FLUX — Facility Load and Utilization Index

The Facility Load and Utilization Index (FLUX) grades the rigor of an organization’s carbon accounting methodology — not the carbon number itself. An organization can report a very low footprint using instruments that have no physical connection to actual renewable generation. The most common example: purchasing unbundled Renewable Energy Certificates with no temporal or geographic matching, then claiming 100% renewable status while consuming from a high-carbon grid. The reported figure may be near zero. The physical reality is the full eGRID intensity times every MWh consumed. FLUX scores the method, computes what the grid actually implies, and flags the gap.

Version

engine v0.1.0public release 2026-05-11MITchecks passing

Primary metric

flux_methodology_score — a continuous 0.0–1.0 score based on the carbon accounting method used and the quality of documentation supporting it.

Formula

methodology_score = min(1.0,
    method_base_score
    + (0.15 if rec_vintage_matched else 0)
    + (0.10 if ppa_documentation else 0))

Bonuses are additive and the total is capped at 1.0.

Carbon accounting methods

Method	Base score	Description
ppa_direct	1.0	Direct PPA — contractual link to specific generation asset
ppa_proxy	0.8	PPA with proxy generation data
marginal	0.7	Time-matched marginal emission rates
grid_average	0.5	Annual average eGRID factor — honest but imprecise
rec_bundled	0.4	Bundled RECs — weak temporal matching
rec_unbundled	0.2	Unbundled RECs — no temporal or geographic match
unknown	0.0	No methodology disclosed

Sources: WRI GHG Protocol Scope 2 Guidance 2015, Rocky Mountain Institute — The Emissions First Principle 2022, RE100 Technical Criteria 2023.

Bonuses

Bonus	Value	Condition
rec_vintage_matched	+0.15	RECs matched to the same year as consumption
ppa_documentation	+0.10	PPA contract documentation provided

Letter grade

Grade	Condition	Label
A	score ≥ 0.85	Excellent
B	score ≥ 0.65	Good
C	score ≥ 0.45	Adequate
D	score ≥ 0.25	Poor
F	score < 0.25	Failing

Worked examples

Direct PPA with documentation:

carbon_method:     ppa_direct    → 1.0 base
ppa_documentation: true          → +0.10
rec_vintage_matched: false

methodology_score = min(1.0, 1.0 + 0.10) = 1.0
Grade: A (Excellent)

Marginal with vintage-matched RECs:

carbon_method:     marginal      → 0.70 base
rec_vintage_matched: true        → +0.15

methodology_score = min(1.0, 0.70 + 0.15) = 0.85
Grade: A (Excellent — at the threshold)

Unbundled RECs:

carbon_method:     rec_unbundled → 0.20 base
ppa_documentation: false
rec_vintage_matched: false

methodology_score = 0.20
Grade: F (Failing)

eGRID carbon intensity

FLUX computes an estimated actual carbon footprint from the facility’s total energy consumption and EPA eGRID 2023 grid intensity for its subregion:

actual_carbon_kg = total_energy_mwh × egrid_intensity_kg_mwh
carbon_delta_kg  = actual_carbon_kg - claimed_carbon_kg

A positive carbon_delta_kg means the organization’s claimed footprint is below what the grid implies.

Selected eGRID subregions (kg CO2/MWh):

Subregion	Intensity	States
MROW	386	MN, WI, IA, IL
RMPA	617	CO, WY
ERCT	386	TX
FRCC	430	FL
NPCC	195	NY, MA, CT
SERC	394	GA, NC, SC, TN, AL
WECC	287	CA, WA, OR, NV, AZ
RFCE	285	PA, NJ, MD, DE, DC
RFCW	500	OH, IN, WV, KY
HICC	650	HI
AKGD	493	AK

Source: EPA eGRID 2023, Summary Tables, Table 1.

Full subregion coverage and international grid intensities in flux/coefficients/flux_coefficients.json.

