Definition
Incremental generation or purchased power cost directly attributable to renewable output forecast error: the dollar cost of committing the wrong generation mix because the solar/wind forecast was wrong.
The Measurement Gap
Traditional dispatch tracking records fuel costs, purchased power expense, and generation unit performance. It does not connect those costs to the upstream forecast decision that determined which units were committed. When a renewable forecast misses by 20%, the resulting gas turbine ramp or spot market purchase appears in the fuel variance report with a notation like "increased renewable variability." The forecast model that caused the variance is never implicated.
The specific failure point is commit-horizon decisions. The generation dispatch plan is locked in hours before the operating interval. At that commit point, the renewable forecast determines how much firm thermal generation is scheduled. A solar forecast that overstates noon output by 150 MW causes the operator to commit less thermal generation than needed. When actual output comes in low, the operator must ramp fast-start units (which are expensive precisely because they were not planned) or buy from the short-term market at peak prices. The forecast made a specific commitment error. That error has a specific dollar cost. RFDC measures it.
RFDC enables the operational decision of whether to supplement the primary renewable forecast with ensemble methods, nowcasting, or a higher-fidelity satellite irradiance model, and whether the cost of that supplementation is justified by the forecast error cost it eliminates.
Without this measurement, forecast vendors report MAPE statistics while fuel overruns are attributed to market volatility. The model improvement case cannot be made because the financial consequence of current model performance is invisible.
AI processing of interval-level dispatch logs, renewable metering, and market settlement data makes the forecast-to-cost attribution computationally feasible at the granularity where decisions are actually made.
Formula
RFDC = Σ (Cost_actual_dispatch_i − Cost_optimal_dispatch_i) [for intervals where |Forecast_i − Actual_i| > threshold]
Variables:
- Cost_actual_dispatch_i: actual fuel, purchased power, and start-up costs for dispatch interval i ($/interval). Source: generation cost accounting, EMS dispatch logs
- Cost_optimal_dispatch_i: modeled cost of the dispatch that would have been chosen with perfect renewable foreknowledge for interval i ($/interval). Source: ex-post dispatch optimization using actual renewable metering
- Forecast_i: renewable output forecast for interval i (MW). Source: renewable forecasting system
- Actual_i: actual renewable output for interval i (MW). Source: SCADA/EMS metering
- Threshold: minimum forecast error magnitude that changes the dispatch decision (e.g., 10% of installed renewable capacity or 50 MW, whichever is lower)
Calculation Frequency:
Daily accumulation, reported monthly and annually
Output: Dollars per period. Track RFDC_over (over-forecast direction, thermal committed unnecessarily) and RFDC_under (under-forecast direction, emergency ramp or spot purchase) separately to reveal forecast bias direction. RFDC below 2% of total generation cost indicates manageable bounds; above 5% indicates a material cost driver warranting model review.
Proposed formula, pending validation pilots.
At a Glance
Decision Supported
How much is the renewable forecast error costing in real dollars, not forecast MAPE as a statistical measure, but the actual fuel, purchased power, and penalty costs caused by the gap between predicted and actual renewable output?