All articles

Energy signature: what your building’s consumption-HDD line says

Plot your building's energy signature, read its deviations and spot heating drift in your BMS trend logs.

Adrien Lafond
13 min read
Illustration: on a screen, a cloud of points aligned on a straight line and one outlier above it, an outdoor thermometer at the frosted window

Since 1 January 2025, the BMS of a commercial building whose HVAC systems exceed 290 kW must be able to detect a loss of efficiency[1][1]RegulationLégifrance, “Decree No. 2020-887 of 20 July 2020 on building automation and control systems in non-residential buildings and automatic heat regulation”, codified in Articles R.175-1 et seq. of the CCH (French Construction and Housing Code)See in sources ↓. The decree does not say what that looks like. The energy signature gives it its simplest form: a line linking heating to cold weather. Here is how to plot it, read its deviations and check your trend logs. Because a line only tells the truth if your meters do too.

The energy signature: defining the consumption-HDD line, then plotting it

A definition: heating consumption as a function of degree days

The energy signature of a building is the line that plots its heating needs against degree days, period after period. Each point pairs the energy consumed with the cold endured, calculated from the average outdoor temperature. In the same climate, the points fall back onto the line: that is the reference behavior.

The energy label (the DPE rating) ranks a commercial asset on a scale. The line, for its part, tracks its actual energy consumption against the cold, winter after winter.

Heating degree days, the thermometer of the heating season

Heating degree days (HDD, called DJU in France) measure how far the outdoor temperature falls below a reference each day. At 8 °C, a day is worth 10 HDD on an 18 °C base. Added up, they quantify the cold of a heating season.

Did you know? The reference HDD published in France (Météo-France, COSTIC) are calculated on an 18 °C base. Some Swiss or Belgian guides use another base, which shifts the slope and the intercept of the line.

The choice of weather station and reference year is covered separately, in our page on HDD and the baseline.

What the slope and the intercept reveal

The slope, in kWh per HDD, expresses the energy added per degree day of cold; divided by 24 h, it gives the building's apparent heat loss coefficient, in kW/K. It measures how sensitive the envelope and the controls are to winter.

The intercept, the baseload, groups everything that consumes with no link to the cold: domestic hot water, process, permanent heat needs. NF ISO 50006 treats this type of model as an energy performance indicator (EnPI)[2][2]RegulationAFNOR, “NF ISO 50006”, energy management systems, evaluating energy performance using energy performance indicators (EnPIs) and energy baselinesSee in sources ↓. All that remains is to plot the line on your own trend logs.

Step 1: extract heating consumption at the right time step

The readings come from consumption monitoring: natural gas bills, energy accounting, an energy monitoring dashboard or the trend logs of your building BMS. The time step decides what you will see. At a daily or weekly step, anomalies show up. At a monthly step, they get smoothed out.

If the heating plant also produces hot water, plan to install separate metering before any plotting.

Step 2: match each period with HDD from the same station

Take the HDD at the same time step as your readings, from the nearest weather station. In a city center, a microclimate sometimes has to be taken into account. Then freeze the base from one year to the next: otherwise, the climate adjustment changes while nothing has moved on site.

Step 3: fit the regression line, a worked example

Worked example · natural gas-heated office building Twelve months, one line: the slope, the baseload and the R²
0100200300400 0100200300400 HDD base 18 per month MWh of heating per month Slope: 950 kWh/HDD Baseload: 4,000 kWh/month R² = 0.98
Summer months cluster near the baseload, winter months climb along the slope. A point that strays stands out at a glance. Illustrative values, unrelated to any real building.

Take an office building heated with natural gas, monitored over twelve months. The values are illustrative. The regression line comes out with a slope of 950 kWh per HDD and a baseload of 4,000 kWh per month. This slope reads as the extra energy consumed for each additional HDD.

The R² coefficient tells you whether the line explains the scatter. Close to 1, it means the cold drives most of the kWh. The expected annual consumption then becomes a benchmark. A gap of 20,000 kWh can then be costed in euros, at the average price of your contract.

Reading the deviations: six signatures that reveal an operating problem

Points above the line and a steepening slope: spotting energy drift

Annotated chart · six profiles

Six signatures that reveal an operating problem

kWh of heating per day
Reference line Normal points Abnormal points Observed trend
What you see
Likely cause on the BMS side
The test to ask the O&M contractor for
Illustrative data.

Drift goes unnoticed on a bill. On the line, it shows: points above the plot, then a slope that steepens from one winter to the next. Here are the six shapes we find in BMS trend logs, plotted against HDD. Each one comes with its likely cause and the test to ask for.

Which one to test first? The outliers: a change caused by a gap in the trend log has nothing to do with building drift. The baseload comes next, because it is the simplest to correct. Slope and scatter come last: they call the control itself into question.

Isolated outliers: the meter before the building

An isolated point far from the line rarely has a thermal cause. Disconnected meter, power cut, gap in the trend log, duplicate: the data lies before the equipment does. Cross-check it with consumption monitoring and energy accounting.

Want to know which of these six profiles looks like your building? We plot the line on your real BMS trend logs.

Schedule an audit

What remains is what this same line has to prove, depending on whether it serves the O&M contractor, OPERAT or a performance contract.

One line, four uses in the French framework

Diagnosis, OPERAT, IPMVP, BACS: what each use requires from the data

In France, the same consumption-HDD regression serves four uses. They require neither the same time step nor the same depth of history.

