Foreword

This specification standardizes how persistent, daily load shifting (e.g., EV charging schedules, HVAC pre-cool/pre-heat, thermal storage control, commercial process rescheduling) is measured and certified using whole-building interval data. It is designed for programs that value when electricity is used, not only how much.

About the OpenEAC Alliance

The OpenEAC Alliance develops open, auditable methodologies for distributed energy resources (DERs) to enable trustworthy Energy Attribute Certificates (EACs). This document aligns with the Alliance’s emphasis on clarity, reproducibility, and market interoperability.

Introduction

This document specifies a statistically rigorous method to quantify recurring, schedule-based load shifting using whole-building meter data. The assignment of load reductions to load increases is inspired by the EnergyTag Storage Allocation Methods, particularly FIFO.

Normative References

• OpenDSM, CalTRACK 2.0 (EEMeter) — daily/interval M&V framework

  • Used in CAISO

• LBNL Time-of-Week and Temperature (TOWT) model

• ASHRAE Guideline 14 — Measurement of Energy and Demand Savings and IPMVP Option C

• EnergyTag Granular Certificate Scheme Standard

Terms and Definitions

• Intervention Date: The date in which deliberate load shifting began. The baseline period is defined in relation to this date.

• Baseline Period: The period before the intervention date, with meter data to be used for fitting a model.

• Measurement Windows: Program-defined hours of the day targeted for measurement. Given additional constraints (such as by day of week or season), these could alternatively be defined by hour-of-week or with specific timestamps. In the case of dynamic windows (such as load shifting based on price signals), an ongoing daily schedule of peak and off-peak windows must be provided throughout the measurement period.

  • Peak Window (P): Program-defined hours targeted for reduction (e.g., 16:00–21:00 local).

  • Off-Peak Window (O): Program-defined hours acceptable for increase (e.g., 00:00–06:00 or 10:00–15:00).

• Counterfactual: Predicted hourly kWh absent the schedule change.

• Increased Consumption Record (ICR): An amount of additional Wh that was estimated to have been consumed at a particular hour within an off-peak window.

• Reduced Consumption Record (RCR): A reduction in the amount of Wh estimated to have been consumed at a particular hour within a peak window. An RCR must be paired with a valid ICR.

Scope

This methodology supports ongoing load-shifting delivered by consuming less energy during one period and additional energy during another period relative to a baseline period. The following are IN SCOPE:

• Time-shifting load from program-defined peak windows to off-peak windows (such as EV charging moved to overnight; HVAC pre-cool shifting cooling load from 4–9 pm to midday).

• Measurements can be conducted using whole-building meter data; submeter/device methods may be used where available but are not required.

• Impacts are meant to occur on a regular basis, windows SHOULD take place at least once per week.

The following are OUT OF SCOPE for this method:

• One-time “event” DR only (see event-based DR spec).

• Permanent load reductions that change total energy unrelated to time shifting (treat as efficiency).

• Frequency regulation or services requiring sub-minute telemetry.

Data Requirement Definitions

• Electricity Interval Data: Hourly (or sub-hourly aggregated to hourly) whole-building Wh.

  • Baseline Data Sufficiency: ≥90% hourly intervals in baseline and ≥90% in each reporting month.

• Weather Data: Hourly dry-bulb temperature representative of the asset’s location.

• Peak/Off-Peak Window Schedule (Optional): A timeseries spanning the measurement timeseries defining dynamic peak and off-peak windows. This is only needed if the windows are not pre-defined.

Methodology

1 Construct a baseline period

1.1 The counterfactual SHALL be trained on a baseline period ending prior to the intervention date.

1.1.1 This baseline period SHOULD be 365 days, particularly for devices with seasonal operations (such as HVAC). If the load shifting derives from a device that isn’t seasonally operated, then it MAY be shortened to as little as 56 days, but should be as long as data availability allows.

1.1.2 Holidays and known atypical schedules MAY be excluded from training if they degrade model stability.

2 Fit a baseline model (Time-Of-Week and Temperature)

2.1 Fit an energy prediction model that incorporates

2.1.1 Temporal patterns (hour-of-day, day-of-week, monthly)

2.1.2 Temperature via segmented splines or bins (cooling and, if present, heating sensitivities).

2.1.3 Solar Irradiation (for locations with behind-the-meter PV).

2.2 Compute CVRMSE of the baseline model evaluated on the baseline data.

3 Use the model to predict hourly counterfactual consumption for each hour in each window.

4 Calculate savings by assigning off-peak window increases to peak window savings following a FIFO framework

4.1 Off-Peak Window: Generate an ICR for each hour where there was increased consumption in the off-peak window.

4.1.1 For each hour h, ICRh = max(0, observed_whh - counterfactual_whh).

4.2 Peak Window: Generate an RCR for each hour where there was decreased consumption in the peak window.

4.2.1 For each hour h, RCRh = max(0, counterfactual_whh - observed_whh).

4.3 Match each RCR to an ICR to create a savings record. ICRs must have the following relationship to the RCR they are matched to:

4.3.1 An ICR must have taken place between 0-24 hours preceding the RCR.

4.3.2 Different Wh from the same ICR can be assigned to different RCRs. Multiple ICRs can be assigned to the same RCR.

4.3.3 Savings records are composed only of the amount of Wh of RCRs that have been matched to ICRs. Any RCRs with no valid ICRs to match are discarded and not counted towards savings.

Model Disqualification

• Single-site eligibility: CVRMSE ≤ 1.0. Sites exceeding the threshold SHALL be excluded from savings claims.

• Portfolio target: Aggregated portfolio SHOULD have a CVRMSE ≤ 0.25.

Appendix

Appendix 1: Consumption Records Example

The off-peak window for a given location is defined as 10 AM - 12 PM. The peak window is defined as 5 PM - 7 PM.

For Day X, the following consumption values are calculated, counterfactual predictions are made, and consumption records are generated:

Hour	Counterfactual	Observed	ICRs Generated	RCRs Generated
10 AM	10 Wh	12 Wh	2WH@10AM
11 AM	10 Wh	13 Wh	3WH@11AM
5 PM	4 Wh	1 Wh		3WH@5PM
6 PM	5 Wh	1 Wh		4WH@6PM

To calculate savings, we go through each RCR and match ICRs to it.

3WH@5PM:

• We start with the earliest valid ICRs generated. The earliest record is 2WH@10AM. Since this record is within 24 hours of the ICR, it is considered valid. The full set of Wh from the RCR is assigned to the ICR. After this process, there are 3 - 2 = 1 Wh left to fill in the RCR.

• We next look at the subsequent ICR, which is 3WH@11AM. We assign 1 of these Wh to this RCR, meaning that the remaining ICR is reduced to 2WH@11AM.

4WH@6PM:

• We check for the earliest valid ICRs generated. There is 2WH@11AM left. This is within 24 hours of this ICR, so it is considered valid. All of these Wh from the ICR can be assigned to this RCR, so there are 4 - 2 = 2 Wh left for this RCR. Since there are no remaining valid ICRs, these 2 Wh are discarded.

The savings calculated for these hours are:

Hour	RCRs	ICRs At Start	ICRs at End	Savings
10 AM	-	-	2WH@10AM	-
11 AM	-	2WH@10AM	2WH@10AM, 3Wh@11AM	-
5 PM	3WH@5PM	2WH@10AM, 3Wh@11AM	2WH@11AM	3 Wh
6 PM	4WH@6PM	2WH@11AM	-	2 Wh