Reducing uncertainties in greenhouse gas emissions from chemical production

Luke Cullen, Fanran Meng, Rick Lupton, Jonathan M. CullenNature Chemical Engineering 2024 · DOI

Reducing uncertainties in greenhouse gas emissions from chemical production

Abstract

Uncertainties in greenhouse gas emissions estimates for petrochemical production have lacked quantification globally, impacting emissions reporting and decarbonization policymaking. Here we analyze cradle-to-gate emissions of 81 chemicals at 37,000 facilities worldwide, assessing 6 uncertainty sources. The results estimate a 34% uncertainty in total global emissions of 1.9 ± 0.6 Gt of CO2-equivalent emissions for 2020, and 15–40% uncertainties across most petrochemicals analyzed. The largest uncertainties stem from the inability to assign specific production processes to facilities owing to data limitations. Uncertain data on feedstock production and off-site energy generation contribute substantially, while on-site fuel combustion and chemical reactions have smaller roles. Allocation method choices for co-products are generally insignificant. Prioritizing facility-level process specification in data collection for just 20% of facilities could reduce global uncertainty by 80%. This underscores the necessity of quantifying uncertainty in petrochemical greenhouse gas emissions globally and outlines priorities for improved reporting.

Introduction

The petrochemical industry is a titan of global commerce, contributing roughly 7% of global GDP, but it is also a massive emitter of greenhouse gases. While the sector is known to be energy-intensive, global emission estimates have historically been characterized by significant uncertainty. Life Cycle Assessment (LCA) practitioners often rely on generic emission factors (EFs) because specific data on what happens inside a particular factory is often a trade secret. This lack of "process specificity"—knowing exactly which chemical route a plant uses—creates massive uncertainties that hamper effective decarbonization policy. This research seeks to move beyond "best guesses" by quantifying exactly how much uncertainty exists, where it comes from, and how we can most efficiently reduce it.

Methodology

The researchers developed a process-based LCA model to calculate cradle-to-gate emissions for 2,043 different chemical manufacturing pathways. They assessed six sources of uncertainty, categorized into two types:

  1. Data Uncertainty: Variations in data for feedstocks, indirect energy (off-site electricity/heat), direct energy (on-site fuel), and direct process emissions (chemical reactions).
  2. Model Uncertainty: The practitioner's choice of "allocation method" (how to split emissions between a main product and its co-products) and "process specificity" (the ambiguity of which process a specific facility actually uses).

To propagate these uncertainties through their global model, they used the Taylor series method to estimate variance. Finally, they mapped these processes to a massive database of 37,379 individual plants from the ICIS Supply and Demand Database, supplemented by process data from the IHS Process Economics Program (PEP) Yearbook.

Key Insights: The "Process Specificity" Challenge

A central finding is that the single largest source of uncertainty is not the chemistry or the energy data—it's the fact that we often don't know which specific process a plant is using. For example, butadiene emissions can range from 0.2 to 11 kgCO2e/kg depending on the production method. Using a "generic" average in an LCA can lead to massive errors.

Emission factor (EF) ranges for primary chemicals (a) and thermoplastics (b). The grey bars represent the range of EFs across different production processes, highlighting the impact of process specificity on uncertainty.

Interestingly, the study found that the choice of allocation method—often a point of intense debate among LCA experts—is actually insignificant compared to other sources of error for most chemicals. Furthermore, uncertainty is highly concentrated: primary chemicals (the basic building blocks) account for 70% of the total uncertainty in the sector. Because downstream products like plastics inherit about 85% of their uncertainty from these upstream inputs, fixing the data for a few "building block" chemicals like ethylene, propylene, and ammonia provides benefits that trickle down the entire value chain.

Key Results and the 80/20 Rule

Total global cradle-to-gate emissions for the analyzed chemicals in 2020 were estimated at 1.9 ± 0.6 Gt CO2e, representing an overall uncertainty of approximately 34%. Most individual chemicals showed uncertainties between 15% and 40%.

Cradle-to-gate emission factors categorized by main feedstock and emission source (FS: Feedstock, IE: Indirect Energy, DE: Direct Energy, DP: Direct Process). The charts illustrate how different feedstocks and energy inputs contribute to the overall carbon footprint and its associated uncertainty.

Among data sources, feedstock production (56%) and indirect energy use (40%) are the dominant contributors to uncertainty, while on-site fuel combustion and reaction stoichiometry play much smaller roles. Crucially, the researchers identified an "80/20 rule" for data collection: by identifying and obtaining specific process data for just the top 20% of facilities (ranked by their uncertainty contribution), global emission uncertainty could be slashed by 80%.

Limitations and Future Work

The study is limited to a "cradle-to-gate" scope, excluding the use phase (e.g., carbon released from fertilizer application) and end-of-life (e.g., plastic incineration). While 81 high-volume chemicals were covered, thousands of specialty chemicals were excluded; these likely have even higher individual uncertainties due to sparser data. Future work should aim to incorporate "system expansion" to account for how co-products displace other products in the market and address the challenge of data access, as much of the underlying industrial data remains behind expensive commercial paywalls.

Summary updated

Abstract

Abstract Uncertainties in greenhouse gas emissions estimates for petrochemical production have lacked quantification globally, impacting emissions reporting and decarbonization policymaking. Here we analyze cradle-to-gate emissions of 81 chemicals at 37,000 facilities worldwide, assessing 6 uncertainty sources. The results estimate a 34% uncertainty in total global emissions of 1.9 ± 0.6 Gt of CO 2 -equivalent emissions for 2020, and 15–40% uncertainties across most petrochemicals analyzed. The largest uncertainties stem from the inability to assign specific production processes to facilities owing to data limitations. Uncertain data on feedstock production and off-site energy generation contribute substantially, while on-site fuel combustion and chemical reactions have smaller roles. Allocation method choices for co-products are generally insignificant. Prioritizing facility-level process specification in data collection for just 20% of facilities could reduce global uncertainty by 80%. This underscores the necessity of quantifying uncertainty in petrochemical greenhouse gas emissions globally and outlines priorities for improved reporting. The dataset generated offers independent emissions factor estimates based on facility-specific information for 81 chemicals, supporting future analyses.
BibTeX
@inproceedings{cullen2024reducing,
  author = {Luke Cullen and Fanran Meng and Rick Lupton and Jonathan M. Cullen},
  title = {{Reducing uncertainties in greenhouse gas emissions from chemical production}},
  booktitle = {Nature Chemical Engineering},
  year = {2024},
  doi = {10.1038/s44286-024-00047-z},
}
Figure