How do you measure and verify smart building energy savings?

A saving is energy that was not used, which means it can never be read off a meter. The Efficiency Valuation Organization (EVO), which maintains the International Performance Measurement and Verification Protocol (IPMVP), puts it this way: savings are determined by comparing measured consumption before and after a measure, making suitable adjustments for changes in conditions. That means estimating what the building would have used without the intervention, under the conditions that actually occurred, and comparing it with what it did use. The method has to be agreed before implementation, while baseline data can still be collected.

The four IPMVP options

IPMVP is a largely non-prescriptive framework, in which each project designs its approach in an M&V plan. It describes four options, which differ mainly in the measurement boundary:

For smart building projects that change controls across many systems, Option C is often the natural fit, because IPMVP associates it with multiple measures that have interactive effects. It does, however, need savings large enough to stand out from the normal variation in whole-building consumption. For a single measure, such as pump speed control, Option A or B drawn around that equipment gives a clearer answer.

What drives consumption

The variables that materially affect energy use, such as weather (heating and cooling degree days), occupancy, opening hours and production or service levels, need recording, because they become the adjustments in the baseline model. One-off changes, such as a new tenant, an extension or a change of use, need a defined treatment too, decided before they happen.

Energy first, money second

Savings should be reported in energy units first and converted to money only as a second step, using an agreed tariff method. A lower bill caused by a tariff change is not an energy saving, and a reduction in demand charges may need separate analysis. The report should include the uncertainty of the result and the cost of the M&V itself, which should be proportionate to the size of the saving and the commercial decision it supports.

New controls are installed. The reporting period is colder, with 2,400 degree days, and measured use is 108 MWh.

• Unadjusted comparison: 113 − 108 = 5 MWh.
• Adjusted baseline: 30 + 0.0375 × 2,400 = 120 MWh of expected use without the controls.
• Adjusted saving: 120 − 108 = 12 MWh, or 10% of the adjusted baseline.

The colder weather hid more than half of the saving in the simple comparison. Even so, the 12 MWh remains an estimate that depends on the model, the boundary and the data, and a full M&V plan would state its uncertainty.

What a workable M&V plan contains

IPMVP leaves the detail to each project's plan. To head off later disputes, the plan should state at least:

• the measure and its intended effect;
• the option, the measurement boundary and the meters used;
• the baseline period, its data and any known anomalies in it;
• the baseline model, its adjustment variables and how its fit was judged;
• how one-off changes will be identified and adjusted for;
• what happens if a meter fails during the reporting period;
• the reporting period, frequency and format;
• the price or tariff method for any financial conversion;
• the expected uncertainty and how it will be reported;
• who collects the data, who calculates the result and who reviews it.

The plan belongs in the project scope, not in a later side agreement, and the meters and data retention it relies on must actually exist before the baseline period starts.

Who calculates the result

If the result determines a payment, a bonus or a guarantee, the party that benefits should not be the only one calculating it. The contract should say who performs the calculation, who checks it, what access the checker has to the data and the model, and how a disagreement is resolved. Independent review costs money, so its depth should match the value at stake.