The solar toolbox: How utilities can find the best planning approach for distributed solar
Studying the locational benefits of distributed solar is one way to plan, an LBNL study says
There are big differences in how utilities are planning for the rise of distributed solar and a lot can be learned from a close look at how they plan to integrate it.
That’s the takeaway from a new Lawrence Berkeley National Laboratory study, which analyzed 30 integrated resource plans from cooperatives, municipal- and investor-owned utilities and the divergent approaches to integrating distributed solar as penetrations rise in certain service territories.
“Utilities are starting to plan for customers adding distributed solar and for procuring it themselves,” Andrew Mills, lead author of the study, told Utility Dive. “We wanted to know how they are doing it.”
The report also assesses planning by five system operators, along with distribution system analysis in California, New York, Hawaii, and Massachusetts. While the primary planning has zeroed in on customer adoption, planners are already beginning to think about how distributed solar can serve system needs, Mills added.
A prime example of this is New York’s Reforming the Energy Vision, where the Public Service Commission is tackling locational marginal pricing to see where distributed resources are most valuable to the state’s grid.
“The innovative practices tend to be more comprehensive and account for multiple factors that drive distributed PV planning,” Mills said.
How planners tackle distributed solar
Developing a forecast for deployment is key, since it guides planning in other areas. When accurately assessed, it can predict how distributed solar can meet generation needs, variability impact, and value and cost to the transmission and distribution system, according to Mills.
The study describes four categories of deployment forecasting. Most planning uses essentially static or linear models that are based on the assumption that customers will continue to add DPV either to meet mandates or at an assumed rate or an historical rate. Such planning relies “on few or no quantifiable predictive factors,” according to the paper.
But other types, specifically customer-adoption planning, recognizes “end user decision making” hinging on photovoltaic economics and resource potential, among other factors, according to the researchers.
By using factors independent of mandates or historical trends, planners have the “ability to generate new, self-consistent DPV adoption forecasts.” That does not, however, completely eliminate uncertainty.
That type of uncertainty is best addressed through IRPs, with the analysis finding more than half of them use multiple scenarios to address possible least-cost resource mixes.
A third of the studies looked at only one forecast, while a few plans evaluated the impact of net load changes on their forecasts but failed to consider expected variations in DPV capacity.
Using multiple scenarios is currently the most comprehensive approach because “it accounts for changes in both net load and the generation portfolio,” the researchers reported. But it can be made more robust to uncertainty by including an “acquisition path analysis.”
The approach, as used by PacifiCorp and the Hawaiian Electric Companies (HECO) adds “trigger events” that could necessitate a change to an alternative planning scenario. But less than half of the forecasts mulled the possibility of proactively deploying distributed solar as a system resource. Even if they did, planners usually dismissed it because of the “higher cost and lower capacity factor relative to utility-scale PV (UPV),” the researchers reported. Only Pacific Gas and Electric (PG&E) seemed to tackle distributed solar factors, including its “avoided losses, transmission deferrals, and distribution-system cost impact,” the report noted. But two major proceedings, the California Distribution Resources Plans and the REV are including these factors in their process.
Failing to consider a wide range of factors, including utility-scale and distributed solar, could result in “a less reliable planning decision,” Mills warned.
“Planners should consider both the up-front capital costs and the costs that are avoided by investment in DPV," he added. "Only PG&E did that."
The crystal ball of future customer distributed solar adoption
The uncertainty of customer demand makes planning for future DPV deployment complicated. Suddenly, utilities’ generation portfolios are no longer entirely in their control.
“The rate of adoption depends on many factors, some of which are changing rapidly, including the upfront cost of DPV systems, availability and level of incentives, and retail rate designs or net-energy-metering [NEM] policies that affect the bill savings,” the study reported.
Customer-adoption planning is the most innovative approach, the study noted, including “historical DPV deployment, location-specific DPV technical potential, various DPV economic considerations, and end-user behaviors,” the researchers reported.
Out of the four drivers in customer-adoption, analyzed in the study, distributed solar economics is the most common, including a comprehensive assessment of technology cost and performance, federal and state incentives, innovative business models, electricity prices, and rate design factors.
