WATTCARBON
WATTCARBON
GS

Grid Scores &
Carbon Scores

Evaluating DER impact across projects, types, and geographies.

OpenEAC Alliance Meeting  ·  June 2026
McGee Young
WattCarbon
Steve Suffian
WattCarbon

Agenda

Where we're headed today

01

From Capacity to Scores

Why we need standardized impact metrics beyond kWh and kW

~5 min
02

Carbon Score

Measuring emissions impact using average and marginal grid intensity

~7 min
03

Grid Score

Six $/kW-year value streams measuring DER contributions to the grid

~10 min
04

Questions for the Alliance

Capacity measurement, units, and the six-element framework

~5 min
01

Section One

From Capacity
to Scores

Bridging hourly savings to comparable impact metrics.

Where We Left Off

Last time: DER capacity requires hourly M&V

Now that we can measure hourly savings, how do we evaluate and compare the impact of different DER projects?

The Scoring Problem

A battery in Texas and a heat pump in New England —
how do you compare?

Battery Storage

ERCOT
DISCHARGE + CHARGE –
  • 50 MWh displaced annually
  • Dispatches during summer peaks
  • Short-term baseline

Heat Pump Retrofit

ISO-NE
WINTER HEATING LOAD REDUCED
  • 120 MWh displaced annually
  • Reduces winter heating load
  • Long-term counterfactual

Which project delivers more grid value? More carbon value? The answer depends on when the savings happen and what the grid needed at that moment.

The Framework

Two families of scores

Carbon Score

How much did this project reduce grid emissions?
  • Multiply hourly savings by hourly emissions intensity
  • Two variants: average emissions, marginal emissions
Units — kg CO₂/hour (or tCO₂/MWh if normalized by savings magnitude)

Grid Score

What is the annualized dollar value of this project's grid contributions?
  • Six value streams, each denominated in $/kW-year with technology-specific coincidence factors
  • Energy arbitrage, congestion relief, capacity/RA, ancillary services, tariff savings, T&D deferral
Units — $/kW-year

Both start from the same input: hourly energy savings from a standardized methodology.

02

Section Two

Carbon Score

Emissions impact using grid intensity data.

Carbon Score — Methodology

How it works

1
Start with hourly savings — counterfactual minus observed energy for each hour
2
Obtain hourly emissions intensity — from grid data providers (EPA eGRID, WattTime, Singularity, EIA)
3
Multiply each hour's savings by that hour's intensity — yields hourly carbon impact (+ avoided, − increased)
4
Average the hourly carbon impacts = Carbon Score
Carbon Score
CS = mean(
  (counterfactual − observed)
  × emissions_intensity
)
for all hours h in the evaluation period  ·  result in kg CO₂/hour

Carbon Score

Two lenses on carbon impact

Average Emissions Intensity

Carbon footprint of electricity displaced
  • Average carbon intensity of all generation on the grid that hour
  • Data: EPA eGRID, EIA, some ISO fuel-mix feeds
  • Easier to obtain; available for most regions
  • Better for accounting — attributing emissions to consumption

Marginal Emissions Intensity

Emissions actually avoided
  • Which generator would ramp up or down in response to this change
  • Data: WattTime, Singularity, REsurety, Cambium (NREL)
  • Harder — requires dispatch modeling or statistical inference
  • Better for impact — the causal effect on emissions

We compute both. They often tell different stories — a project in a coal-heavy region may have a high average score but a moderate marginal score if the marginal generator is gas.

Carbon Score

Carbon Score in practice

kg CO₂ +5.0+6.02pmsavings +20 kWh+40.0+52.06pmsavings +80 kWh-12.0-15.02amsavings −30 kWh Average emissions Marginal emissions
Hourly carbon impact = savings × emissions intensity. The 2am bars go negative — the battery is charging.
HourSavingsAvg impactMarg. impact
2pm+20 kWh+5.0 kg+6.0 kg
6pm+80 kWh+40.0 kg+52.0 kg
2am−30 kWh−12.0 kg−15.0 kg
Average11.0 kg/hr14.3 kg/hr

The 6pm hour is worth more in both lenses — the grid is dirtier and the savings are larger. The 2am penalty hurts less because the grid is cleaner at night.

