Impact of carbon dioxide removal technologies on deep decarbonization of the electric power sector

The study investigates how carbon dioxide removal (CDR) technologies—specifically bioenergy with carbon capture and sequestration (BECCS) and direct air capture (DAC)—affect investments, operations, costs, and emissions in the U.S. electric power sector under deep decarbonization goals. Contextualized by the Paris Agreement’s temperature targets, prior scenario literature often relies on negative emissions but lacks technological, temporal, and spatial detail to evaluate power system impacts. The purpose is to quantify the role and value of a portfolio of CDR options within detailed power system planning, addressing uncertainties in speed, scale, cost, and acceptability of CDR. The research question centers on how availability and costs of DAC and BECCS influence least-cost generation portfolios, storage needs, emissions trajectories (including net-negative targets), and policy costs across varying CO₂ reduction stringencies.

Existing analyses of CDR have largely used integrated assessment models (IAMs), which lack the detailed representation of power system operations needed to capture variability, storage, and dispatchable low-carbon technologies. Prior detailed power system studies typically include only BECCS and do not consider a portfolio of CDR options. The literature acknowledges uncertainties in CDR’s speed, scale, cost, and societal acceptability, and discusses the potential need for CDR to neutralize residual emissions and draw down historical CO₂. Studies highlight power sector decarbonization as pivotal to economy-wide mitigation via electrification, but show gaps in assessing CDR within high-resolution capacity planning and dispatch frameworks. This paper fills that gap by jointly considering BECCS and DAC in a detailed U.S. power sector model and conducting systematic sensitivities on policy stringency, technology availability, and costs.

The analysis uses EPRI’s US-REGEN model, a linear-programming capacity planning and hourly dispatch model of the U.S. electric sector linked to an end-use demand model. The version applied is a single-year (2050) static equilibrium with hourly operations, allowing endogenous investment in generation, energy storage, transmission, CO₂ transport and storage, and CDR capacity. The model aggregates the U.S. into 16 regions and includes existing hydropower, nuclear (about 73 GW with 80-year extensions), pumped hydro, and interregional transmission as endowments. Variable renewable outputs are based on gridded hourly meteorological data (NASA MERRA-2) synchronized with hourly load from the end-use model. The model enforces a reserve requirement that firm capacity (excluding variable renewables) exceed regional peak load. Technologies represented include wind, solar PV (multiple configurations), nuclear (existing and new), natural gas combined cycles (NGCC), NGCC with CCS, gas turbines (NGGT), BECCS, and a range of storage options (lithium-ion batteries with endogenous energy-to-power ratio, compressed air, pumped hydro, and hydrogen pathways via electrolysis, storage, and hydrogen turbines). Transmission expansion is endogenous with assumed costs; CO₂ transport and storage include regional storage cost curves, NATCARB-based storage capacity limits, and endogenous pipeline investments with regional variation; regions without storage (e.g., New England) require pipelines. Emissions factors exclude lifecycle emissions. CDR technologies modeled: BECCS (based on Johnson and Swisher) with capital cost $5870/kW (≈$568/t-CO₂/year net removal capacity), heat rate 14.8 MMBtu/MWh, availability factor 60–80% monthly, coproduct electricity 0.85 MWh per t-CO₂ removed, and regionally varying CO₂ transport/storage costs; DAC (high-temperature liquid solvent design per Larsen et al.) with reference capital cost $614/t-CO₂/year (low-cost sensitivity $107/t-CO₂/year), electricity use 0.3 MWh/t-CO₂, natural gas use 5.6 MMBtu/t-CO₂, 30-year lifetime, and endogenous electricity prices. Both technologies assume permanent geologic storage of captured CO₂. Scenario dimensions: (1) Electric sector CO₂ caps from 70% to 140% reductions relative to 2005 emissions (2430 MtCO₂), with additional resolution near 100%; (2) Technology choice set: technology-neutral (full portfolio) vs. renewables-only; (3) Wind/solar/storage costs: reference vs. breakthrough (low cost) for VRE and batteries; (4) DAC cost: reference and low; (5) BECCS capital cost ($3250–$10,000/kW) and heat rate (6.8–17 MMBtu/MWh) sensitivities; (6) Biomass availability: reference plus ±50% regional supply step adjustments using FASOM-GHG-derived piecewise supply curves; (7) CO₂ storage infrastructure/costs: sensitivities restricting pipelines and equalizing storage costs across regions. Additional sensitivities extend caps to net-negative levels (up to 140%) to approximate offsetting residual nonelectric emissions. Costs reported as incremental policy costs are differences in annualized electric sector expenditures between policy scenarios and a no-CO₂-policy reference, including investment, fuel, O&M for generation, transmission, storage, and CDR; distribution costs are excluded. Natural gas price assumed $4/MMBtu; economy-wide CO₂ price in end-use model starts at $50/tCO₂ in 2020 and escalates at 7% real.