Use History required Who requires or uses it What you get out of it Common pitfall
Operational diagnosis Daily or weekly time step The building owner and its O&M contractor Drift, as it happens A monthly step smooths everything out
Décret tertiaire (OPERAT) Annual consumption The Éco Énergie Tertiaire scheme (France's tertiary decree on energy savings in commercial buildings, reported on the OPERAT platform)[4][4]RegulationLégifrance, “Decree No. 2019-771 on obligations to reduce final energy consumption in commercial buildings” (Décret tertiaire)See in sources ↓ Weather adjustment of consumption A poorly adjusted reference year
IPMVP Option C baseline Baseline period covering a full cycle The parties to a performance contract[5][5]FrameworksEVO, “International Performance Measurement and Verification Protocol (IPMVP)”, Option C “whole facility”See in sources ↓ Measured savings An undocumented change of use
BACS Hourly time step, kept at a monthly step for five years[6][6]RegulationLégifrance, “French Construction and Housing Code (CCH), Article R.175-3”See in sources ↓ Article R.175-3 of the CCH (French Construction and Housing Code)[6][6]RegulationLégifrance, “French Construction and Housing Code (CCH), Article R.175-3”See in sources ↓ Detection of efficiency losses Confusing capability with method

On BACS, the deadlines have moved. Buildings above 290 kW have been subject to it since 1 January 2025[1][1]RegulationLégifrance, “Decree No. 2020-887 of 20 July 2020 on building automation and control systems in non-residential buildings and automatic heat regulation”, codified in Articles R.175-1 et seq. of the CCH (French Construction and Housing Code)See in sources ↓. Those from 70 to 290 kW will be by 1 January 2030 at the latest, after a postponement[3][3]RegulationLégifrance, “Decree No. 2025-1343”, postponing the BACS deadline for 70 to 290 kW buildings to 1 January 2030See in sources ↓, or as soon as their heating or air-conditioning system is replaced. New buildings are not covered by the postponement.

The text requires the ability to detect a loss of efficiency: the decree imposes a detection capability, not a method. The line is only one means among others. In this regulatory environment, a reliable trend log serves all four uses. A trend log full of gaps serves none.

Before trusting the line, seven checks to run

Seven points to check with the O&M contractor

Checklist box · to do with the O&M contractor

Before trusting the line

0/7
Check the points already verified with your O&M contractor.

A clean regression on wrong trend logs is still wrong. Before presenting the line to a committee, we review seven points with the O&M contractor, metering first. Check the ones you have already verified.

Two empty boxes are enough to doubt the plot. The first job then concerns the state of the metering and the separation of domestic hot water. At Foobot, the order does not change: make reliable first, optimize next and AI only if the complexity justifies it. The building owner gains a line they can discuss point by point with their O&M contractor.

What Foobot does with an energy signature on your BMS trend logs

BMS advisory: making meters reliable and removing manual overrides

When the line lies, the cause is rarely thermal. Our BMS advisory starts by making the meters reliable: disconnections, gaps in the trend log, duplicates, unexplained peaks. It continues on the operations side, with the removal of manual overrides, sequences made consistent again and alarms that are useful. The building owner then has a trend log on which to plot the line. This work is worthwhile in itself, with no AI phase behind it.

A line to replot every heating season

Plotting the line takes little tooling. Checking it takes more: metering, HDD base, separation of uses. Once this baseline review is done, each deviation points to a precise place to act. It may be a schedule, a forgotten manual override or an optimal start that is too early. The issue is not settled once and for all. A line that is right at handover drifts as soon as operations slacken. Replot it every winter and compare it with the previous one. Would you rather have a model watch it between two seasons? We do it on your existing BMS, without heavy works.

detect drift continuously

Further reading:

FAQ - energy signature

What is an energy signature?

It is the line that links a building's heating consumption to the HDD of each period. Its slope measures sensitivity to cold, its intercept the baseload. Points that stray from it signal an operating fault to investigate.

How can energy consumption be optimized?

Tune before you invest: setpoints, time schedules, optimal start time. These practical steps reduce energy consumption without works. Then convert each kWh avoided at the average price of your contract to decide on the next actions. 👉 Control comes after tuning.

What are energy performance indicators?

On the line, three indicators are enough: the slope in kWh/HDD, the baseload and the scatter of the points. NF ISO 50006 classes this type of measure among EnPIs[2][2]RegulationAFNOR, “NF ISO 50006”, energy management systems, evaluating energy performance using energy performance indicators (EnPIs) and energy baselinesSee in sources ↓. Track how they change from one season to the next to manage energy performance.

How do you read the energy label?

The energy label gives an annual rating of the building. In commercial buildings, it shows where you stand, without saying how you operate. The consumption-HDD line measures that actual behavior, period by period. It also shows where the gap is lost.

Which tools for energy monitoring?

Energy monitoring relies on three sources: BMS trend logs, energy accounting based on bills and a dashboard shared with the O&M contractor. A spreadsheet is enough for consumption monitoring, provided the metering is reliable.

How can energy consumption be reduced?

Correct drift first: manual overrides never lifted, heating circuits kept running out of season, optimal start too early. Each correction shows on the line from the next period. 👉 Savings without works start with operations.

What is at stake in the energy transition?

For a commercial portfolio, the energy transition goes through two texts. The first is the Décret tertiaire[4][4]RegulationLégifrance, “Decree No. 2019-771 on obligations to reduce final energy consumption in commercial buildings” (Décret tertiaire)See in sources ↓. The second is the BACS decree, applicable since 1 January 2025 above 290 kW[1][1]RegulationLégifrance, “Decree No. 2020-887 of 20 July 2020 on building automation and control systems in non-residential buildings and automatic heat regulation”, codified in Articles R.175-1 et seq. of the CCH (French Construction and Housing Code)See in sources ↓. The common challenge remains proving measured rather than declared savings.

Sources

Sources and references

7 sources.


Energy newsletter

Get our articles on AI-driven HVAC control

One case study a month, on fine-grained control and its impact on building energy performance.