Skip to content

methodology.md

Latest commit

 

History

History
166 lines (127 loc) · 7.72 KB

methodology.md

File metadata and controls

166 lines (127 loc) · 7.72 KB
Raw

Methodology & Limitations

This document describes how the CE services compute their results, which values are drawn from published sources, and — just as importantly — which are illustrative assumptions. It is intended to make the system reproducible and to support an honest description in any paper or report that uses it.

One-line summary: the quantitative services use published, cited emission factors and standards-based dimensions, combined with deliberately simple, configurable model coefficients. The system is a transparent demonstrator, not a certified ISO 14067 / ISO 14040 LCA tool.

1. Product Carbon Footprint (calculate_carbon_footprint)

Structure

The footprint follows the standard cradle-to-gate product-carbon-footprint stage decomposition (cf. ISO 14067, ISO 14040/44, GHG Protocol Product Standard):

materials_CO2 = Σ (emission_factor[material] × mass_fraction × weight_kg)
manufacturing = materials_CO2 × 0.18
transport     = 0.25 × weight_kg × region_factor
total         = materials_CO2 + manufacturing + transport

mass_fraction is expected on a 0–1 scale, but the composition may be supplied as percentages instead (values summing to ~100); the service detects this and normalizes to fractions, so a caller can use either convention. The circularity service applies the same tolerance to its percentage and 0–10 score inputs.

Emission factors — published & cited

Per-material cradle-to-gate GWP-100a factors (kg CO₂e/kg) are defined, with a source tag on every entry, in ce_services/constants.py:

Group Source
Metals (Al, Fe→steel, Cu, Ni, Co, Li, Ti, Ag, Au) Nuss & Eckelman (2014), PLOS ONE — Table S38, GWP-100a, supply-mix-weighted primary production, GLO
Plastics & common materials (glass, paper, board, rubber, timber, textile, ceramic, HDPE/PP/PVC/ABS/PC/nylon/epoxy) ICE database v2.0 (Hammond & Jones, Univ. of Bath)
Steel worldsteel LCI — global average crude steel ≈ 1.9 kg CO₂e/kg
PET NAPCOR / U.S. LCI (2020) ≈ 2.23 kg CO₂e/kg

Two deliberate source choices are documented in the code: titanium uses the titanium-metal figure (45.1) rather than the pigment-weighted supply-mix total (8.1), because the tool models metal parts; and PET uses NAPCOR/PlasticsEurope rather than ICE's anomalous carpet-fibre entry (5.56).

Illustrative coefficients — not from a source

The following are reasonable order-of-magnitude assumptions for demonstration and are exposed for tuning. They should not be cited as measured values:

Coefficient Value Status
Manufacturing energy overhead 18% of material CO₂ assumption
Base transport intensity 0.25 kg CO₂/kg assumption
Regional transport multipliers 1.0–1.8 relative, illustrative
Industry-average benchmark 4.5 kg CO₂e/kg assumption
A–E rating thresholds 2.0 / 3.5 / 5.5 / 8.0 kg CO₂e/kg illustrative bands
Recycled-material saving 40% of material CO₂ assumption

Factors are primary-production averages and do not adjust for a product's actual recycled content (recycled metals/plastics are typically far lower).

2. Circularity Indicator (calculate_circularity_indicator)

Structure

A composite index on 0–1, a weighted sum of four normalised dimensions:

CI = 0.30·recycled_content + 0.35·recyclability + 0.20·durability + 0.15·repairability

Dimensions — grounded in standards

The choice and assessment of each dimension map to the CEN-CENELEC material-efficiency standards (CEN-CLC/JTC 10) for the EU Ecodesign framework:

Dimension Standard
Recycled content EN 45557:2020
Recyclability / recoverability EN 45555:2019
Durability EN 45552:2020
Repairability (repair/reuse/upgrade) EN 45554:2020
Terminology CLC/TR 45550:2020

Assumptions — not standardised

No EN 4555x standard prescribes how to aggregate the dimensions into one index. The following are author-defined and configurable (see CI_WEIGHTS, GRADE_THRESHOLDS, REFERENCE_INDEX in ce_services/circularity.py):

  • the weights (0.30 / 0.35 / 0.20 / 0.15);
  • the A–E grade thresholds;
  • the reference index (0.42) used for the comparison string. This is an illustrative value, not an official figure. There is no published 0–1 product-level "EU average circularity index." The closest official EU metric is the Circular Material Use Rate (Eurostat cei_srm030, 11.5% in 2022), which is an economy-wide material-flow share and is not comparable to this product composite.

This composite is distinct from the Ellen MacArthur Foundation Material Circularity Indicator (MCI), which is defined as a Linear Flow Index modified by a utility factor — a different formula, not a weighted sum.

CE data points. The product/material inputs these services consume are a small, demonstrator-level subset. The production-level, AAS-mapped catalog of Circular Economy data points is maintained separately by DFKI: Datapoints-for-Circular-Economy.

3. Other services

  • CE Data Extraction (extract_ce_data) — a rule-/keyword-based parser that pulls CE-relevant fields (materials, recycled content, recyclability, certifications, hazardous substances, end-of-life) from unstructured text. It is a heuristic information-extraction demonstrator, not a validated NLP model.
  • Material Reuse Potential (analyze_material_reuse) — returns reuse scores, pathways, value-recovery percentages and CO₂ savings from a curated per-material lookup table. The tabulated values are illustrative reference figures.
  • Service Registry (get_available_services) — metadata only; no calculation.

4. Limitations (summary)

  1. Calculations are simplified models intended to demonstrate the agent/MCP architecture, not to deliver certified LCA results.
  2. Emission factors are cited cradle-to-gate primary-production averages; they ignore product-specific recycled content, use phase, and end-of-life.
  3. Several model coefficients (Section 1) and the circularity aggregation (Section 2) are transparent assumptions, not standardised or empirically fitted.
  4. The circularity reference value and any "benchmark" comparisons are illustrative.
  5. The extraction and reuse services are heuristic/lookup-based.

All assumptions are isolated in named constants so they can be replaced with project-specific or empirically calibrated values.

References

  • Nuss, P. & Eckelman, M. J. (2014). Life Cycle Assessment of Metals: A Scientific Synthesis. PLOS ONE 9(7): e101298. https://doi.org/10.1371/journal.pone.0101298
  • Hammond, G. & Jones, C. (2011). Inventory of Carbon & Energy (ICE) Database, v2.0. University of Bath / Circular Ecology. https://circularecology.com/embodied-carbon-footprint-database.html
  • World Steel Association — Life Cycle Inventory (LCI) data and eco-profiles. https://worldsteel.org/
  • NAPCOR — Virgin PET life-cycle inventory (U.S. LCI, 2020 revision).
  • CEN-CENELEC EN 45552:2020, EN 45554:2020, EN 45555:2019, EN 45557:2020; CLC/TR 45550:2020. (CEN-CLC/JTC 10, material efficiency for ecodesign.)
  • Eurostat — Circular material use rate (cei_srm030).
  • Ellen MacArthur Foundation & Granta Design — Circularity Indicators: Methodology (Material Circularity Indicator).
  • ISO 14067:2018; ISO 14040:2006 / ISO 14044:2006; GHG Protocol Product Standard.
  • DFKI — Datapoints for Circular Economy (production-level, AAS-mapped CE data points). https://github.com/DFKI/Datapoints-for-Circular-Economy