This IREC Insight blog is focused on a relatively uncharted frontier in energy planning: distributed energy resource forecasting (aka DER forecasting). We will examine what a DER forecast is, why it matters, and what states, regulators and utilities should consider as they navigate future grid planning efforts.

by Erica McConnell and Allison Johnson, attorneys for IREC through Shute, Mihaly & Weinberger, LLP

As states, communities and diverse stakeholder groups across the country develop long-term plans for solar photovoltaics (PV), they aim to identify pathways for strong market growth as well as any remaining barriers and potential pitfalls that require more proactive attention.

For example, the Vermont Comprehensive Energy Plan sets forth a statewide goal to achieve 90% renewable energy by 2050, and the Vermont Solar Pathways Project undertook comprehensive technical analysis and scenario modeling to determine the physical and financial investments, grid upgrades, efficiency, and flexible load needed to progress Vermont toward that 2050 goal. Similarly, the Wasatch Solar Team in Utah, led by Salt Lake City and Utah Clean Energy, developed a 10-year solar deployment plan for the state to provide a clear path to overcome new challenges associated with growing solar penetrations and guide the development of a sustainable and robust solar market across Utah. Both of these efforts were supported by the Solar Market Pathways initiative, part of the U.S. Department of Energy Solar Energy Technologies Office.

Beyond solar, states and local governments are adopting energy storage procurement requirements, clean peak standards, regional electric vehicle (EV) infrastructure plans, and energy efficiency portfolio standards. In addition, more consumers and businesses are adopting plans of their own to deploy distributed energy resources (DERs), such as solar PV, energy storage, EVs, and energy efficiency, to achieve personal economic, environmental and resiliency goals.

The future growth of DERs on the electric grid does not have historical precedent. Therefore, critical to the adoption (and ultimate success) of any new plans to drive the integration of more DERs, is that utilities and states make adjustments to how they plan for and invest in their electric systems over the long-term.

Most utilities across the country undertake planning efforts of their own on a regular basis, to identify generation and transmission and distribution system infrastructure needs to meet anticipated growth in electricity demand. Sometimes done through Integrated Resource Plans (IRPs), or more recently through Distribution Resource Plans (DRPs), these utility-led planning efforts (which are typically overseen by state regulatory commissions) underpin their requirement to pursue cost-effective, least-risk and reliable electric systems for their customers.

In the new paradigm of state, community and consumer-driven clean energy plans and investments, utility planning efforts must be revised –  to ensure evolution towards a future grid that is notably different from the grid of the past. To avoid an overbuilt (and costly) system and enable more streamlined integration of customer-centric DERs, utility planning should proactively take into account current DER growth trends and develop more sophisticated and granular DER forecasting methodologies. Such efforts should not just consider anticipated customer-led DER growth, but also broader solar and other DER deployment plans, as well as related state and local energy plans, such as state energy efficiency plans and energy storage mandates, among others.

What is DER Forecasting?
A key component to future grid planning efforts is being able to accurately predict when, where and in what quantities DERs will be deployed on the grid. The term “DER forecasting” refers to the process (and underlying methods and assumptions) of making predictions about adoption of DERs on the distribution grid, specifically to inform system planning. Ideally, accurate DER forecasts will help utilities and stakeholders answer three related questions:

  • When will DER growth occur over time?
  • Where on the grid will that growth occur?
  • How will these new DERs operate?

Like traditional load forecasting, DER forecasting requires utilities to use the best information about what has happened in the past and what may happen to develop a picture of what is likely to happen in the future. But DER forecasting diverges from traditional load forecasting when it comes to the inputs; the historical data used for traditional load forecasting is simply not available or necessarily accurate for most DERs.

DER forecasts form the basis for a wide variety of grid planning and market activities and are central to ensuring that updates and modifications to the grid are done with DERs in mind. For example, California Public Utilities Commission (CPUC) staff have proposed an evolving vision for a new distribution planning process built on DER forecasts and growth scenarios. A simplified version of this vision is captured in the graphic below (presentation available in full, here).

