A Data-Driven Framework Shows How the U.S. Can Retire Coal Plants Smarter and Faster

A new study from the University of California, Santa Barbara introduces a detailed, data-heavy framework designed to speed up the retirement of the United States’ remaining 198 active coal plants. Even though coal power has been in long-term decline, more than 114 plants—representing around 105 gigawatts of generating capacity—still have no retirement dates and are projected to keep running through 2035. That timeline is out of sync with national climate targets, which require a complete coal phaseout by 2035 to stay on track for net-zero emissions.

The researchers behind this work set out to understand why some plants shut down while others linger, and how targeted action could push the U.S. closer to its decarbonization goals. Their central idea is that coal retirements are not driven by a single factor like age or profitability, but by a complex mix of technical, economic, environmental, health-related and political forces. So instead of relying on broad, one-size-fits-all policies, the authors argue for a more contextual, data-sensitive approach.

What Makes This Framework Different

The research team—led by Sidney Gathrid and guided by Grace C. Wu—built a model that analyzes 68 different variables for every coal plant in the country. These variables cover everything from fuel flexibility, emission levels, and debt load to local pollution impacts, state policies, and regional energy-market trends.

Using a combination of graph theory and topological data analysis, they grouped the entire national coal fleet into eight clusters. Each cluster represents a type of plant with shared characteristics, and each cluster responds to different retirement pressures.

This led to the creation of what they call a contextual retirement vulnerability score—or CRV score—which measures how similar an active plant is to a coal plant that has already announced retirement. A higher score means the plant is more likely to be retired early if pushed by the right mix of strategies.

According to the study, about 28% of coal plants without retirement plans already fall into the category of highly vulnerable. These are realistic “quick wins” where focused policy or advocacy could meaningfully speed up the closure timeline.

Understanding the Eight Plant Groups

While the paper includes all eight clusters in technical detail, the broad categories include:

• High Health Impacts Plants
  These plants are located in regions with significant asthma rates, air-quality violations, or heavy particulate emissions. Public-health organizations could target these plants with enforcement pressure and awareness campaigns.
• Expensive Plants
  These are plants that are simply not profitable, often because they’re old, inefficient, or burdened with financial liabilities like stranded debt. Economic incentives or debt-restructuring programs can help move these toward retirement.
• Plants in Anti-Coal Regions
  These operate in states or counties where political sentiment and public policy already favor renewables. Clean-energy targets, solar or wind growth, and supportive regulation make these easier to retire.
• Fuel Blend Plants
  One example cluster contains plants capable of co-firing with natural gas, which may seem flexible but also reveals a unique set of vulnerabilities. These plants often remain financially shaky and technologically outdated.

Each cluster comes with its own “retirement archetype”—essentially a profile describing why plants in that group retire and what tools tend to work.

A Closer Look: Belews Creek as a Case Study

One of the clearest examples in the study is the Belews Creek power plant in North Carolina. It is a 2.49-gigawatt, nearly 50-year-old facility that can burn up to 50% natural gas. Despite this fuel flexibility, it has some serious weaknesses:

• It is among the top particulate matter polluters in the country.
• It ranked 26th out of 198 plants for fine-particle emissions.
• It was carrying approximately 46 million dollars in debt in 2020.
• It operates in a state rapidly expanding solar power.
• North Carolina has adopted coal-debt securitization policies, which help utilities close uneconomic plants.

Interestingly, there were early conversations about replacing Belews Creek with a small modular nuclear reactor, but the owner—Duke Energy—has since postponed the plant’s retirement. This delay illustrates one of the study’s main points: coal-plant decisions involve complicated financial calculations, political trade-offs, and reliability concerns, so retirement strategies need to address those specific issues rather than rely on general assumptions.

Why Many Coal Plants Are Still Running

The researchers note that while market declines have shut down many coal facilities over the past decade, the plants that remain often have unique local conditions that keep them online. Some examples include:

• Heavy financial entanglements like outstanding debt
• The plant’s role in grid reliability
• Local political support for coal jobs
• State-level regulatory barriers
• Regional lack of renewable alternatives or transmission infrastructure

This explains why age alone cannot predict retirement. Using age-based rules would still leave an estimated 37 gigawatts of coal capacity running in 2035.

Suggested Strategies for Policymakers

Because each cluster responds to different pressures, the study suggests a range of targeted strategies:

• For health-impact clusters: stronger environmental enforcement, EPA scrutiny, local health-impact reporting
• For unprofitable plants: economic transition packages, buyouts, securitization programs, market-based mechanisms
• For regions with political opposition to coal: leveraging state-level clean-energy mandates
• For plants with fuel flexibility: incentives to transition from coal permanently rather than operate in hybrid mode

The framework allows decision-makers to match plants to the tools most likely to work, rather than wasting resources on ineffective pressure points.

Why This Matters Beyond Coal

A major contribution of this work is that it bridges advanced mathematical modeling with practical policy guidance. The authors emphasize that their method could be applied not just to coal retirements but to any complex decarbonization challenge—such as industrial emissions, renewable siting decisions, or grid-modernization planning.

The research team even released the analysis as an open-source toolkit, encouraging policymakers, NGOs, utilities and researchers to adapt the method.

Additional Context: The Broader State of U.S. Coal Power

To expand the article beyond the study itself, here are some broader insights that help frame the national landscape:

The Decline of U.S. Coal

Coal has steadily dropped from being the dominant U.S. electricity source decades ago to making up a much smaller share today. This shift is driven by:

• Rapidly falling costs of solar and wind
• The growth of battery storage
• The expansion of natural-gas generation
• Increasing public and regulatory pressure over air quality
• Economic challenges facing aging plants

Despite this decline, coal still plays a role in certain regions due to reliability concerns and local politics.

Why Grid Reliability Adds Complexity

Some coal plants operate as baseload anchors in regions with limited transmission capacity or limited renewable deployment. This means retirements must be paired with careful planning to maintain stability, which is why coordinated, data-driven frameworks like the UCSB model can be useful.

The Future of Coal Retirements

With many utilities setting net-zero goals, and with states adopting clean-energy standards, more retirements are expected. However, progress varies widely by state, utility and region, making strategic decision-making extremely valuable.

Research Paper

Strategies to accelerate US coal power phase-out using contextual retirement vulnerabilities
https://www.nature.com/articles/s41560-025-01871-0