Public policy is the second category of drivers. State mandates and environmental policies are among the most common considerations in this category. HECO used “the overall level of public policy support for clean energy as one of the two dimensions (the other being the price of oil) to create four broad scenarios in its IRP,” the researchers report.
A third driver is customer preference, which is one of the most volatile, the study reports. And the last category factors in the indirect impact of macro factors, such as load growth, oil and gas prices and economic growth.
PacifiCorp, PG&E and Puget Sound Energy are three utilities that incorporate customer-adoption planning. This type of planning is particularly suited to how planners approached the uncertainty of the Investment Tax Credit before it was granted an 11th hour extension last year, Mills noted.
“If planning is based only on projections from past numbers, there is no place to enter the potential impacts from the ITC expiring or not expiring,” he said. “With customer-adoption planning, planners can create scenarios that show what adoption would be with and without the ITC.”
Improving statistical techniques can allow for more factors to be considered, Mills added. Other methods include a better comprehension of customer decisions by using modern marketing techniques to evaluate customer behavior and technology valuation as well as understand third-party ownership, loans and community solar appeal.
A close-up of distributed solar’s locational value
Once planners ascertained the amount of looming distributed solar, they will need to understand how its value permeates most of the rest of planning. In turn, it becomes necessary to figure out where the resource will be on the system because its location increasingly determines its benefits and impacts.
Most IRPs focus on system-level needs and therefore have not been concerned with the location of distributed solar. But as planners begin to think about its role in utility portfolios, concerns about over where distributed solar can be placed to provide the most system benefit are likely to become relevant.
Location was discussed in some resource planning studies “to distinguish different DPV options or distinguish DPV from other resources,” the researchers report. It is also very much a part of estimating avoided losses, avoided transmission and distribution (T&D) costs, the DPV generation profile, and the capacity credit.
Independent system operators (ISOs) need locational data for future planning “down to the granularity of dispatch or load zones” and where distributed solar is on “specific feeders is important for distribution planners,” the researchers report.
Planners are largely using three methods. 80% of the studies project future distributed solar locations based on where current load, population, or resource is. The third is a “propensity-to-adopt” approach that is more innovative, adding predictive factors like demographics and customer load, the report noted.
Distributed solar adoption has not followed maps of load or population, instead aligned with demographic factors like home ownership and cluster by neighborhood, Mills said. “Simple assumptions that DPV will grow in proportion to load or customer population could lead to less informed decisions than would come from a more comprehensive approach.”
Because of wide variation in current and anticipated distributed solar penetrations, the need to understand its impact on future T&D investments is very different across the planning studies, the researchers report. The most innovative planning in use estimates future feeder hosting capacity.
In a study from from Dominion and Navigant, representing 14 feeders, evaluates their hosting capacity as well as doing cost-benefit calculations for advanced inverters and energy storage.
These findings make it clear that advanced technologies can allow higher penetration of distributed solar at a lower cost, but with one caveat, Mills added. “It requires that the utility invest in system-wide telecommunications and distribution management systems that are not included in the upgrade cost estimates.”
This makes the specific numbers less useful to planners but they “will allow for more informed planning,” Mills said
A twist in high-end distributed solar projections
What surprised Mills was the significant difference between distributed solar adoption forecasts for 2020 and 2030 made by recognized expert analysts and those made by utilities and system planners.
The forecasts from Bloomberg New Energy Finance, GTM Research-Solar Energy Industries Association, and the National Renewable Energy Laboratory is above the high end of planning forecasts "about two thirds of the time," the researchers report.
While different methodologies are part of the disparity, utilities are more likely to be more conservative “because the risks associated with forecasting too little DPV (which can lead to extra costs from overbuilding the system) are less acute than the risks of forecasting too much (which can lead to reliability issues if insufficient resources are available),” the researchers reported.
“There is no way to say one is more accurate than the other," Mills said. "But there is enough credibility in the third-party forecasts to suggest the utilities should at least benchmark their highest forecasts, though not necessarily their base forecasts, against them.”