Customer Baseline Load (CBL)

The algorithm that defines which historical days are eligible for the baseline window. This is the most critical characteristic, as it determines the comparability of the baseline to the event day.

• Day Matching: Most methodologies require that baseline days match the event day's type (e.g., weekday, weekend, holiday).
• Exclusion Rules: Days with prior DR events, outages, or anomalous consumption are automatically excluded to prevent baseline contamination.
• Example: The California ISO's 10-in-10 method selects the 10 most recent eligible non-event weekdays.

A multiplicative or additive factor applied to the raw historical average to correct for load bias—the systematic difference between baseline window conditions and event-day conditions.

• Morning-of Adjustment: Uses consumption data from the hours immediately preceding the DR event to scale the baseline up or down.
• Weather Adjustment: Applies regression models to correct for temperature deviations between baseline days and the event day.
• Purpose: Without adjustment, a methodology systematically over- or under-pays participants, creating perverse incentives.

Defines whether the methodology applies the same rules to both positive performance (load reduction) and negative performance (load increase).

• Symmetric: The same baseline and adjustment logic applies regardless of performance direction. Common in capacity markets.
• Asymmetric: Applies a cap or floor only to positive performance to prevent gaming through intentional baseline inflation.
• Impact: Asymmetric rules protect market integrity but can penalize legitimate operational variability.

The temporal resolution of meter data required to calculate the baseline, which directly impacts accuracy and infrastructure cost.

• Interval Metering: Typically requires 5-minute, 15-minute, or hourly data from advanced metering infrastructure (AMI).
• Proxy Day Methods: Used when interval data is unavailable; substitutes a single historical day's load shape.
• Trade-off: Higher granularity enables more precise morning-of adjustments but demands robust data pipelines and validation.

A tolerance range around the baseline within which deviations are considered statistically insignificant and not compensated.

• Purpose: Prevents micro-payments for random load noise and reduces settlement disputes.
• Example: A ±5% deadband means a customer must reduce load by more than 5% below CBL to receive any payment.
• Calibration: Deadbands are often set based on the coefficient of variation of the customer's historical load.

A classification mechanism that identifies customers whose load is highly correlated with temperature, humidity, or wind speed.

• Segmentation: Weather-sensitive customers (e.g., those with electric HVAC) are assigned different baseline methodologies than non-weather-sensitive customers.
• Thresholds: A customer may be flagged if a linear regression of load against temperature yields an R² above 0.7.
• Rationale: Applying a non-weather-adjusted method to a weather-sensitive load produces wildly inaccurate baselines during heat waves or cold snaps.

The rigorous analytical process of quantifying the actual load reduction delivered by a demand response resource against its Customer Baseline Load. M&V protocols determine financial settlement by comparing observed meter data during an event to the counterfactual CBL estimate.

• Accuracy metrics: Bias, precision, and total error
• Adjustments: Weather normalization and opt-out corrections
• Standards: IPMVP and regional ISO-specific methodologies

A centralized software platform used by utilities or aggregators to manage, dispatch, and monitor demand response events across a portfolio of enrolled assets. The DRMS relies on pre-calculated CBLs for each participant to forecast available capacity and measure post-event performance.

• Function: Event dispatch, baseline calculation, settlement reporting
• Integration: Connects to smart meters and building management systems
• Scale: Manages thousands to millions of behind-the-meter assets

A graphical representation of a customer's electrical demand variation over a specific chronological period. CBL algorithms typically sample the most recent 10 non-event weekdays to construct a representative load profile, excluding outliers and prior event days.

• Granularity: 15-minute, hourly, or daily intervals
• Segmentation: Weekday vs. weekend, seasonal variations
• Application: Foundation for all baseline calculation methods

The backend financial system that calculates payments and penalties for demand response participants based on verified performance data and market rules. The engine compares actual meter data against the CBL during event hours to determine the net load reduction.

• Inputs: CBL, actual load, event duration, performance factor
• Outputs: Capacity payments, energy payments, non-compliance penalties
• Markets: PJM, CAISO, ERCOT, and bilateral contracts

A fully automated system where a utility signal directly controls customer loads based on pre-programmed permissions, eliminating manual intervention. CBL accuracy is critical in ADR because there is no human override to validate the fairness of automated dispatch decisions.

• Protocol: OpenADR 2.0b standard
• Latency: Sub-second to minute-level response
• Assets: HVAC, lighting, industrial pumps, EV chargers

The strategic reduction of power consumption during periods of highest grid demand to avoid capacity charges. CBL establishes the reference point against which peak shaving performance is measured, directly impacting a facility's demand charge savings.

• Trigger: Real-time pricing thresholds or grid stress signals
• Duration: Typically 2–4 hour windows
• Benefit: Reduced transmission and distribution charges