recognizing that some level of forecast uncertainty is inevitable, a utility interested in evaluating the potential benefits from improving its DPV adoption forecasting methods must compare the expected costs of DPV misforecasting under its current approach to the expected costs under an improved approach. Using the modeled costs of DPV misforecasting presented above, one can apply a relatively simple probabilistic method to estimate the cost savings from reducing DPV forecast uncertainty by a specified amount. As one example, Figure 2 estimates that a utility planning for DPV growth equal to 3.5% of total generation over a 15-year period could expect present-value savings of $0.4 million per TWh of retail electricity sales by reducing its DPV forecast uncertainty from roughly +75%/-55% (line A) to ±25% (line B). Thus, for a large utility with 10 TWh per year of sales, this would amount to a present value savings of $4 million.
For further details on these findings and others, please refer to the report and accompanying slide deck briefing, which can be downloaded here. In addition, a webinar summarizing key findings will be held on May 24 at 10 am Pacific / 1 pm Eastern. Register for the free webinar here.
We appreciate the funding support of the U.S. Department of Energy’s Solar Energy Technologies Office for making this work possible.