03

Section Three

Grid Score

Six $/kW-year value streams measuring DER contributions to the grid.

Grid Score

Previous Iteration of Grid Score

8,760 hourly savings values · one bar ≈ 73 hours stress hours — filtered into the numerator
Take one year of hourly savings, keep only the stress hours, sum them, divide by all positive savings.
1
Start with hourly savings (same input as Carbon Score)
2
Define "stress hours" using an external signal specific to that score
3
Filter savings to those hours, then sum = numerator
4
Divide by Total Displaced Electricity = sum of all positive-savings hours
Grid Score
GS =
  Σ( savings )  for h in stress_hours
  ÷ Total Displaced Electricity
Total Displaced = Σ savingsₙ for all h where savingsₙ > 0  ·  no weighting term

A Grid Score of 0.30 means 30% of the project's total displaced electricity occurred during grid stress hours. Higher is better. All six scores are pure, unweighted kWh ratios.

Grid Score — The General Pattern

How Grid Scores work

1
Geo-assign the asset — resolve to Balancing Authority, pricing node, tariff zone, AS region, or feeder (each stream uses a different geographic level)
2
Look up the stream's $/kW-year benchmark — from market data (LMP, BRA, AS clearing prices), tariff schedules, or NWA filings
3
Apply a technology-specific coincidence factor — the fraction of theoretical value this DER archetype realistically captures
4
Scale by scenario — low/mid/high escalators act as error bars (e.g. 0.85 / 1.00 / 1.25)

Data availability varies significantly by geography. Wholesale-market scores require ISO data. Capacity/RA is broadly available via EIA-930. Tariff savings is computable everywhere. T&D deferral is the most constrained — limited to CA and NY for distribution-level data.

Grid Score (per stream)
Value = $/kW-yrbenchmark
  × coincidence_factor(tech, mode, stream)
  × capacitykW
  × scenario_factor

A Grid Score of $60/kW-yr for Capacity/RA means each kW of DER capacity delivers $60/year of capacity value to the grid. The coincidence factor adjusts for how well the technology actually captures that value. Higher is better.

Scenario escalators

Low / mid / high (e.g. 0.85 / 1.00 / 1.25) are error bars around the medium estimate, not distinct scenarios. They capture uncertainty in load growth and market evolution.

Grid Score

Six dimensions of grid impact

Grid ScoreWhat It Measures$/kW-yr Data SourceGeographic Level
Tariff SavingsRetail rate structure alignmentUtility tariff schedules (TOU, demand charges)Utility service territory
Energy ArbitrageValue of shifting energy across timeEnergy component of LMPPricing node or zone
Congestion ReliefWholesale market congestion reliefCongestion component of LMPPricing node or zone
Capacity / RAPeak demand contributionRA clearing price or BRA auction resultsBalancing Authority
Ancillary ServicesGrid balancing contributionComposite AS clearing priceZone or system-wide
T&D DeferralLocal infrastructure reliefNWA filings, LNBA, E3 ACC fallbacksTransmission node / feeder

Each score is denominated in $/kW-year — the annualized value a 1 kW DER delivers through that stream. The coincidence factor for each technology adjusts for realistic operating constraints. A rooftop solar project might score well on energy arbitrage but poorly on capacity/RA — the multi-dimensional view is the point.

Grid Score — Eligibility & Coincidence Factors

Not every DER captures the same value

Technology archetypes & control modes

Each DER is classified by technology (battery, HVAC, smart-thermostat, solar+storage) and control mode (wholesale-bid, self-consumption, demand-response). The combination determines which value streams the project is eligible for and how much of each stream it captures.

What is a coincidence factor?

The fraction of a stream's theoretical value that the archetype captures under realistic operating constraints. Not the fraction of theoretical capacity — the fraction of theoretical value. A factor of 0.00 means the project is ineligible for that stream.

Coincidence factors vary by stream — a battery might be 0.90 for energy arbitrage but 0.60 for ancillary services (must reserve state-of-charge headroom). A wholesale-bid battery gets 0.00 for tariff savings (no retail exposure).