International support

For non-US facilities, supply the egrid_subregion field directly with an international grid code:

Code	Grid	Intensity (kg CO2/MWh)
EU_AVG	EU-27 average	276
DE	Germany	385
FR	France	58
NL	Netherlands	298
SE	Sweden	13
NO	Norway	7
GB	United Kingdom	207
CN	China national	537
JP	Japan	453
CA_ON	Canada Ontario	29
SG	Singapore	408
IN	India	708

Sources: ENTSO-E Transparency Platform 2023 (Europe), IEA World Energy Outlook 2023, China MEE National Carbon Emissions Report 2022.

Greenwashing detection

FLUX flags greenwashing when all three conditions hold:

1. carbon_method is rec_unbundled or rec_bundled
2. claimed_renewable_pct > 0
3. claimed_carbon_kg < actual_carbon_kg × 0.50

The flag is a separate finding (greenwashing_flag = 1.0). It does not change the methodology score — the score already reflects the weakness of the REC method. The flag surfaces in the GRADE assessment report with the reason string, which includes the claimed percentage, the actual grid-implied carbon, and the claimed carbon.

The greenwashing check does not apply to grid_average, marginal, ppa_proxy, or ppa_direct — those methods do not involve certificate-based accounting.

Findings exported to GRADE

Metric	Unit	Description
flux_methodology_score	score_0_to_1	Carbon accounting quality score — GRADE input
carbon_accounting_grade	grade_points_0_to_4	A=4, B=3, C=2, D=1, F=0
actual_carbon_kg	kg_CO2	total_energy_mwh × egrid_intensity
claimed_carbon_kg	kg_CO2	Organization self-reported footprint
carbon_delta_kg	kg_CO2	actual_carbon_kg − claimed_carbon_kg
greenwashing_flag	boolean_as_float	1.0 if flagged, 0.0 if not
egrid_intensity_kg_mwh	kg_CO2_per_MWh	Grid intensity used for estimate
carbon_method_code	method_code_0_to_6	Integer code for ATLAS recommendation routing

What FLUX rewards

Traceability. Can the organization document that the renewable power it claims corresponds to actual generation delivered at the time its cluster was running? A direct PPA with a generator in the same grid region is traceable. Unbundled RECs purchased from another region after the fact are not — they are a financial instrument, not a physical connection to renewable power.

The methodology quality score rewards progressively better traceability: from no accounting (0.0) through grid averages (0.5) to time-matched marginal rates (0.7) to direct PPAs (1.0).

Input schema

Field	Type	Required	Description
organization	string	yes	Organization name
period	string	yes	Reporting year (YYYY)
location_state	string	yes	2-letter US state abbreviation, or "" for international
total_energy_mwh	float	yes	Gross facility energy consumption
carbon_method	string	yes	One of the seven method enum values above
claimed_renewable_pct	float	yes	Claimed renewable percentage (0–100)
claimed_carbon_kg	float	yes	Organization self-reported carbon footprint
egrid_subregion	string	no	Override state lookup — use for international facilities
ppa_documentation	bool	no	PPA contract documentation provided
rec_vintage_matched	bool	no	RECs matched to the same year as consumption
utility_provider	string	no	Utility provider name (for display)

CLI usage

# Score a single facility from a flux_input_v1.json file
flux analyze \
  --input flux_input.json \
  --output flux_output.json

# Run all five synthetic organizations and print summary table
flux demo

# Validate a ptl_output_v1.json file against the PTL schema
flux validate --input flux_output.json

Example flux_input.json:

{
  "organization": "Midwest Research Computing",
  "period": "2026",
  "location_state": "MN",
  "total_energy_mwh": 18500,
  "carbon_method": "ppa_direct",
  "claimed_renewable_pct": 100,
  "claimed_carbon_kg": 0,
  "ppa_documentation": true
}

International facility:

{
  "organization": "European AI Institute",
  "period": "2026",
  "location_state": "",
  "egrid_subregion": "FR",
  "total_energy_mwh": 12000,
  "carbon_method": "grid_average",
  "claimed_renewable_pct": 0,
  "claimed_carbon_kg": 696000
}

Source

• View FLUX source on GitHub
• FLUX methodology document
• Report an issue