Meeting Summary

California Public Utilities Commission (CPUC)

Investor Owned Utilities’ (IOU) Distribution Resources Plan (DRP)[1]

Distributed Energy Resources (DER) Growth Scenarios Working Group (GSWG)

Opera Plaza Community Room – Mezzanine Level, 601 Van Ness Ave., San Francisco, CA 94102

May 3, 2017

*DRAFT*

These notes summarize the Growth Scenarios Working Group (GSWG)meeting facilitated by More Than Smart. A total of 57 individuals attended the meeting representing 27 different organizations, both in person and via conference call. For meeting audio, stakeholder list and presentations for this meetings, go to Documents from prior GSWG meetings, other working groups facilitated by More Than Smart and upcoming events are also available there, or contact Laura Wang at for additional information.

Agenda: GSWG Meeting #3 (May 3): The purpose of this meeting was to review and discuss proposed IOU methodologies for incorporating distributed generation, energy storage, and electric vehicles into their 2017 distribution plans, compare proposed approaches to prior cycles, and discuss options and issues for future cycles.

  • Introductions (Laura Wang, More Than Smart) and Background/Overview (Dina Mackin, CPUC)
  • Distributed generation (IOUs)
  • Energy storage (IOUs)
  • Electric vehicles (IOUs)
  • Next steps

Meeting Summary:

Background/Overview

On February 27, 2017, CPUC published an Assigned Commissioner’s Ruling (ACR) establishing several guiding principles and setting a schedule for the IOUs to develop DER growth scenarios for their 2017 Distribution Resource Plans.[2] This series of meetings is being held to publicly discuss, review, and eventually improve and/or better align the approaches (if needed), and will lead to a revised document prepared and submitted by the IOUs in Q3, 2017.

The April 17, 2017 GSWG meeting provided an opportunity for the IOUs to present the proposed framework, describe and compare their general assumptions and approaches to distribution planning both at the system-wide level and disaggregated to distribution circuits. The next several meetings focus on growth scenarios for specific demand-modifying activities/resources (e.g., demand response; demand-side distributed generation, storage and electric vehicles; and energy efficiency).

  • Discussion
  • CPUC Energy Division clarified that the GSWG’s mission is to explain, compare, and if needed, improve (both long- and short-term) methodologies for developing IOU distribution planning forecasts for DERs, including modeling approaches, data, and relationship(s) with forecasts in related planning processes.
  • The GSWG is not designed to determine how much DER(s) should be procured. The CPUC’s Integrated Resource Plan (IRP) proceeding is the primary venue for those decisions (although a wide number of proceedings are inter-related, including grid updates, general rate cases, etc.). Admittedly there could be a dynamic/iterative relationship between the GSWGand these other processes – information developed in GSWG may in some cases affect IRP (and other) analyses and outcomes. However, the distinction between the primary goals of these processes is an important, foundational notion for GSWG. GSWG is for discussing, comparing, coordinating, vetting, and ideally improving the growth scenarios used for distribution planning.
  • Part of the GSWG discussions is to determine whether there is adequate information flowing “upward” from the IOU distribution planning processes to the IEPR forecasts, as well as to determine how/whether/to what degree the IOUs begin with the IEPR forecasts for disaggregating for distribution planning.
  • In general, CPUC’s expectation going in to the GSWG has been that the IOU service territory-level assumptions will reconcile with IEPR forecasts. Prior discussions have shown that:
  • Differences in timing between IEPR and IOU forecasts lead to significant differences in methodologies. The IOUs update their internal forecasts often, with new data, prepared for a variety of submissions, each using the newest and/or most appropriate data. Meanwhile, the Energy Commission’s IEPR forecast is produced as a full forecast every two years, with an update of key variables in the mid-cycle. These updates for products produced in different regulatory cycles/timing accounts for some differences between IOU and Energy Commission forecasts, but there still seem to be differences between the method(s) the utilities use between IOUs.
  • The GSWG has not yet examined the degree of differences between growth forecasts/assumptions. Throughout the course of meetings thus far we’ve observed a variety of results and approaches that in various ways are similar to and/or differ between IOUs and/or IEPR forecasts, but not systematically analyzed the degree, of difference(s).
  • Understanding the regulatory cycles and which updates feed into forecasts used for those regulatory products will be instructive.However, it is still necessary to understand differences between IOUs and IEPR forecast in side-by-side comparisons.