Example: Capacity / RA coincidence factors
TechnologyControl ModeFactorCitation
batterywholesale-bid0.85CAISO 2024 RA QC; PJM ELCC 4-hr derate
batteryself-consumption0.00Not counted towards capacity
hvacdemand-response0.70Internal est.; FERC-201 participation
smart-thermostatdemand-response0.40Internal est.; seasonal availability
solar+storagewholesale-bid0.80NREL ATB 2024; coupled capacity credit

Grid Score

Settlement vs. Forecast

Two modes for computing grid value. Forecasting uses coincidence factors to estimate what a DER will deliver.

Settlement (historical)

value = Σ( actual_price × actual_savings )
  • Granularity: location × hour
  • Observed market data × metered savings
  • “What was this DER worth last month?”

Forecast (forward-looking)

value = $/kW-yrbenchmark × coincidence × capacity
  • Granularity: location × quarter
  • Deemed loadshapes × benchmark prices × coincidence factors
  • “If I built a battery here, what would it be worth?”
Which streams use which mode?
StreamSettlementForecast
Tariff SavingsMetered actualsDeemed loadshape × coincidence
Energy ArbitrageHourly actualsDeemed loadshape × coincidence
Congestion ReliefHourly actualsDeemed loadshape × coincidence
Capacity / RAEntirely deemed (settlement = forecast)
Ancillary ServicesEntirely deemed (settlement = forecast)
T&D DeferralEntirely deemed (settlement = forecast)

Deemed streams always use coincidence factors. Hourly-settled streams only use them in forecast mode — settlement has real data instead.

Grid Score — Component 1

Tariff Savings

How much a DER reduces a customer's retail electricity bill.

Settlement
value = Σ( counterfactual − observed ) × tariff
Forecast
value = deemed_loadshape × tariff_schedule × coincidence
Key points
  • Geography: Utility service territory
  • Captures TOU differentials, demand charges, and critical peak pricing
  • Wholesale-bid batteries get 0.00 — no retail tariff exposure
  • Self-consumption batteries: 0.80 (demand-charge avoidance)
Data availability
  • All regulated utilities — tariff schedules are public; most universally computable score
  • OpenEI — U.S. Utility Rate Database covers >3,000 utilities
  • Critical peak pricing — event calendars vary by utility

Unique challenge

Tariffs vary enormously by utility, rate class, and vintage — the score must reference each project's specific applicable tariff.

Grid Score — Components 2 & 3 (LMP-based)

Energy Arbitrage & Congestion Relief

Both derived from LMP decomposition: LMP = Energy + Congestion + Losses. Each component is scored separately.

Energy Arbitrage
Settlement: Σ( energy_LMP × savings )
Forecast: energy_LMP × deemed_loadshape × coincidence
  • Uses energy component only, not full wholesale price
  • Value concentrates in evening ramp and solar oversupply periods
Congestion Relief
Settlement: Σ( congestion_LMP × savings )  where > 0
Forecast: congestion_LMP × deemed_loadshape × coincidence
  • Uses congestion component only; negative hours set to 0
  • A node can be low on arbitrage but high on congestion
ISO Region — Pricing Nodes congested $32 $35 $30 $34 $31 $78 $65 DER projects Low congestion High congestion
Same LMP, two scores. The congested node ($78) has high congestion value; the $32 node has mostly energy value. DERs near congested nodes score higher on congestion relief.

Grid Score — Component 4

Capacity / Resource Adequacy

The value of committing dispatchable capacity to meet peak demand.

Entirely deemed (settlement = forecast)
value = $/kW-yrclearing × nameplate × coincidence × scenario
Key points
  • Geography: Balancing Authority
  • Source: CAISO RA (CPUC), PJM BRA auction clearing
  • Requirement: device must prove dispatchability to grid operator
  • Scenarios: low 0.85 · mid 1.00 · high 1.25

How it works

The DER commits a fixed capacity (kW) to the grid operator. The clearing price sets $/kW-yr for that commitment. The coincidence factor derate reflects how much of nameplate the technology can reliably deliver (e.g. a 4-hr battery can't sustain output through an 8-hr peak).

