Customer Baseline Load (CBL) Calculation

The foundational algorithm for selecting representative baseline days from a look-back window of recent, non-event days. The methodology identifies the Y most recent non-event, non-holiday weekdays and then selects the X days with the highest energy consumption from that subset.

• Common variant: 10-of-20 baseline for commercial and industrial customers
• Purpose: Captures the customer's peak demand potential, ensuring the baseline reflects a conservative estimate of what load would have been
• Exclusion rules: Automatically filters out prior event days, weekends, holidays, and days with anomalous meter data to prevent contamination of the baseline

A multiplicative or additive correction factor applied to the raw baseline to account for temperature-driven load divergence between the baseline period and the event day. Without this adjustment, a demand response event called on an unusually hot day would systematically overpay participants.

• Heating/Cooling Degree Days: The primary independent variable for regression-based adjustments
• Morning-of adjustment: Uses actual event-day weather observed up to the hour before curtailment begins
• Non-linear breakpoints: Advanced methodologies apply piecewise linear regressions with distinct slopes above and below a reference temperature (typically 65°F/18°C)

A load-matching scalar that aligns the baseline's magnitude with actual metered load during a defined pre-event window on the event day itself. This corrects for day-to-day variability not captured by weather alone.

• Calculation: Ratio of actual load to unadjusted baseline load during the 2-4 hours immediately preceding the event
• Capping constraint: Typically limited to ±20% to prevent gaming or over-correction from anomalous pre-event behavior
• Critical function: Eliminates systematic bias where customers could artificially inflate their baseline by increasing load on selected high-X-of-Y days

A rigorous data quality gate that verifies the integrity of interval meter data before calculating performance payments. Corrupted or missing data can invalidate an entire settlement cycle.

• Completeness check: Requires a minimum threshold of valid intervals (typically 95%+) for the event period
• Pulse overflow detection: Identifies meter register rollover events that create artificial negative consumption spikes
• Voltage sag correlation: Cross-references consumption anomalies with power quality event logs to distinguish real load drops from meter errors
• Backstop estimation: Applies linear interpolation for short gaps; flags extended outages for manual review

The practice of applying distinct CBL methodologies to different customer segments based on their load profile characteristics. A single algorithm cannot accurately model both a steady-state industrial process and a weather-volatile commercial building.

• Industrial: Often uses a simple averaging baseline (e.g., 5-of-10) due to flat, predictable load shapes
• Commercial: Requires weather-sensitive adjustments and SAA due to HVAC-driven variability
• Residential: May use a matched-pair control group methodology where non-participating neighbors serve as the counterfactual
• Aggregation rule: Portfolio baselines are calculated as the sum of individual customer baselines, not by applying the methodology to aggregated load

The post-event calculation that quantifies actual load reduction by comparing metered consumption against the finalized CBL. This is the settlement-grade output that determines financial payments.

• Load reduction formula: Reduction = max(0, CBL_adjusted - Actual_Load) for each settlement interval
• Deadband exclusion: Reductions below a minimum threshold (e.g., 1 kW) are zeroed out to eliminate noise
• Persistence requirement: The customer must sustain the reduction for the full event duration; early snapback disqualifies payment for non-compliant intervals
• Audit trail: Every input—selected baseline days, weather station data, adjustment factors—must be immutably logged for regulatory review

The automated dispatch of load reduction signals to enrolled participants during grid stress events. CBL serves as the settlement backbone for these programs—without an accurate baseline, incentive payments cannot be calculated. Modern orchestration platforms use OpenADR 2.0b to send signals and retrieve meter data for CBL computation.

• Event-triggered load curtailment
• Relies on CBL for performance measurement
• Common in capacity markets and ancillary services

A control logic that dispatches battery energy storage systems (BESS) to discharge during periods of highest site load, reducing demand charges. CBL is critical here because it establishes the counterfactual consumption profile—what the load would have been without the battery discharge. This allows facility managers to quantify the financial value of their storage assets.

• Reduces kW demand charges
• Uses CBL to measure avoided peak
• Common in C&I rate optimization

The use of targeted DERs to reduce peak load on a specific substation or feeder, deferring traditional infrastructure upgrades. CBL methodologies are essential for measurement and verification (M&V) of NWA projects. Regulators require statistically rigorous baselines to confirm that the DER portfolio delivered the promised peak reduction.

• Avoids transformer and feeder upgrades
• Requires CBL for regulatory M&V
• Often uses 10-day baseline windows

The strategy of charging batteries during low-price off-peak periods and discharging during high-price on-peak periods. CBL calculations help isolate the incremental impact of storage dispatch from normal load variability. Without a reliable baseline, it is impossible to distinguish arbitrage savings from coincidental load reductions.

• Captures energy cost differentials
• CBL separates storage impact from noise
• Requires sub-metering for precision

A privacy-preserving ML technique that trains forecasting models across decentralized edge nodes without centralizing raw customer data. CBL models benefit from this approach because baseline calculations require historical load data that customers may be reluctant to share. Federated learning enables collaborative model improvement while keeping meter data on-premises.

• Preserves customer privacy
• Improves CBL accuracy across portfolios
• Edge nodes compute local model updates

Standardized frameworks such as IPMVP and ASHRAE Guideline 14 that define how energy savings must be quantified. CBL is the foundational input to Option C (whole-facility) M&V approaches. These protocols specify acceptable baseline adjustment methods—including weather normalization and occupancy corrections—to ensure savings claims are statistically defensible.

• Governs savings quantification rigor
• Specifies baseline adjustment techniques
• Required for performance contracting