Unaddressed non-energy use in the chemical industry can undermine fossil fuels phase-out

The COP28 Global Stocktake called for transitioning away from fossil fuels in energy systems to achieve net zero by 2050, but did not explicitly address the non-energy use of fossil fuels. Approximately 30 EJ (about 13%) of global fossil fuel production is used for non-energy purposes, two-thirds of which serve as feedstocks for primary chemicals, contributing roughly 1 GtCO2 per year. Because refineries co-produce fuels and materials feedstocks (e.g., naphtha co-produced with gasoline and diesel; refinery propylene and aromatics), decarbonising energy systems without tackling non-energy uses risks sustaining fossil fuel infrastructure. Concurrently, demand for materials (polymers, chemicals, fertilisers, lubricants, asphalt, solvents) is expected to grow, including for energy transition technologies. This paper investigates, from an integrated assessment perspective, how a growing chemical sector intertwined with oil refining affects fossil fuel phase-out and chemical decarbonisation, and how chemicals compete for biomass, hydrogen and CO2 with other mitigation options. The hypothesis is that while primary chemicals are hard-to-abate and hard-to-defossilize, the sector can contribute strategically to deep decarbonisation if alternative feedstocks and enabling policies are available.

Prior studies have examined mitigation in primary chemicals via fuel switching for process heat, feedstock substitution, CCS/CCU, and circular (bio)economy strategies, individually and in combination. The IAM literature has explored declining fossil fuel demand under climate policies and some sectoral decarbonisation strategies, yet has largely not addressed the co-production dynamics between refineries and chemicals under stringent climate scenarios. The authors note a gap in IPCC AR6 scenarios regarding detailed representation of fossil fuels as feedstocks for primary chemicals alongside refinery co-production. The paper situates its contribution against studies highlighting the potential of biomass conversion to materials, the importance of biogenic carbon storage assumptions, and the role of recycling and circularity in reducing virgin plastics demand.

The study uses the COFFEE model, a global, perfect-foresight, linear programming, least-cost integrated assessment model based on the MESSAGE framework with explicit representations of energy, land use (AFOLU), greenhouse gases (CO2, N2O, CH4), and carbon dioxide removal options (land sinks, BECCS, DAC, and material carbon storage). COFFEE features detailed oil and gas sector resolution, including crude quality types, trade, and refinery typologies (Existing Topping, Cracking, Hycon; New Cracking and Hycon) with different activity modes geared toward diesel, gasoline, or kerosene. This work augments refining to include regional propylene and BTX co-production from FCC and catalytic reforming, and allows new refineries to increase HVC output up to 15% for higher petrochemical integration. Industry process representation explicitly models primary chemicals: high-value chemicals (HVCs: ethylene, propylene, butadiene, aromatics), ammonia, and methanol. For HVCs, 12 technologies cover various feedstocks and routes: steam cracking (ethane, naphtha), naphtha catalytic cracking (NCC), FCC and catalytic reforming (integrated and standalone), propane dehydrogenation (PDH), dimerisation, Catadiene, metathesis (MTT), ethanol to butadiene (ETB), methanol-to-olefins (MTO), methanol-to-aromatics (MTA), and bioethanol dehydration (BDH). For syngas-based products, technologies include SMR, coal gasification, biomass gasification, partial oil oxidation (ammonia only), electrolysis (ammonia only), and CO2 hydrogenation to methanol (CDH), with ammonia via Haber–Bosch. Feedstock categories: fossil-based (coal, natural gas, refinery products including LPG, naphtha, heavy oil, with gasoline–naphtha interconversion), bio-based (bioethanol, bio-naphtha, bio-LPG via FT-BtL from various crops), and CCU-based (methanol via CO2 hydrogenation from DAC or concentrated sources). Captured CO2 enters a regional pool for storage or utilisation. Scenarios assume SSP2 socioeconomics and static plastics demand and end-of-life shares to isolate supply-side dynamics. The baseline National Policies implemented (NPi) scenario includes policies enacted by 2020. A 1.5C scenario uses an IPCC AR6-consistent carbon budget (>66% likelihood of limiting warming to 1.5°C). Sensitivity cases under 1.5C include: gCCS (restricted global CCS rollout), PBIO (biomass capped below 100 EJ/yr), MNEToff (biogenic carbon storage in biomaterials not credited), and all (combining all restrictions). Global CCS caps and biomass bounds reflect protocol settings from IAM intercomparison projects (ENGAGE, NAVIGATE).