- CDR availability lowers the cost of achieving deep CO₂ reductions and has larger effects at higher policy stringency by placing a ceiling on marginal abatement costs and flattening the marginal abatement cost curve. - Deployment thresholds: CDR is only deployed for electric sector reductions ≥90% vs. 2005. For 100% reductions with both CDR options at reference costs, BECCS is preferred over DAC; when DAC is the only CDR option at reference cost, 91 MtCO₂/year DAC is deployed. - DAC cost sensitivity: At low DAC capital cost ($107/t-CO₂/year), DAC deployment exceeds 340 MtCO₂/year for the 100% reduction case (≈14.2% of 2005 electric sector CO₂), substantially displacing BECCS. - Portfolio impacts: With CDR, capacity and generation from gas (with and without CCS) increase—especially low-capacity-factor gas turbines—while advanced nuclear, renewables, and long-duration storage (e.g., hydrogen) decrease relative to no-CDR cases. Example: Advanced nuclear capacity in the 100% cap scenario falls from 117 GW (no CDR) to 47 GW (DAC+BECCS) or 73 GW (DAC only). - Storage impacts: Battery deployment remains high across scenarios, but long-duration storage grows nonlinearly without CDR and is substantially limited when CDR is available. CDR enables cost-effective reliance on gas turbines instead of long-duration storage for firm capacity at very low emissions. - BECCS scaling: Under the 100% cap, total BECCS capacity ranges from 0–41 GW across BECCS cost and heat rate assumptions; under 140% CO₂ reductions, BECCS reaches 42.2 GW in both dispatchable and must-run cases due to high utilization driven by carbon removal value. - Net-negative targets and CDR mix: With increasing reductions beyond 100%, BECCS is preferred until marginal biomass costs rise; with reference costs, BECCS saturates near 110% reductions (−243 MtCO₂/year), after which DAC becomes the marginal least-cost CDR. The crossover occurs earlier with low biomass availability (≈105%) and much earlier with low-cost DAC (≈90%). For 140% reductions, total CDR is 1050 MtCO₂/year: ~79.6 MtCO₂/year offsets residual power sector fossil emissions; the remainder offsets other sectors. - Economic outcomes: For the 100% cap, annual policy cost savings from CDR are $21.2 billion/year (DAC only) and $28.3 billion/year (DAC+BECCS) relative to no CDR. Renewables-only decarbonization is costlier than technology-neutral approaches: +$33.5 billion/year (44.9%) without CDR under reference costs (+$14.3 billion/year or 45.7% under breakthrough costs). CDR availability reduces CO₂ allowance prices markedly and makes total costs increase approximately linearly with abatement. - Operations: DAC runs at very high capacity factors (~8000 hours/year), indicating limited value as a flexible, intermittent load for absorbing surplus VRE. BECCS plants also operate at high utilization; the value of carbon removal exceeds electricity value, leading to near-constant operation. - System load: DAC electricity consumption is modest relative to total load and storage losses. In the 100% cap DAC-only case, DAC uses 24.8 TWh/year (0.42% of end-use demand); in the 140% cap DAC-only case, 322 TWh/year (5.39%). By comparison, net storage losses without CDR in the 100% cap case are 548 TWh/year. - Biomass use: BECCS at the 100% cap consumes ~1.81 quads of biomass; 2019 U.S. biomass production was 4.82 quads, indicating a significant but manageable scale under assumptions. - Spatial deployment: BECCS is distributed across regions with high biomass availability and CO₂ storage access (Gulf, Southeast, Ohio Valley, Midwest). DAC concentrates in regions with lower combined electricity and storage costs (South Atlantic, California, MISO South, Texas). Pipeline constraints mainly affect New England due to lack of local storage. - Flexibility sensitivity: Treating BECCS as must-run vs. dispatchable under 140% reductions shows minimal system differences because BECCS already operates at high utilization driven by carbon value.

The findings show that including CDR technologies in detailed power system planning materially reduces the cost and complexity of achieving deep decarbonization and net-zero targets. CDR flattens the marginal abatement cost curve by providing a backstop mitigation option, enabling residual positive emissions from low-capacity-factor gas assets while avoiding extensive deployment of high-cost, low-utilization technologies such as long-duration storage and large additions of firm low-carbon capacity. The results address the research question by quantifying how availability and relative costs of BECCS and DAC reshape least-cost portfolios, reduce sensitivity of capacity mixes to abatement stringency, and enable net-zero and net-negative outcomes at lower cost. They also demonstrate that DAC’s economic value stems from sustained high utilization rather than flexibility services, and that BECCS’s carbon value drives near-continuous operation. Spatially, deployment aligns with regional cost and resource endowments, and pipeline/storage constraints can influence siting where geologic storage is absent. Overall, CDR provides system flexibility and optionality to hedge uncertainties in other emerging technologies while maintaining renewables as the backbone of generation.

This study contributes a high-resolution assessment of how a portfolio of CDR options—BECCS and DAC—affects U.S. electric sector decarbonization. It demonstrates that CDR lowers total and marginal costs of stringent CO₂ targets, reduces dependence on long-duration storage and certain firm low-carbon technologies, and enables cost-effective pathways to net-zero and net-negative emissions. BECCS tends to dominate up to net-zero under reference assumptions, with DAC increasingly attractive as biomass costs rise or DAC costs fall in higher CDR-demand scenarios. Policy implications include the value of incorporating CDR into planning and policy design to cap abatement costs and provide flexibility. Future research should extend to intertemporal transition dynamics, operational and ancillary service constraints, broader environmental tradeoffs (land, water, lifecycle emissions), policy mechanisms and financing for first-of-a-kind CDR, and interactions with nonelectric decarbonization and potential CO₂ utilization pathways.

- Modeling uses a single-year (2050) static optimization, abstracting from transition dynamics and learning-by-doing over time. - Operational detail is limited: no explicit sub-hourly constraints, ancillary services markets, or sub-state granularity. - Demand is fixed across scenarios in the electric sector analysis; demand-side flexibility is not endogenously optimized. - Lifecycle emissions (e.g., biomass supply chains), land-use change, and water impacts are not included in emissions accounting. - Geological storage characterization is simplified to regional cost curves and capacities; site-specific risks and permitting are not modeled. - Policy, RD&D financing, and public acceptance considerations are beyond scope; CO₂ utilization options are not considered (captured CO₂ assumed stored). - Results depend on assumed technology costs, fuel prices, biomass availability, and storage costs; while sensitivities are explored, real-world uncertainties remain.