Building energy benchmarking: what it takes, and what it actually shows

Benchmarking a building means comparing its energy use to something. The something is what people skip over, and it is the whole question. There are two comparisons, they answer different things, and a report that blends them produces a number nobody can act on.

Against itself: is this building using more energy than it did last year, once weather and calendar differences are accounted for. Against its peers: is this building reasonable at all for its type, size and climate. The first tells you whether something changed. The second tells you whether there is room to improve. Neither substitutes for the other.

What you need before you can benchmark anything

The inputs are short to list and are where most of the effort goes.

• A complete consumption record. Every fuel, every meter, every month, with no gaps. A single missing month distorts an annual intensity by roughly eight percent, and the direction is always flattering.
• Periods you can align. Utility billing periods do not match calendar months, so a year of consumption requires apportioning the bills that cross the year boundary, consistently, with a rule you can state.
• A floor area you can defend. Intensity is consumption divided by area, so the denominator matters as much as the numerator. Gross versus rentable versus conditioned area can differ by fifteen percent or more in the same building, and the benchmark you compare against assumes one of them.
• Consistent units. Gas in cubic metres, electricity in kilowatt-hours, district energy in whatever the supplier uses, converted with factors you have written down and dated.
• Knowing what the building does. Occupancy hours, principal use and any process load, because the peer comparison is only meaningful within a use type.

The uncomfortable part is that none of this is analysis. It is data assembly, and it is the reason benchmarking projects take longer than expected. We cover the record itself in energy data management, Green Button and bill PDFs and the checks that come before reporting in utility bill data validation.

The peer comparison, and where the peer group comes from

Energy use intensity is the common currency: energy per unit of floor area per year. ENERGY STAR Portfolio Manager explains how it defines energy use intensity, and its 1 to 100 score positions a building against similar buildings, where 50 is median performance.

For Canadian buildings the peer group behind that score comes from Canadian survey data rather than United States data, which matters because climate and construction differ. Natural Resources Canada maintains the Canadian side of the program. A score calculated against the wrong peer group is not a small error; it can move a building from below median to above it.

Weather, and why last winter is not comparable to this one

A building in a colder year uses more heating energy without anything being wrong. Comparing raw consumption year over year attributes that to the building, which is how a perfectly maintained site gets flagged as deteriorating.

The correction is weather normalisation: expressing consumption as a function of degree days so the comparison is made at equivalent conditions. Done properly it separates the weather from the building, and the residual is what you should be looking at. We describe the method, and the cumulative sum technique that makes drift visible, in weather normalisation, CUSUM and RETScreen M&V.

One caution worth stating: normalisation is a model, and a model fitted to a building whose operation changed mid-year will attribute the change to weather. The static facts about the building, its schedule and its systems have to be tracked alongside the consumption, or the normalisation quietly absorbs real changes.

What benchmarking does not tell you

• Why. An intensity figure identifies which building is expensive and says nothing about what is causing it. That requires interval data or a survey.
• Whether a fix worked. Annual intensities move too slowly and are too noisy to verify a measure. That is a monthly comparison against a baseline model.
• What to do about it. A benchmark is a ranking, not a recommendation. Turning it into work requires someone to look at the worst performers and form a view.
• Whether the data is right. A confident benchmark on incomplete data is the most common failure mode, and the report will not warn you.

This is why benchmarking works best as the entry point to a routine rather than as an annual exercise on its own. The benchmark says where to look, interval data and a survey say what is wrong, and a monthly review says whether it stayed fixed. We set out what that routine contains in the monthly energy review.

Reporting obligations that use the same figures

Several Canadian jurisdictions now require large buildings to report energy and water use, and those returns are built from exactly the record described above. Doing the benchmarking properly means the reporting is a by-product rather than a separate project, which we cover in building energy reporting rules in Toronto, Montreal and Vancouver.

The practical advice is to build the record once, to the standard the strictest requirement demands, and then use it for everything: internal ranking, peer comparison, statutory reporting and incentive applications. Building four parallel records is how organisations end up with four different numbers for the same building.

FAQ

What is energy use intensity?

Energy per unit of floor area per year, usually expressed per square foot or per square metre. It allows buildings of different sizes to be compared, provided the floor area basis is consistent and stated, because gross, rentable and conditioned area can differ substantially in the same building.

How much data do we need to benchmark?

Twelve consecutive months with no gaps, for every fuel and every meter, plus a defensible floor area. Two years is better because it lets you separate a genuine change from a one-off. A single missing month distorts an annual intensity by around eight percent.

Do we need weather normalisation?

For comparing a building to its own history, yes. Without it a colder year looks like deterioration. For comparing against a peer group, the benchmark itself usually handles climate, so raw annual intensity is what the program asks for.

Is an ENERGY STAR score comparable between Canada and the United States?

The metric is the same but the peer group is not. Canadian buildings are scored against Canadian survey data. Using the wrong peer group can move a building across the median, so it is worth confirming which is being applied.

Can we benchmark without a full year of data?

You can rank buildings against each other on partial data if the periods are the same for all of them, which is often enough to decide where to look first. You cannot produce a defensible annual intensity or a peer score without twelve complete months.