When combined with hosting capacity analysis, another key tool in the modern grid toolbox, DER forecasts can help inform when and where the grid may face capacity constraints. In turn, this analysis may be used to determine where future DER growth should be directed, and what grid changes and investments may be needed to accommodate expected growth. Such efforts feed into related analyses to identify optimal grid locations for DERs, and develop locational valuation tools and mechanisms, such as solicitations and pricing structures, to drive DER to those locations, as illustrated in the below graphic.

DER Forecasting is a key step to integrated distribution planning (Optimizing the Grid: A Regulator’s Guide to Hosting Capacity Analyses, IREC December 2017)

Key Considerations for DER Forecasting
While the concept and vision for incorporating DER forecasting into planning and policy activities are relatively straightforward, the details are not. Lessons and insights are emerging that will help improve DER forecasting as a tool for building the modern grid.

Modeling Distinct Impacts of DERs
The distinct performance characteristics and related consumer behaviors associated with particular DERs pose a unique challenge for DER forecasting. DERs may contribute to or lessen the load on the grid (or both), and may otherwise impact its operation. For example, distributed solar PV generates energy that may offset customer load, similar to adoption of energy efficiency measures, and may also help to decrease system-wide peak demand, depending on when and where the generation occurs.

When distributed solar PV is paired with distributed storage, its effect on the grid may be modified, shifting system charging or discharging in time. Smart inverters may be engaged to modify the impacts of distributed solar PV on load by providing voltage regulation and other grid stability services. In addition, the localized effects of DERs depend on where they are installed, and when and how they operate. For example, EVs typically draw energy from the grid—they increase load—but their impacts vary based on when, where and what time they recharge.

Many utilities have incorporated projections for at least some DERs, such as distributed solar PV, into their annual forecasting processes as “load modifiers” – factors that may cause a peak load forecast to deviate from the expected trajectory. Now, in response to state policies, and at the urging of commissions and stakeholders, a few utilities across the country are diving into the next generation of DER forecasting for use in planning and beyond.

Developing DER Forecasts: Sources of Information
DER forecasters may look to a wide variety of information sources, including:

  • Models for predicting adoption of new technologies (e.g., S-curves), which may be applied to adoption of new DERs.
  • Economic and demographic information about customers, which may correlate with their adoption of DERs.
  • Weather data and locational factors (e.g., roof slopes and shading), which may affect generation potential and the attractiveness of certain DERs to customers, affecting where they are installed.
  • Technical studies, which may help elucidate likely applications and locations for DERs.
  • Local government plans and laws (e.g., zoning), which may encourage or limit DERs in certain locations.
  • Permitting information that may indicate the number of certain DERs, such as distributed solar PV, approved and under consideration by local governments.
  • Interconnection queues that may indicate the number of certain DERs, such as distributed solar PV and storage, approved and under consideration by utilities.
  • DMV registration data for EVs.
  • Regional, state or local electrification plans.
  • Adoption rates of state, utility or local incentives for DERs.
  • State and local policy changes, both adopted and anticipated, such as changes to Renewables Portfolio Standards.
  • Economic projections, which may indicate the general market strength and appetite for DERs.
  • Information from DER providers, such as planned marketing campaigns or expected growth.
  • DER industry market forecasts, developed and published by reputable third-party consultancies.

Not all sources of information are appropriate or available for forecasting growth of every type of DER. For example, while market forecasting tools and recent historical data are relatively reliable for well-developed DERs like solar PV and energy efficiency, these tools tell us little about newer DERs, like distributed storage and EVs.

The DER industry and affiliates (including consultancies, non-profit organizations, and other groups evaluating industry trends) may have a better sense of realistic growth trajectories for these newer technologies. Some data are only available for certain types of projects, such as the number of interconnection applications or permitting information.