No hourly settlement

Unlike energy or congestion scores, there are no hourly price signals. Value is fixed at the auction clearing price × the promised capacity. This is a contract-based value stream, not a market-price-based one.

Grid Score — Component 5

Ancillary Services

The value of providing frequency regulation and reserves to keep the grid balanced.

Entirely deemed (settlement = forecast)
value = avg_AS_price$/kW-yr × coincidence × nameplate × scenario
$/kW-yr = $/MW-hr × 8760 / 1000  (sum across reg + reserves)
AS products
  • Frequency regulation — seconds timescale
  • Spinning reserves — respond within 10 min
  • Non-spinning reserves — respond within 10–30 min
  • Emerging: uncertainty reserves (NYISO 2025)

Only batteries on wholesale-bid participate (all others = 0.00 coincidence)

AS regions vary by ISO

  • CAISO — 3 regions
  • MISO — 7 regions
  • NYISO — system-wide + NYC/LI locational
  • PJM, ERCOT, SPP — primarily system-wide

Grid Score — Component 6

T&D Deferral

The value of avoiding or deferring utility transmission and distribution infrastructure investment.

Entirely deemed (settlement = forecast)
value = $/kWNWA filing × nameplate × coincidence
Key points
  • Geography: defined by NWA RFP (transmission node / feeder)
  • Value = avoided capital expenditure on T&D infrastructure
  • Fallbacks: E3 ACC (California), Synapse PJM study
  • vs. Congestion Relief: congestion uses wholesale price signals; T&D Deferral is about physical infrastructure limits

Two distinct layers

Transmission — visible through LMP spreads; requires mapping DER location to relevant constraint

Distribution — feeder/substation congestion via hosting capacity analysis

Data availability
  • CA, NY (distribution) — available via hosting capacity maps
  • ISO regions (transmission) — partially derivable from LMP patterns
  • Most other regions — no public feeder-level or constraint data
  • Non-ISO regions — unavailable

The most locally valuable score, and the hardest to compute. NWA programs in CA and NY already use similar logic.

Putting It Together

What a multi-score DER evaluation looks like

Project: 500 kW Battery Storage (wholesale-bid), PJM (Northern Virginia)  ·  Period: January 2026  ·  Scenario: mid

Grid Value Breakdown ($/kW-yr, post-coincidence) $20.24 $13.50 $9.25 $8.98 $7.00 Ancillary Services Energy Arbitrage T&D Deferral Capacity / RA Congestion Relief Tariff Savings: $0 Total grid value: ~$59/kW-yr = ~$29,500/yr for a 500 kW system at scenario mid
ScoreBenchmarkCoinc.$/kW-yr
Carbon (Average)0.38 tCO₂/MWh
Carbon (Marginal)0.52 tCO₂/MWh
Ancillary Services$33.730.60$20.24
Energy Arbitrage~$150.90~$13.50
T&D Deferral$18.500.50~$9.25
Capacity / RA$10.560.85~$8.98
Congestion Relief~$100.70~$7.00
Tariff Savings0.00$0

This battery's grid value is dominated by ancillary services and energy arbitrage. Tariff savings is zero because wholesale-bid batteries have no retail tariff exposure. The stacked view shows where the value comes from — buyers weight streams by what they care about.

Questions for the Alliance

What we need your input on

1

How do you think about measuring capacity?

What does “capacity” mean for different DER types? How should we handle resources that can’t sustain output for extended periods?

2

Does $/kW-year make sense as a unit?

Or should grid scores use a unit closer to energy (e.g. $/MWh, $/kWh)? What are the trade-offs for different market participants?

3

Do these six elements track for measuring grid impact?

Are there value streams missing? Are any of these redundant or better combined? Does this decomposition match how you think about DER value?

In Closing

Scoring the grid impact of DERs

Hourly savings are the foundation. Carbon Scores measure emissions impact. Grid Scores measure the $/kW-year value a DER delivers across six value streams — making grid contributions meaningful, comparable, and actionable for different buyers.

Next Steps

openeac.org   ·   methods.openeac.org
A

Reference

Appendix

Formula reference, data sources, and the NREL ELCC research.

← → navigate  ·  N — speaker notes
Speaker Notes — Slide 1