Fabric production accounts for nearly two-thirds of a cotton T-shirt’s manufacturing emissions and close to 70% of its data uncertainty, a new cradle-to-gate assessment has shown. Total emissions were estimated at 1.3706 kg CO2e for a 150 g garment, with overall uncertainty of ±13.81%, equivalent to ±0.1892 kg CO2e. Weaving and spinning emerged as the most emission-intensive and uncertainty-sensitive stages within the manufacturing boundary.
- Fabric production contributed 0.8459 kg CO2e, representing 61.71% of total emissions across yarn, fabric and final assembly stages within the defined boundary.
- Yarn production added 0.3667 kg CO2e and T-shirt assembly 0.1581 kg CO2e to the cradle-to-gate footprint within the defined manufacturing system.
- The assessment applied a physical-based carbon accounting framework and a pedigree matrix to evaluate uncertainty across five data quality indicators.
- The findings were published in ‘Uncertainty Analysis of Physical-Based Carbon Accounting in Cotton T-Shirt Manufacturing’ in Scientific Reports.
THE STUDY: The paper examines how uncertainty is embedded in process-based carbon accounting for cotton garment manufacturing. It models emissions from fibre conversion to finished apparel using secondary datasets, applying a structured data-quality scoring system to test robustness across production stages. The analysis focuses on parameter reliability rather than expanding the boundary beyond manufacturing processes or redefining system limits.
- The case study draws emission and uncertainty data from India (2015) for yarn and fabric production and from China (2015) for garment assembly, while excluding agriculture, inter-stage transport and capital infrastructure.
- A pedigree matrix translated qualitative data-quality scores into quantitative uncertainty ranges across five evaluation dimensions.
- The work was authored by Emmanuel Olugbemi and Natanael Favero Bolson of the School of Engineering, University of Birmingham, UK.
BREAKING DOWN THE FOOTPRINT: Emissions are concentrated in fabric manufacturing, which contributes the largest share of greenhouse gases within the defined production boundary. Yarn processing forms the second-largest share, while final garment assembly accounts for a comparatively smaller portion. The uncertainty profile follows a similar structure, with stage-level uncertainty reported at 26.08% for yarn production, 18.59% for fabric manufacturing and 27.80% for garment assembly.
- Weaving represents the single largest process-level contributor, accounting for 48.97% of total uncertainty and nearly 71% of stage-level uncertainty within fabric production.
- Spinning dominates uncertainty within yarn processing, accounting for 25.11% of total uncertainty, while other preparatory steps contribute only marginally.
- Sewing accounts for the full uncertainty share within garment assembly, reflecting higher emission-factor sensitivity relative to underlying activity data inputs.
TECHNOLOGY VARIANTS COMPARED: Alternative production configurations shift how uncertainty is distributed across stages without materially altering the overall confidence range. Comparing ring-spun and rotor-spun yarn, alongside woven and knitted fabrics, shows only marginal changes in total uncertainty. However, the share attributable to fabric processing increases when rotor spinning and knitting are introduced.
- The ring-spun and woven baseline recorded the lowest overall uncertainty among the configurations assessed.
- Rotor-spun and knitted combinations raised the fabric stage’s share of total uncertainty above 80% in the comparative scenario.
- Despite these shifts in distribution, aggregate uncertainty moved only fractionally across the four technology variants examined.
METHODOLOGICAL CONSTRAINTS: The assessment is based entirely on secondary datasets rather than primary factory measurements, which affects representativeness across regions and technologies. The manufacturing boundary excludes cultivation, transport, machinery production, construction and maintenance of facilities and equipment, and auxiliary inputs. The pedigree-matrix method assumes parameter independence and was used rather than Monte Carlo simulation, relying partly on qualitative scoring that constrains how interaction effects are captured.
- Reliance on generic process inventories may understate site-specific variability in energy use and operational performance.
- Excluding upstream agriculture and downstream use phases distinguishes the assessment from cradle-to-grave textile studies where use-phase impacts often dominate.
- The scoring framework translates qualitative judgements into quantitative spreads, introducing subjectivity into uncertainty characterisation.
IMPLICATIONS FOR CARBON DISCLOSURE: Concentrated uncertainty in a small set of manufacturing processes affects how product carbon footprints are interpreted and compared across supply chains. Presenting results as ranges rather than single values aligns with greenhouse gas accounting guidance and reduces overconfidence in reported figures. The findings also highlight uneven data reliability within process-based approaches used for disclosure and benchmarking.
- Prioritising high-impact stages for primary data collection can improve robustness in industrial reporting and supplier engagement.
- Harmonising emission factors and scoring practices would support comparability across regions, databases and assessment frameworks.
- Greater transparency in uncertainty reporting strengthens credibility as digital product passports and disclosure rules expand.
