The power sector is crucial for economy-wide decarbonization. Natural gas has historically aided emissions reductions in the U.S. electric sector, but its long-term role under deep decarbonization, especially net-zero targets, is debated. Utilities are setting net-zero targets that may include new gas-fired capacity, raising questions about gas's compatibility with decarbonization goals. Previous studies have examined natural gas's emission-reducing potential, but few have addressed zero-emission goals or accelerated decarbonization aligned with the U.S. target of "100 percent carbon pollution-free electricity by 2035". This study aims to fill this gap by using a detailed energy systems model (US-REGEN) to assess natural gas's role in deeply decarbonized U.S. electricity systems, considering various sensitivities (accelerated zero-emissions goals to 2035, regional power system contexts, technology costs, and policy assumptions). The study's analysis features hourly temporal resolution, endogenous end-use decisions and load shapes, and a comprehensive suite of technological options. This detailed approach allows for a more accurate representation of the interplay between variable renewables, energy storage, and dispatchable low-carbon technologies.

Prior research examined natural gas's role in power sector emissions reduction, but lacked the focus on zero-emission goals and accelerated decarbonization timelines considered here. Existing literature didn't account for recent cost declines in renewables and storage, nor the evolving natural gas prices that impact economic feasibility.

The study utilized the US-REGEN energy systems model, which has hourly resolution and considers a wide range of generation technologies (variable renewables, energy storage, firm low-emitting technologies like hydropower and geothermal, and thermal units with or without CCS). The model also incorporates carbon removal (BECCS and DAC), endogenous end-use decisions, load shapes, and inter-regional transmission. The scenarios considered three policy targets: Reference (existing policies), Carbon-Free (no CO2-emitting technologies after the target year), and Net-Zero (emissions balanced by sequestration). Target years were 2035 and 2050. Sensitivity analyses explored variations in natural gas prices, renewable and storage costs, CCS availability, methane leakage rates, policy incentives (45Q tax credits), and long-duration energy storage availability. The model endogenously determines electricity demand and hourly load shapes, reflecting deep decarbonization contexts with CO2 pricing for non-electric sectors.

Natural gas capacity and generation are robust elements of least-cost decarbonization portfolios, both during the transition and at net-zero. However, the extent depends on policy design, methane mitigation, and technological change. Natural gas plays a role in least-cost paths in most sensitivities, even in zero-emission systems, unless explicitly excluded by policy. It provides firm, flexible capacity crucial for meeting demand as coal retires and electrification increases. Regions with lower-quality renewable resources rely more on natural gas or face higher decarbonization costs without it. Wind and solar have substantially larger generation shares than natural gas in most net-zero scenarios (52–66% vs. 0–19%). The study also quantifies transition risks for various stakeholders. Carbon dioxide removal (CDR), mainly via BECCS (with DAC playing a larger role under specific assumptions), is vital for achieving net-zero goals. Accelerated decarbonization (targeting 2035 instead of 2050) leads to relatively higher natural gas contributions. Regional differences exist; regions with lower-quality renewable resources have higher natural gas shares and experience higher electricity price increases under Carbon-Free scenarios. Prohibiting natural gas significantly increases decarbonization costs (over $1.6 trillion vs $1.3 trillion for generation in 2035). Electricity prices are heavily influenced by policy targets and timetables, with Carbon-Free policies leading to substantial price increases that could disincentivize electrification. The system value of natural gas shifts from energy provision to capacity provision over time; capacity factors for natural gas decline significantly as renewable penetration increases. The study explores sensitivities to wind and solar costs, finding that lower renewable costs don't always translate to increased energy storage deployment. Higher end-use electrification increases generation from various technologies but reduces natural gas's role by 2050. Hydrogen's role remains modest due to high marginal abatement costs.

The findings highlight natural gas's potential role in least-cost decarbonization pathways, contingent on policy design and carbon removal availability. Policies that allow for carbon removal offsets from natural gas offer a more cost-effective transition. The analysis challenges assumptions that natural gas will be a stranded asset in a zero-emissions future, demonstrating its continued value in providing firm capacity even with high renewable penetration. Regional variation in resource quality and policy significantly influence natural gas's role. The study's hourly resolution and consideration of endogenous end-use decisions provide a more nuanced understanding of natural gas deployment compared to previous studies.

Natural gas can play a role in a net-zero electricity sector, particularly when paired with carbon removal technologies. Restricting natural gas increases decarbonization costs. The extent of natural gas's contribution is highly sensitive to policy design, technology costs, and market uncertainties. Future work should investigate operational constraints, ancillary service markets, market design for high-renewable scenarios, and the role of natural gas in net-zero economy-wide futures.

The analysis assumes exogenous fossil fuel price trajectories, doesn't explicitly model all operational constraints or ancillary services markets, considers hydrogen only for new investments, lacks subhourly or sub-state detail, and uses a single historical weather year. These limitations could affect the precision of the results, but the overall findings remain robust.