Forecasting the impacts of policy changes can pose a complex challenge, particularly when new policies have not been in place for very long or even finalized by the time the forecast is generated. Ultimately, for all types of data, the level of confidence in the data and how robust it is are both important, as forecasts are only as reliable as the underlying data. However, excluding the impacts of such policies from DER forecasts is equivalent to predicting they will have zero impact on the growth of DERs—a highly unlikely outcome.

Assumptions & Growth Scenarios
As with any modeling effort, the assumptions behind DER forecasts strongly influence the results. Unfortunately, in many cases, the assumptions that should be made for DER forecasting purposes (and how they should be made) remain highly uncertain. An example from an investor-owned utility in New York, Orange and Rockland, shows that forecasts can vary considerably depending on assumptions made regarding DER interconnection applications. The graphic below compares the actual load curve from a peak summer day in 2016 with six possible load forecasts resulting from changes in PV forecasts. The forecasts vary based on (1) the percentage of applications in the interconnection queue projected to actually materialize, and (2) the projected impacts of changes to New York’s queue management rules. Shifts in just these two variables generate significantly different forecasts, both in the actual peak load and the timing of that peak.

New York Stakeholder Engagement Session on Long-Term Load and DER Forecasting
(July 14, 2017) Orange and Rockland Case Study

As states and utilities are working to find ways to deal with uncertainty in the context of DER forecasting, developing DER growth scenarios is emerging as a critical starting point for discussions. California, for example, has directed utilities to develop three 10-year growth scenarios that capture a wide range of potential growth: a most likely scenario that projects autonomous growth based on today’s market realities; a high growth scenario that incorporates additional information about anticipated developments in the energy marketplace; and a very high growth scenario that projects the DER growth necessary to meet state energy and emissions goals.

To date, only the most likely scenario has been explored in any detail; it will serve as the basis for this year’s planning process. However, each of these growth scenarios could eventually play a role in planning, helping utilities and stakeholders understand and plan accordingly for the scope and likelihood of potential future outcomes.

Probabilistic Forecasts
Relying on a single growth scenario may give a false sense of certainty in a forecast. In contrast, probabilistic forecasts, which can indicate the probability that a particular scenario may occur, offer a more nuanced picture of the DER growth likely to materialize and can help manage the uncertainty of DER forecasting efforts. However, the practicability and the implications of probabilistic DER forecasting are still under debate and some argue it would present major analytical challenges if applied to more traditional utility planning approaches.

In addition, some utilities have expressed concern that acknowledging uncertainties in their forecasts will open the door for challenges to their spending proposals. For example, utilities traditionally must identify investments as “necessary” in order to be compensated for them. If the need for an investment is assigned a probability depending on the likelihood of DER growth, utilities have expressed concerns as to the impact on their ability to justify that investment to ratepayer advocates and commissions. Employing probabilistic forecasting then, may also require changes in how all stakeholders, including regulators, think about utility planning, investments and cost recovery.

Nonetheless, many stakeholders maintain that probabilistic forecasting would facilitate safe, modular grid updates, particularly where DER (or load) growth is most uncertain. For example, a single forecast might indicate that a utility should make a large capital investment – such as building a new substation – in order to accommodate DER growth within a certain time frame (say four years), while a probabilistic forecast might show that a smaller DER project in two years will allow the utility to delay the larger investment for an additional three years. In that interim period, forecasts may change, and the larger investment may not materialize at all. In other words, probabilistic forecasting could facilitate more efficient planning and the incorporation of more DERs as non-wires alternatives.

Granularity & Accuracy
DER forecasts can be developed from two opposite vantage points. System-level “top-down” forecasts begin with predictions for cumulative DER growth across the distribution system and then may be disaggregated to predict where that growth will occur. In contrast, local data-driven “bottom-up” forecasts begin by looking at where particular DER growth is expected and then compile that information into a bigger picture prediction. Each approach has significant implications for the certainty, accuracy and granularity of forecasts, and utilities may rely on either or both, depending in part on the information available to them and how they plan to use the forecasts.