Distributed Generation (DG) – IOUs

Each IOU presented its proposed approach for estimating DG growth, discussing both system-level forecasting and disaggregation methods. The IOUs jointly addressed challenges in determining appropriate methodology.

  • IOU presentations
  • PG&E – DG System Level Forecasts (PV)
  • For 2017-2018distribution planning, PG&E performed internal modeling to forecast Distributed Generation. DG is defined for the forecast as behind the meter* (BTM) systems less than 20 MWs in size, designed to offset customer load, and includes the following technologies:
  • Solar PV
  • Non-PV Technologies
  • Fuel Cells
  • Combustion and Heat to Power Technologies
  • Wind
  • PG&E applies a Bass Diffusion technology adoption model to forecast customer-driven “organic” DG adoption based on projected DG cost-effectiveness and other characteristics, such as the suitability of the building for DG installations (technical potential), cost effectiveness (economic potential), and, propensity indicators including owners vs. renters, credit ratings, customer exposure to advertising and/or word of mouth endorsements, etc.
  • For solar, in addition to forecasting “organic” adoption, PG&E projects BTM PV installations associated with mandates and programs, specifically:
  • Solar on new homes based on anticipated Zero Net Energy (ZNE) requirements
  • Solar associated with low-income programs
  • PG&E’s 2017 DG forecast is substantially higher than the most recent Energy Commission forecast (2016 IEPR update). PG&E’s 2017 forecast is approximately 8,000 cumulative installed MW by 2026 whereas the 2016 IEPR update is approximately 5,000.
  • Differences include statewide policy decisions made after CEC 2016 forecast (income tax extension, NEM 2.0), so CEC will presumably capture the effects of these policies in their 2017 IEPR, and may also begin including ZNE.[3]
  • IOUs have greater granularity than CEC, and use information in their interconnection requests and rates requests queues to help identify the amount, and to the extent possible, location of expected growth.
  • PG&E – DG System Level Forecasts (non-PV)
  • There are a limited number of customer sites with Non PV technology interconnections – these are “early adopters.” This makes predicting future adoption occur difficult. As of 2016, in PG&E territory there are approximately:
  • Fuel Cells ~250
  • Distributed Wind ~240
  • Turbines and Engines ~240
  • PG&E – DG Disaggregation
  • PG&E disaggregates DG service territory forecasts using “propensity scoring” and other local information through a process being developed and refined each year – the LoadSEER tool/forecasting model plays a role in organizing a wide variety of inputs and outputs.
  • For 2017 distribution planning, PG&E will start with the IEPR forecast, evaluate differences between the IEPR and PG&E corporate forecasts, and determine the best approach. Local planning knowledge and Rule 21 interconnection queue may be sufficient.
  • PG&E has 5.3 million electric customers, an average of 6,900 customers per substation and an average of 1,700 per feeder.
  • SDG&E – DG System Level Forecasts
  • SDG&E begins with the Energy Commission’s IEPR forecast. For SDG&E the IEPR forecast area is the same as SDG&E’s service territory so no conversions are necessary.
  • For the 2017-18 distribution planning cycle, SDG&E will use annual DG growth provided in 2017 IEPR (currently being formulated).
  • Evaluation on the distribution level will include Residential PV, Commercial PV and Combined Heat and Power.
  • Information from the queue for interconnections is also used for larger scale >1MW requests.
  • SDG&E – DG Disaggregation
  • SDG&E uses the Bass diffusion model similar to PG&E’s but it is disaggregated by zip code then to feeders. Small generators are allocated by customer class/type for each feeder.
  • The interconnection queues go out for about one year – but can be as long as three years. Average nameplate capacity is used to obtain kW, which is then disaggregated to feeders.
  • SCE – DG System Level Forecasts
  • SCE performs internal modeling to forecast Distributed Generation and compares methodology and forecast with the CEC’s California Energy Demand Forecast as part of the IEPR.
  • Two types of DG:
  • Solar PV - estimates of PV system economics and a Generalized Bass Diffusion modeling framework to forecast organic residential adoption. Historical trends are utilized to forecast non-residential growth. SCE’s most recent forecast has also included a consideration of new installations due to Zero Net Energy requirements. SCE uses NREL Rooftop Solar Photovoltaic Technical Potential in the United States” (2016) study for “technical potential” – 91 percent of buildings in SCE service territory meet the requirements for PV technical potential. PV growth assumptions are based Energy Commission’s 2016 IEPR update. SCE’s 2017 IEPR submission is much higher, based on new information. Presumably the 2017 IEPR forecast will also be much higher. The IOUs produce/update their internal forecasts more frequently.
  • Other (CHP, Wind, Fuel Cell etc.) - Forecasts for these resources are done by trending historical adoption. SCE has a limited amount of BTM non-PV DG growth and has seen little change in overall growth trajectory.
  • SCE – DG Disaggregation
  • For residential PV they assess status of current interconnection data, use statistical analysis to group circuits with a “propensity” to grow PV penetration, use a Bass diffusion model to forecast growth then normalize to match system level forecast. Title 24 ZNE requirements are allocated to areas projected to have residential new construction.
