Best Practice Modeling to Achieve Low Carbon Grids
Transitioning to a zero carbon electricity grid is likely to be a multi-trillion dollar undertaking1. Nearly one in three Americans currently have difficulty paying their energy bills,^2 underscoring the importance of making this transition at least cost and ensuring that electric utilities, regulators and grid operators have the best possible analytical tools available to plan future energy resource investment.
Utilities use a type of tool called a capacity expansion model to plan least-cost portfolios of energy resources (including generation, transmission and energy storage) to meet forecasted energy demand and clean energy goals. Many of the models in wide use today are ill-equipped to cost effectively guide the transition to a low-to-zero-carbon grid because they were not designed to account for the variability of renewable energy resources from hour to hour over a year, or across multiple years.
In this study, we articulate our view of best practices that modelers can use to plan cost effective and reliable low carbon grids. This view is backed by leading academics and practitioners. We highlight some key limitations of many of the commercially available tools today, summarize research underscoring the costs of these issues, and point to methods that modelers should use to plan lower carbon, lower cost, more reliable grids. Finally, we summarize a case study using Form Energy’s capacity expansion tool and data from one of Form Energy’s commercial partners^3 , to demonstrate the measurable cost and reliability benefits that best-practice modeling methods can bring to utilities and their customers. The case study underscores many of the major findings from academia.
Limitations of incumbent capacity planning tools
The capacity expansion models in wide use today were designed around a planning mindset that assumed that thermal power plants are predictable and available when needed. Thus, if the electric grid had enough resources to meet peak demand, the grid would be capable of meeting demand at any other time. These tools were also built in an era that lacked the computational power and analytical methods available today. As a result, the tools include certain simplifications to save computational time and analytic complexity. Most notably, legacy capacity expansion models:
- Design resource portfolios based on limited time samples: Rather than make investment decisions based on a model of at least one full year, incumbent models design resource portfolios using a small sample of hours or days, and assume that this trimmed down time series accurately captures the full intra-year variability of renewable resources and storage.
- Design portfolios using ‘typical’ operating conditions: Incumbent models optimize portfolios for using ‘typical’ weather data, relying on reliability models to ensure the resulting portfolios are reliable across weather conditions. However, renewable generation and demand varies significantly from year-to-year, and portfolios designed for a single snapshot are less cost effective and reliable than a portfolio designed for diverse grid conditions.
- 1 One recent cost estimate comes from Chloe Holden, 2019. The Price of a Fully Renewable US Grid: $4.5 Trillion.
- 2 U.S. Energy Information Administration, 2018. One in three U.S. households faces a challenge in meeting energy needs.
- 3 The case study relies on sensitive data. We have anonymized the data and partner identification as a result.
Capacity expansion model capabilities needed
Academic and industry progress in building new capacity expansion models has led to an emerging set of best practices about how to plan low carbon grids that rely substantially on renewables and storage. Where possible, capacity expansion models should:
- Make investment decisions based on at least one full year of grid operations at hourly resolution , including weather and load variability that reflects day-to-day, week-to-week, and season-to-season fluctuations.
- Make investment decisions based on multiple weather years and key future system conditions , such as technological availability, commodity prices, or other variables.
Incorporating this level of granularity often requires modeling trade-offs, and the academic literature points to advanced modeling techniques that can avoid the need to capture multiple years of weather and system data at 8,760-hour granularity. Where these techniques are employed, it’s critical that their efficacy is benchmarked against the full granularity model. Despite the impact of model simplifications on planning outcomes, few commercial models today use the advanced methods pursued in academia and none provide any guarantee of the performance of the model simplifications employed.
The benefits of modern capacity expansion modeling
Lower Costs: Models that represent hourly grid operations and can co-optimize portfolios across multiple scenarios produce lower cost portfolios than models that use time sampling and typical weather years. Our case study confirms existing research and finds that, for one particular utility, full year, hourly resolution modeling produces portfolios that are more than 10% cheaper than time sampled portfolios.
Accurate Technology Representation: Models that preserve the full time chronology of a year can accurately model technologies like long duration energy storage, which can produce energy continuously over days and can shift energy across seasons. By contrast, models that break apart the year’s chronology often can’t accurately model such technologies. Time sampling techniques overestimate baseload value, underestimate flexibility value, and often exclude long duration storage technologies altogether.
Increased reliability: Full year, hourly resolution modeling and co-optimization across scenarios of future system conditions produces portfolios that are more reliable than those produced by less capable models that consider single snapshots of the future.
The U.S. spends roughly $200 billion each year^4 on electricity generation to meet growing electric demand and to replace old power plants. These investments translate to costs for customers, and, if insufficient, the electric grid’s reliability can suffer. Further, investments in new fossil fueled power plants can lock utilities into producing high levels of greenhouse gas emissions and causing other environmental and health damages that last decades. It is essential that decision makers make the best, most informed investments possible.
Electric utilities and their regulators rely on capacity expansion models
- computer models that help utilities identify the least cost portfolio of power infrastructure investments needed to meet demand
- as one of the primary tools to inform their investment decisions.^5 Unfortunately, the vast majority of the capacity expansion models used today were developed for electric grids with fossil fuel backbones and embed many assumptions that reflect this fact.
The power system is changing. In 2010, wind and solar contributed less than 2.4% of U.S. electricity generation capacity, but this quadrupled to 9.9% by 2019. Renewables and storage comprise the majority of planned power investments around the country, portending a continuation of these trends (see Figure 1). As the power sector transitions, the models utilities and developers use to guide their investment decisions need to adapt as well.
This study compares new, best-in-class capacity expansion modeling approaches with existing modeling tools to evaluate how their differences impact electric resource needs, portfolio costs and reliability in grids with high levels of renewables. The study reviews the current state of capacity expansion planning and highlights some of the primary shortcomings of these planning techniques. This paper then recommends capacity expansion modeling best practices drawn from academic and industry research. Finally, this paper summarizes a simple case study based on data from one of Form Energy’s utility partners, to underscore the value of these recommendations.
Figure 1: Renewables and storage as a fraction of planned generation capacity by state
To read the full whitepaper you can download it here.