Local data can provide accurate and precise information about grid needs and growth at particular locations, especially in the short-term. Utilities can use location-specific information, such as interconnection applications, permits, demographics and zoning, to predict, with reasonable certainty, what is likely to occur at locations on the grid in the next few years. However, particularly for larger utilities, it may be impractical to extrapolate local data to develop forecasts for every location in the system. In addition, while forecasts for particular locations may be fairly accurate, errors may compound when localized forecasts are aggregated into a system-level forecast. For example, slight overestimates in growth for locations throughout the system may be negligible at those locations. However, they may add up to a system-level forecast that predicts significantly more growth than is likely to materialize.

Top-down system-level forecasts manage the cumulative uncertainty problem presented by bottom-up approaches, particularly for longer-term growth. While it may be difficult to predict whether load growth will occur in one neighborhood versus another, system-level forecasts can more accurately predict that a certain amount of growth will occur system-wide within a certain period of time, based on market projections, policy incentives, and technology costs. For DER-specific forecasts, performance characteristics or consumer behaviors may make it difficult to predict the impact of DERs at different locations, and system-level forecasts can mitigate these challenges. For example, customers with EVs may charge in multiple locations, so forecasting their impacts at particular locations is prone to high levels of uncertainty; system-level forecasts for EV growth can spread this uncertainty across the grid to minimize overall error.

At the same time, system-level forecasts are inherently less granular. In order to use these top-down forecasts to predict local grid needs, system forecasts must go through a disaggregation process that assigns predicted growth to particular locations, based on local data (similar to the data used to generate bottom-up forecasts) and/or models or assumptions. Depending on how a system-level forecast is disaggregated, the end result may or may not fully account for true “on the ground” conditions.

California is currently moving toward a predominantly top-down approach to DER forecasting, while New York’s forecasts continue to be calculated primarily from the bottom-up. However, each state uses pieces of both approaches to inform their DER forecasting efforts.

CPUC staff has proposed to link DER forecasts to system-level forecasts from the California Energy Commission’s (CEC) biennial Integrated Energy Policy Report (IEPR). The utilities will disaggregate these system-level forecasts to develop preliminary local DER forecasts. Then they will adjust those forecasts based on local data, such as interconnection queue information, and other models and assumptions, such as customer adoption patterns. This approach will ensure that distribution planning aligns with state transmission planning and other planning processes that also rely on the IEPR forecasts.

New York utilities, on the other hand, develop their short-term distribution forecasts based on local data; longer-term forecasts often rely on system-level factors, such as general economic growth predictions. New York has not proposed a process for generating separate DER forecasts yet, but some utilities have begun to incorporate DERs into their general forecasts as load modifiers.

The granularity of DER forecasts will be important as DER planning processes evolve. Grid investments (and potential sourcing of DERs to defer those investments) will depend on identifying particular times and places where the grid will face constraints. [bctt tweet=”Developing reliable DER forecasting methodologies may be resource intensive up front, but the resulting forecasts will support a suite of grid benefits in the long-term.” username=”IRECUSA”]

Pioneering the Future of DER Forecasting
Mitigating and managing uncertainty in DER forecasting requires more practice and real-world applications, and pioneering new approaches are illuminating what works, and what doesn’t. California and New York have demonstrated that DER forecasting already plays a role in utilities’ annual planning processes. However, in both states there are still untapped opportunities to ensuring such forecasting efforts are as transparent as possible and developed in ways that allow the utilities to use them to better integrate DERs into their systems. Confidence in DER growth scenarios and underlying assumptions remain integral to both states’ efforts.

For states that have not yet considered the role of DER forecasting in their utility planning processes, requiring utilities to develop DER forecasts is clearly an important first step that will help identify data gaps and barriers to reliable forecasting. Requiring utilities to reconcile their forecasts with actual growth on an annual basis can help inform the strengths and weaknesses of the forecasts and underlying methodologies.

As more states and utilities evolve their plans to reflect the fact that consumer- and community-driven energy resources are a growing part of our energy picture, DER forecasting will continue to improve and be a cornerstone of future grid planning.