  • Non-residential PV is allocated based on usage by customer segment.
  • Per above, there is little non-PV DG growth in SCE service territory.
  • Joint IOU Challenges
  • Significant uncertainty driven by PV policies, projected costs, estimates of technical potential and customer demand for PV
  • The small sample sizes for emerging and less prolific technologies render it difficult to forecast adoptions
  • Uncertainty is magnified at more granular levels in the distribution asset hierarchy where fewer number of customers associated with a given asset makes it more difficult to predict which customers will invest in DG and when.
  • Discussion
  • IOU internal forecasts are updated more frequently than Energy Commission IEPR forecasts (which are produced every two years, in odd-numbered years). Even though Energy Commission now produces “IEPR forecast updates” in even-numbered years, those forecasts only update a few key inputs, e.g., economic/demographic inputs. The IOU internal forecasts include numerous updated features throughout the year. E.g., the IEPR forecast does not currently include ZNE but probably will in 2017.
  • The IOUs work closely with Energy Commission in producing forecasts via the Demand Analysis Working Group (DAWG).
  • It was remarked that seeing the example in these presentations where there is a 40 percent difference in growth (e.g., several of the 2017 IOU PV forecast vs. the 2016 IEPR update) is surprising. On one hand, Energy Commission is in the process of updating the 2017 IEPR. On the other hand, depending on the point in a regulatory cycle, certain work products could diverge significantly from the IEPR forecast – e.g., if an IOU uses its internal forecasts to produce a document such as a general rate case (GRC) for example.
  • Stakeholders had questions regarding the influence of NEM and FITC. IOUs clarified that a major difference in higher IOU projections for PV vs. Energy Commission’s 2016 IEPR forecast update (based on 2015 IEPR) is FITC extension and application of NEM 2.0. Note that NEM 2.0 is the same as NEM 1.0 – it is extended. No one knows what NEM 3.0 might look like. Also, ZNE requirements and continuingly declining prices are driving the higher IOU PV adoption forecasts vs. the 2016 IEPR update forecast.
  • Residential PV is dominating the PV growth. Residential adoption requires the use of the bass diffusion model rather than relying on historical data, as adoption rates change quickly.
  • Stakeholders asked whether it was possible toextract out and determine how much each factor is contributing to growth rates. IOUs responded that while it may be possible to extract out ZNE contributes to growth rates in forecasts, other factors may be embedded within the model.
  • IREC commented that it would be helpful to establish a framework on how to include/incorporate policies, such as changing TOU rates, in the future.
  • Not all IOUs include a TOU shift, which impacts adoption rates. CALSEIA commented it does not make sense to include declining cost of solar technology, but not account for declining benefits (TOU rates) within the model.
  • NRDC asked what is the “sweet spot” for accuracy, particularly given that not all data is available, and what is the appropriate forum to address this. Establishing a formal method to understanding this is useful. Topic to come back to in further WG discussions
  • IOUs do not currently conduct “scenarios” (e.g., “high-” or “low-“) growth scenarios for distribution planning – this would be too complicated. When the IOU start with the CEC forecast for distribution planning (which happens for most analyses) they use the mid- case.
  • ZNE compliance rates are uncertain but some internal SCE research shows 90 percent compliance for ZNE homes, which is high relative to expectations.
  • Strategy Innovation:once there is sufficient AMI data, the IOUs will have customer characteristics, loads, segments of customer. Using this information it is possible to add probabilities of load, weather, DER performance, etc. and this bottom up forecast is very reliable. Duke Energy has published analyses showing the value/reliability of this type of analysis. GSWG participants would like to see this report (or reports).
  • With regard to the point above about using AMI data, IOUs suggest that it could be problematic for distribution planning since even a single large customer coming on to a circuit can swamp prior predictions.
  • CPUC Energy Division provided a clarificationregarding the relationship between DER growth scenarios and IRP -- IRP adjusts growth scenarios in the long-term. If the IRP determines that a cost-optimal resource mix for GHG goals would include more DER, CPUC may change policies that affect [one or more] DER(s). Outputs of the IRP will be the main driver in Round 2 of the GSWG; this WG is also an internal input into IRP. Essentially, IRP will have a bigger impact on distribution planning than distribution planning will have on IRP. However, part of both processes will be to work out a feedback loop between these two proceedings (and other proceedings as appropriate).
  • Strategy Innovation: Disagree with CPUC regarding the point above about IRP being a venue for optimizing DER. Right now, the growth scenario results don’t have enough granularity to allow optimization of DER within the IRP.