• Alternative feedstock potential: Across mitigation scenarios, up to 62% of total chemical feedstock could shift to alternatives by 2050 (upper bound in gCCS), with a lower bound of 28% when all restrictions apply. Achieving this requires significant scale-up of biomass utilisation and carbon capture technologies.
• Emissions outcomes: The chemical sector can reach net removals of about -0.73 GtCO2/yr (1.5C) to -1 GtCO2/yr (gCCS) by 2050 if biogenic carbon storage in non-recycled and non-incinerated biomaterials is credited, driven largely by BECCS in ammonia and biogenic carbon storage. Under combined constraints (all), the sector has residual emissions in 2050 and turns net-negative only by 2070.
• System-wide stringency under constraints: Scenarios limiting biomass, CCS deployment, and biogenic material storage require at least 6 GtCO2/yr additional reductions by 2030 relative to the unconstrained 1.5C scenario to remain within the carbon budget.
• Fossil fuel use trajectories: By 2050 under 1.5C variants, coal use falls 95–99%, gas 19–47%, and oil 40–60% versus 2020. By 2100, oil shows smaller reductions (22–42%) compared with gas (22–87%) and coal (82–97%). The non-energy share of fossil fuel use rises from 14% (2020) to 30–38% (2050) and 40–59% (2100) as fossil feedstock demand persists amid limited alternative feedstock expansion and gasoline–naphtha substitution dynamics.
• Refining dynamics: Refining capacity declines over the century; utilisation falls from ~70% (2020) to 30–50% by 2030, then rises toward 100% by 2070 as new, optimised capacity is built. Total oil input capacity declines from about 191 EJ/yr (2010) to 103–132 EJ/yr (2050) and 44–77 EJ/yr (2070). Surviving greenfield refineries are more complex (Hycon) and up to 15% more integrated with petrochemicals, even as refinery-sourced chemicals drop 49–77% by 2050.
• Feedstock and technology shifts in chemicals: Biomass use in chemicals rises to 6–25 EJ/yr by 2050 (12–25 by 2070). CCU becomes important for methanol; BECCS is prioritised for ammonia in constrained contexts. Methanol demand for non-energy uses grows, reaching up to 368 Mt/yr by 2070 as an intermediate for HVCs.
• HVC pathways: Steam cracking remains dominant. Transport electrification drives gasoline oversupply, facilitating gasoline-to-naphtha pool shifts and growth of Naphtha Catalytic Cracking (NCC) to 32–52% of HVC market by 2070. On-purpose routes (PDH, MTT, MTO) expand to fill propylene/aromatics gaps. By 2050, bio-based feedstocks produce 35% (1.5C) to 51% (gCCS) of HVCs, though fossil dependence persists, especially under tight resource constraints.
• CCS allocation: Under gCCS, CCS use is prioritised in the chemical sector due to lower abatement costs; with combined biomass and CCS constraints (all), this advantage diminishes, and DAC plays a larger role when biomass alone is constrained.
• Regional nuance: Asia uses up to three times more biomass for non-energy purposes than OECD90+EU by 2050 when biogenic storage in biomaterials is considered.

The findings confirm that while primary chemicals are hard-to-abate and hard-to-defossilize, the sector can support system-wide decarbonisation if alternative carbon feedstocks and enabling policies are scaled. Addressing non-energy uses is essential to avoid prolonging oil refining and co-production of fuels, which could undermine fossil fuel phase-out. Ambitious feedstock substitution can materially reduce refinery utilisation and associated emissions, but constraints on biomass, CCS, and biogenic material storage impede defossilisation, maintaining fossil feedstock reliance. Technology restructuring is product-specific: steam cracking persists while NCC and on-purpose routes (PDH, MTO/MTA, BDH/ETB) grow; BECCS in ammonia offers significant negative emissions, and CCU underpins methanol expansion. The results underscore the need to improve IAMs’ representation of fossil fuel–materials linkages to capture interdependencies and guide integrated policy design that recognises trade-offs and synergies across energy, materials, land use, and carbon management.

This study shows that unaddressed non-energy uses can slow the fossil fuels phase-out by sustaining refinery operations via feedstock demand. Under stringent climate policies, up to 62% of chemical feedstocks could shift to alternatives by 2050, enabling the sector to deliver up to about -1 GtCO2/yr by mid-century if biogenic carbon storage in durable biomaterials and BECCS are realised. However, combined constraints on biomass, CCS, and material storage delay net-negative outcomes and increase pressure on other sectors. Policymakers should integrate energy and materials strategies, targeting both energy and non-energy fossil uses, and plan for large-scale deployment of biomass, CCU/CCS, and infrastructure for alternative feedstocks. Future research should deepen demand-side and end-of-life integration (reduction, reuse, recycling), assess sustainability dimensions of biogenic carbon storage, evaluate emerging pathways such as crude-to-chemicals (COTC), and explore shifts toward oxygenated bio-based materials platforms.

The assessment focuses on supply-side measures with static demand and end-of-life shares for plastics, excluding demand-side shifts (reduction, reuse) and expanded recycling pathways. It explores a limited technology set and does not include crude-oil-to-chemicals (COTC) due to data limitations. The effectiveness and broader sustainability of biogenic carbon storage in materials are not comprehensively assessed. These choices simplify analysis of supply dynamics and systems integration but may affect the generalisability and timing of inferred pathways.