Demand Response Measurement & Forecasting

Data Inputs & Configurations Guide

---

About This Guide

This guide is designed to help program administrators understand what’s required to enable the FLEX Program’s Demand Flexibility (DR) Measurement and optional Event-Based Forecasting capabilities. It explains the data inputs needed, the configuration options available, and how these elements align with your program objectives, regulatory requirements, and your organization’s operational priorities.

Measurement Overview

Confidence in Event Performance

FLEX Programs’ Measurement capability provides a clear, verified view of how demand response events actually performed. Using interval-level metering data, the platform calculates true and observed load reductions for every event at every meter, delivering near real-time insights into performance across providers, technologies, and grid nodes.

By accounting for non-participation effects through Recurve’s open, industry-leading modeling framework, Measurement produces a complete and defensible representation of each dispatch’s impact that can be leveraged across grid operations, procurement, and regulatory reporting.

Using Measurement, you’ll know:

• How much load was reduced at each interval and in total

• Where reductions occurred geographically or by feeder

• Which customer segments, technologies, or aggregators drove the most impact

• How performance trends evolve across the program season, including fatigue effects.

• The total impact of an event across the day, such as energy reduction, pre-event behaviors, and snapback or rebounding effects.

Each analysis is fully traceable, with every input dataset, baseline, and adjustment factor logged, version-controlled, and auditable. This transparency allows evaluators to independently reproduce the same results, ensuring confidence in settlement, regulatory filings, and future planning.

Event-Based Forecast Overview

Demand Flexibility Forecasting Add-On

FLEX Programs Demand Flexibility Forecasting provides transparency into a given program's available resources by projecting the future flexibility that can be delivered across your distribution system, technologies, and customer base. Once the measurement foundation is in place, Recurve’s forecasting capability uses the same data structures, models, and validation processes as our Measurement feature to anticipate load-reduction potential under various future conditions.

The Forecasting module enables your teams to simulate the daily, hour-by-hour, region-by-region, and technology-by-technology delivery of demand response resources. It transforms measured performance into forward-looking insights, revealing how tomorrow’s weather, participation rates, and program maturity will shape tomorrow’s dispatchable capacity.

Grid operations teams can use forecasts to identify available MWs before events are called. Understanding not only the energy reduction potential, but also how that performance persists across event hours and those hours before and after an event window. Ensuring procurement and dispatch decisions are based on realistic expectations rather than static assumptions. Planning teams can evaluate whether expected load relief aligns with reliability targets or capacity commitments. With Forecasts, Program Administrators can communicate confidently with regulators and stakeholders about anticipated deliverability and seasonal performance expectations.

Core Data

CATEGORY	DESCRIPTION
Interval Load Data	Hourly or 15-minute AMI data for all participating service points. This data provides the observed load used to calculate event impacts and serves as the basis for performance measurement. Historical data from up to one year before the start of the program’s season should be provided to calibrate models. The data flow should be continuous during the program duration to provide ongoing model and event measurement.
Customer Metadata	Foundational customer and site attributes that allow segmentation, normalization, and contextual analysis. These fields enable the assessment of performance across various customer types, rate structures, and equity groups. Customer metadata should be provided at a minimum of monthly frequency when providing the entire service territory population dataset, to ensure that new accounts, tariff changes, and classification updates are captured. If the customer provides data only for participants, updates should occur after each event dispatch to align with the latest enrollment and event activity. Fields: • Service account ID • Utility’s Premise or location ID • Customer class (residential, commercial, industrial) • Tariff or rate schedule Income or equity flag (e.g., CARE/FERA, low-income indicator) • Building type or NAICS sector code • Climate zone
Enrollment and Program Participation Data	Defines which customers are enrolled in which programs, when they became active, and through which delivery channel or aggregator. Links participants to the specific events, dispatches, and performance groups that will be measured. Enrollment data should be continuously updated to ensure each event’s magnitude is properly captured with the latest enrollment information. Fields: • Program ID and name • Enrollment start and end date • DERMs Provider, or Aggregator (optional) • Enrolled Technology (optional) • Dispatch Group (optional)
Event Data	Defines the timing, scope, and structure of each demand response event, forming the foundation for performance measurement. Event data should be provided promptly after dispatch to enable the quickest possible measurement results. Fields: • Event ID and event name • Program ID reference • Event start and end date/time (with time zone) • Dispatch groups or networks • Participation list
Non Participant Data Optional	Customer metadata and AMI data from accounts not enrolled in the measured program. While optional, it is strongly encouraged to include this data to significantly improve measurement accuracy by accounting for external factors that influence energy use beyond program activity, such as economic conditions or broader system trends. This distinction enables the platform to separate true demand response impacts from background variation, producing more precise and defensible measurement and forecasting results.

Forecast Data

CATEGORY	DESCRIPTION
Historic Events	FLEX Forecasting is grounded in measured performance from past demand response events at the individual participant level. To enable accurate forecasting at launch, historical event records and participation data are required, allowing the system to learn how customers and technologies have responded under real conditions within this program context.
Historic AMI Data	To generate accurate forecasting models, FLEX requires at least 45 days of AMI data preceding the earliest historic event and a total of 12 months of historic AMI consumption per meter. This pre-event data establishes baseline consumption patterns, disaggregated load inputs, and load variability at the participant level, providing the foundation for model training and forecast calibration.
Technology Identifiers Optional	Identifies DER technologies associated with each participant, enabling technology-specific modeling and attribution. Includes device type (e.g., thermostat, EVSE, battery, heat pump), vendor, and installation date.
Distribution Node Identifiers Optional	Associate each meter or premise with its corresponding grid asset (feeder, substation). Enables locational DR measurement and constraint analysis.

Measurement Configuration

CONFIGURATION AREA	DESCRIPTION
Program Season	Specifies the months when your demand response program is active and when FLEX should measure and forecast performance. Setting the correct season ensures baselines, events, and reports reflect your actual operational schedule.
Aggregation Levels	Specifies how individual participant results are combined to provide program-wide insights. Results can be rolled up by technology, region, grid node, delivery provider, income or equity segment, and customer sector to support both operational and regulatory reporting.
Comparison Baseline Methodology Optional	Recurve’s measurement approach provides a transparent, defensible view of how much energy was actually delivered by each participant. Comparison baselines—typically the settlement baselines required by programs or regulators—allow results to be viewed through the same lens used for official reporting and compliance. Day-In-Day: Examples include 5 in 10, CAISO 10-in-10, where the baseline is derived from the average of the highest prior similar days within a specified date range. Optionally, accounting for weekends, holidays, and excluding prior event days. Weather Adjustment: Dynamically adjusted baselines with same-day temperature correction based on pre-event hour energy usage. Custom Jurisdictional Baseline: Baselines designed to align with local evaluation or regulatory requirements

Forecasting Configuration

CONFIGURATION AREA	DESCRIPTION
Technology Grouping Optional	Defines how participants are grouped by technology type for forecasting. Recurve provides guidance and recommended defaults, while also inviting customer input to align groupings with program priorities.