Energy storage - IOUs

Each IOU presented its proposed approach for estimating behind-the-meter (BTM) storage growth, and discussed both system-level forecasting and disaggregation methods. IOUs presented joint challenges.

  • IOU presentations
  • PG&E – Energy Storage System Level Forecasts
  • Energy Commission does not include customer-sited BTM storage forecast in the IEPR process. Therefore PG&E has not filed a BTM storage forecast as part of the 2017 IEPR submissions
  • PG&E forecasts economically-driven (“organic”) customer-sited BTM storage
  • Bass technology diffusion framework
  • Estimated cost-effectiveness
  • Use-cases Modeled: peak shaving, rate arbitrage and back-up services
  • Use-cases not modeled: multi-use applications (e.g., wholesale markets)
  • The PG&E model uses storage-optimized inputs e.g., loadshapes, PV profiles, rates, storage system size; financial calculations, etc. This results in a market potential model which is translated into adoption curves using a Bass diffusion/S-curve approach.
  • PG&E – Energy Storage Disaggregation
  • Existing projects are allocated based on interconnection data
  • BTM storage is disaggregated to a county level with existing projects for non-residential installations and a Bass diffusion curve for residential adoptions.
  • At this time there are only about 400 BTM storage projects in PG&E service territory so forecasting based on historic trends is limited.
  • Near-term planning horizon, local information is useful (e.g., pending interconnection requests).
  • Long term planning horizon interconnection and RFO data will be available for larger projects (>500kW) and PG&E is considering a propensity model at the feeder level for smaller projects.
  • These methods will be revisited in subsequent planning cycles.
  • SDG&E – Energy Storage System Level Forecasts
  • The 2017-18 DPP Forecast will use ACR (AB 2514) targets to 2020. IEPR is currently does not provide specific energy storage forecast. Early adoption exceeding ACR targets will be analyzed and the forecast will be adjusted if trajectory continues into 2017.
  • SDG&E – Energy Storage Disaggregation
  • Small data set means this forecast is prone to errors during disaggregation – particularly given that many areas have no current adoption.
  • SDG&E subtracts larger projects (>500 kW) from total forecast.For the remainder, SDG&E uses expected correlation between existing PV adoption and electricity storage adoption.
  • Effectiveness of the “S-Curve” forecast will be evaluated when larger datasets become available.
  • SCE – Energy Storage System Level Forecasts
  • IEPR does not currently provide a specific ES forecast. SCE did not submit a separate storage forecast in 2017 IEPR demand forms.