What are degree days?

Degree days turn outdoor temperature into a weather workload for heating and cooling. A heating degree day (HDD) accumulates how far the daily mean sits below a base temperature. A cooling degree day (CDD) accumulates how far it sits above a (usually higher) cooling base. Add those °C·days over a week, month or season and you get a simple signal of weather-driven load. The number is not a thermostat setting and it is not a comfort score. It is a proxy for how hard the envelope and plant had to fight the outdoors.

Base temperature: building energy balance, not the room setpoint

The base temperature is a modelling choice about when the building starts needing heat or cool, given insulation, internal gains and plant behaviour. It is not the occupied setpoint on the wall. Regional practice differs, and one number does not fit every building. In the UK, Met Office and CIBSE-facing work often uses about 15.5°C for heating degree days, with a higher cooling base (around 22°C is common when cooling is analysed). CIBSE TM41 is the theory reference for degree-day methods; keep the detail in the guide, not in a marketing page. German practice is heating-led: the familiar 20/15 convention (VDI 3807 and related guidance) uses an indoor reference near 20°C and a heating threshold near 15°C, so a heating day is counted when the daily mean falls below 15°C. Swedish estates talk about graddagar and normalårskorrigering in energideklaration and district-heat follow-up; SMHI Energi-Index is often preferred over raw graddagar when the job is a full weather correction.

Weather normalisation: why raw year-on-year kWh lies

A cold year burns more heat. A mild year looks like a saving even when nothing in the plant changed. Weather normalisation (also called weather-normalized energy) adjusts consumption so you can compare periods on a fair weather footing. The simple ratio method scales weather-dependent use by current degree days over normal (or baseline) degree days. A regression form is clearer when you have enough history: kWh ≈ a + b·HDD + c·CDD, often with a weekend or occupancy flag. The intercept a is the weather-independent base load; b and c are the weather slopes. Limitations matter. Better insulation and higher internal gains weaken how much degree days explain. Cooling-dominated sites, solar gains and humidity need more than HDD alone. If the model only ever saw heating season, do not trust it on a hot August.

What to check before you trust a normalised figure

Mismatch the base temperature between your HDD series and the one baked into a published "normal" and the ratio is noise. Skip the base-load split and you weather-correct lights and plugs that barely care about outdoor temperature. Use a remote weather station when the site sits in a different microclimate and the residuals look like plant faults. Incomplete meter periods, estimated gaps and mixed calendar months break both ratio and regression. Applying an HDD-only correction to a cooling-dominated load is the classic trap: the weather signal is pointing the wrong way.

Where degree days sit in monitoring and targeting

Degree days are the weather covariate. They do not own the operating method. Energy monitoring and targeting (M&T), CUSUM charts and degree-day regression as a weekly exception process live in our Field Note on energy monitoring and targeting. This glossary stops at defining HDD, CDD and weather normalisation so that method page can stay the home for the workflow.

How FrostLogic Explore uses them. Explore's Energy Savings Loop measures approved changes at 7, 14 and 28 days against a weather-normalised baseline of the form kWh = a + b·HDD + c·CDD + d·weekend. There is no estimated-savings tier: money shows only when the customer set a tariff. Forecasts can use the same weather covariates. The commercial path for continuous meter reads is energy monitoring in Explore; the buildings-shaped commercial page is energy management for commercial buildings.