Weather-normalized baselines are the foundation of credible energy performance analysis. Without them, you're comparing apples to oranges — a mild winter will always look like "savings" even if nothing changed operationally.
Pro tip: Weather normalization is required by ASHRAE Guideline 14 and IPMVP for any M&V report that will be used in an energy performance contract or utility incentive claim.
Why Weather Normalization Matters
Buildings consume energy primarily for heating, cooling, and baseload operations. Weather directly impacts the first two, which means raw year-over-year comparisons are inherently misleading.
ASHRAE Guideline 14 provides the statistical framework for creating regression models that separate weather-dependent consumption from operational changes. EdiMono implements these models automatically, using heating and cooling degree days (HDD/CDD) as independent variables.
The Regression Approach
The most common model is a changepoint regression — sometimes called a "3-parameter" or "5-parameter" model depending on whether the building has heating only, cooling only, or both.
3-Parameter Heating Model: Energy = Baseload + Heating_Slope × max(0, Balance_Point - OAT)
5-Parameter Model (Heating + Cooling): Energy = Baseload + HS × max(0, HBP - OAT) + CS × max(0, OAT - CBP)
EdiMono fits these models using least-squares regression and validates them against ASHRAE thresholds for R² (>0.7) and CV(RMSE) (<20%).
Important: The balance point temperature (the OAT at which heating or cooling kicks in) varies by building and must be found by optimization, not assumed. EdiMono sweeps candidate values from 10°C to 22°C to find the best fit.
Setting Up Baselines in EdiMono
- Import 12+ months of utility data — monthly bills or interval data
- Select the baseline period — typically the year before an efficiency project
- EdiMono auto-selects the best model — testing 3P, 4P, and 5P fits
- Review the statistical diagnostics — R², CV(RMSE), t-statistics
- Generate normalized comparisons — see true performance changes
Key Takeaway
Always use at least 12 months of baseline data. Fewer data points dramatically increase uncertainty and may not satisfy ASHRAE or IPMVP requirements for M&V reports.
Common Pitfalls
- Too few data points: You need at least 12 monthly readings for statistical validity
- Occupancy changes during baseline: If the building's use changed mid-baseline, split the period
- Multiple fuels: Model electricity and gas separately — they respond differently to weather
Key Takeaway
Weather normalization isn't optional — it's the difference between guessing and knowing. With EdiMono, the math is handled automatically so you can focus on making decisions, not building spreadsheets.