MCR

Home > Energy Products and Services > Client Stories > Forecasting electric vehicles and photovoltaic systems for Indianapolis Power & Light’s Integrated Resource Plan

Forecasting electric vehicles and photovoltaic systems for Indianapolis Power & Light’s Integrated Resource Plan


“Having a third party complete this analysis was essential to IPL; MCR was knowledgeable, provided a well-done report, and presented the material really well. MCR is very organized and on top of project management. They have great communication. They are very easy to work with.”

—Erik Miller, Senior Research Analyst


Download the IPL client story

Background

While developing its 20-year integrated resource plan, Indianapolis Power & Light Company (“IPL”) needed to incorporate electric vehicles (“EV”) and behind the meter distributed solar photovoltaic technology (“PV”) into its load and sales forecasts. The regulator and stakeholders were interested in seeing the effect of these two electro-technologies on the IPL forecast that would drive development of its preferred resource portfolio. With the goal of developing a reasonable and defensible forecast that could be easily explained, IPL approached MCR for help.

Solution

The MCR team brought deep research and analysis experience with EV and PV technologies, as well as experience with statistical modeling and utility load forecasting. We began the project by gathering background information on EV, PV, rates, and market data, including interval sales (kWh energy) data from IPL, MCR’s data libraries, and online resources, such as the National Renewable Energy Laboratory’s (“NREL”) PVWatts system. From this data, we defined prototypical EV and annual charging consumption, as well as residential and commercial PV systems and sizes. We then applied IPL sales data for applicable rates and riders to assemble 8,760-hour (annual) load curves for light duty (i.e., cars, pick-up trucks, etc.) EV charging and, adding PVWatts data for PV production. This analysis established a baseline and drove MCR’s approach to first forecast the annual number of units of EV, residential PV, and commercial PV, and then apply the prototypical system load data to develop the energy and capacity impacts by rate period. The key modeling challenge, therefore, was to develop 20-year forecasts of the number of units of EV and PV.

To resolve this challenge, MCR considered approaches to forecasting numbers of units of emerging technologies (market adoption), often referred to as the Sigmoid Adoption or S-Curve. Such modeling is usually driven by logistic and similar regression techniques and mathematical models, such as the Bass Imitation Model. Forecasting technologies that are early in their market adoption, such as EV and PV, depends on many inputs and assumptions (e.g., pace of innovation, maximum achievable market penetration, etc.) that have a significant impact on the forecast as a whole. As MCR began to develop initial forecasting models, we quickly found that the necessary input data to our calculations was either not readily available, not transparent enough to allow validation, or both. Therefore, we concluded that developing forecasting models based on adoption or S-Curves would require significant effort and would likely produce unreliable results.

As an alternative for developing the PV forecast, we examined several online tools and calculators, including Project Sunroof, Deep Solar and NREL’s dGen. From this review we concluded that these publicly available tools lacked adequate transparency and granularity for utility-specific forecasting.

Given the lack of confidence in either an S-Curve modeling or online tools approach, MCR conducted an extensive EV and PV forecasting methodology literature review that examined over 60 references on EV forecasting and over 30 sources on PV forecasting. Based on the review, we decided to use some well-known national forecasts of EV and PV adoption, and scale them to IPL’s service territory. The scaling was based on factors such as the installed base and historical growth rates of EV and PV in IPL’s territory, population, income, policy/regulatory considerations, etc. IPL agreed with the approach; according to Erik Miller, Senior Research Analyst, “It was a linear approach applying some other source data to create our forecast.”

Results

MCR successfully developed forecasts for numbers of EV and PV systems, EV and PV energy impacts and PV capacity impacts for the 20-year forecast horizon. We presented our results to the IRP stakeholder group; the results were generally well received. IPL will integrate MCR’s forecast with their native load and energy forecast to use in developing their 20-year IRP.