NMBE and CV-RMSE: Energy Model Calibration Metrics
How ASHRAE Guideline 14 uses NMBE and CV-RMSE to judge whether an energy model is calibrated to actual utility data.
A calibrated energy model is one whose predicted energy use closely matches actual utility bills. ASHRAE Guideline 14 provides the standard metrics for judging that match.
**NMBE — Normalized Mean Bias Error** NMBE measures overall bias. A positive NMBE means the model over-predicts consumption; a negative NMBE means it under-predicts. For monthly calibration, Guideline 14 recommends NMBE within ±5%.
**CV-RMSE — Coefficient of Variation of the Root Mean Square Error** CV-RMSE measures the typical size of month-to-month errors relative to the average consumption. For monthly calibration, Guideline 14 recommends CV-RMSE within 15%.
Meeting both thresholds gives reviewers confidence that the model is a fair representation of the building and that savings forecasts built from it are defensible.