Measurement and verification of energy savings

Energy savings cannot be measured directly. You can measure how much energy a building uses, but savings are the difference between what it used and what it would have used without the project, and that second number never happens. Measurement and verification (M&V) is the discipline of estimating that counterfactual credibly, using metered data, adjustments for weather and other changes, and a documented method that others can check.

M&V matters whenever money depends on savings: energy performance contracts, utility incentive programs that pay for performance, green bonds, internal capital approvals and carbon reporting. It also matters for learning, because many projects deliver less than predicted and only measurement reveals why.

The core equation

All M&V methods rest on the same idea:

Savings = baseline energy (adjusted to reporting period conditions) minus reporting period energy, plus or minus adjustments

The baseline period is a representative period before the project, usually 12 months. The reporting period is after implementation. Because weather, occupancy and operating hours differ between periods, the baseline must be adjusted so both periods are compared on the same conditions.

Two types of adjustment are used:

• Routine adjustments for factors expected to change and that can be modelled, such as heating and cooling degree days, production volume or occupancy.
• Non-routine adjustments for one-off changes, such as an added floor, a new tenant with a server room, or a change in operating hours.

The IPMVP options

The International Performance Measurement and Verification Protocol (IPMVP), maintained by the Efficiency Valuation Organization, defines four options. Most Canadian utility programs and performance contracts refer to it.

Choosing an option

• Use A when the uncertain parameter can be measured cheaply and the others are well known or agreed. It is the least expensive but relies on stipulated values.
• Use B when operation varies, such as a VFD whose savings depend on actual load profile.
• Use C when savings are large relative to the normal variation in the whole-building meter, commonly cited as more than about 10 percent of baseline use, or when many interacting measures make isolation impractical.
• Use D when there is no baseline (new buildings), when the baseline will change significantly, or when savings must be attributed to individual measures within a whole-building project.

ASHRAE Guideline 14

ASHRAE Guideline 14, Measurement of Energy, Demand, and Water Savings, provides technical detail that complements IPMVP. It covers regression modelling, uncertainty analysis, instrument requirements and calibration criteria for simulation models. For calibrated simulation, it sets limits on the coefficient of variation of the root mean square error, CV(RMSE), and the normalized mean bias error, NMBE, between the model and measured data. Commonly used limits are 15 percent CV(RMSE) and plus or minus 5 percent NMBE with monthly data, and 30 percent and plus or minus 10 percent with hourly data.

Building a baseline

For whole-building methods, the baseline is usually a regression model of energy use against weather and other drivers.

1. Collect at least 12 months of utility or interval meter data before the project, covering all seasons.
2. Collect matching weather data from a nearby station and calculate degree days or use hourly temperatures.
3. Record operating conditions: occupancy, hours, major equipment changes, vacancies.
4. Fit a model, often a change-point regression that separates base load, heating and cooling. Interval data allows more detailed time-of-week and temperature models.
5. Test the model's statistical fit and check residuals for patterns.
6. Document every assumption.

Electricity and gas should be treated separately because they respond to different drivers. A gas baseline in a Canadian climate is usually dominated by heating degree days, while electricity may depend more on cooling degree days, schedules and plug loads.

Writing an M&V plan

An M&V plan should be agreed before the project starts. A good plan includes:

• The measures, their intended effect and expected savings.
• The selected IPMVP option and measurement boundary.
• The baseline period, energy data and independent variables.
• Any stipulated values and their source.
• Meters and loggers to be installed, their accuracy and calibration.
• The calculation method, including routine adjustments.
• How non-routine adjustments will be identified and handled.
• Reporting period length and frequency of reports.
• Responsibilities and the expected cost of M&V.

Common pitfalls

• No baseline data: meters or loggers installed after the retrofit can only support Option D or stipulation.
• Untracked changes: tenant changes and new equipment show up as lost savings unless recorded.
• Too short a reporting period: heating measures need at least one full winter.
• Interactive effects ignored: a lighting retrofit increases heating energy in winter; isolated measurement misses it.
• Over-reliance on stipulation: stipulated hours that are wrong make the savings wrong.
• M&V that costs more than it is worth: effort should match the value of the savings and the risk involved.

Using building automation data

BAS trend data and interval meters make M&V cheaper and more informative. Logged fan speeds, pump power and plant output let Option B measurements run continuously without separate loggers, provided sensors are calibrated. Continued tracking also catches savings erosion as overrides creep back in. See ongoing and monitoring-based commissioning for how this fits into building operations.

Frequently asked questions

Do utility incentive programs require M&V?

Many custom and performance-based programs do, often using an IPMVP option. Prescriptive rebates for standard equipment usually rely on deemed savings instead. Program rules vary by utility.

How long should M&V last?

Long enough to capture seasonal effects, usually at least 12 months. Performance contracts often require M&V for several years or the contract term.

Can I use an energy audit estimate as proof of savings?

No. An audit estimate is a prediction. M&V compares it with measured results, which is how you learn whether the prediction was right. See HVAC energy audits for the difference.

Key takeaways

• Savings are the difference between adjusted baseline energy and measured reporting period energy.
• IPMVP Options A to D trade measurement cost against accuracy and scope.
• ASHRAE Guideline 14 provides statistical criteria for regressions and calibrated models.
• Agree an M&V plan, baseline period and adjustment rules before the project starts.
• Track building changes and use BAS and interval data to keep savings from eroding.