Whole Facility Approach to M&V
The whole-facility approach (Option C) uses the facility's utility meters - typically electricity and fuel - and compares adjusted baseline consumption to reporting-period consumption. Because it captures all end uses, it inherently accounts for interactive effects and multiple measures, but it requires that total savings be large enough relative to whole-building variability to be distinguishable from noise.
Key concepts: regression modeling (often with weather variables such as heating and cooling degree days or temperature bins), coefficient of determination (R²) and its limits, standard error and uncertainty, baseline period length (usually 12 months minimum to capture seasonality), data quality and interval (monthly vs. hourly), and non-routine adjustments for changes in the facility. A common rule of thumb is that savings should exceed roughly 10% of the whole-facility baseline use for Option C to be reliable; below that, isolation options are preferred. Regression models should be checked for autocorrelation, outliers, and appropriate variable selection.
Traps: using R² alone as a measure of model quality (it can be high with few points or autocorrelated data); ignoring non-routine changes such as new tenants or added equipment; using too short a baseline; and failing to check that the model's independent variables are still valid in the reporting period. Also, questions may imply Option C cannot be used for multiple measures - it can, but attribution among measures is not possible.
Worked example: Baseline regression: monthly kWh = 50,000 + 8 × CDD, R² = 0.92, 12 months of data. Reporting period has CDD = 1,200 and actual use = 55,000 kWh. Adjusted baseline = 50,000 + 8 × 1,200 = 59,600 kWh. Savings = 4,600 kWh, about 8% of baseline - near the threshold where Option C reliability is marginal.