In Measurement and Verification (M&V), accurately tracking energy savings means you cannot just look at raw data. Adjustments must be made to the baseline or reporting period to account for normal or random variations in energy consumption affected by independent variables like weather, production output, or facility occupancy.

To keep your M&V models accurate, there are two types of adjustments that can be used: routine and non-routine.

What is a Routine Adjustment?

A routine adjustment is a regular, expected change in an independent variable. Because these changes are predictable, they are predefined when creating your initial M&V plan.

Examples of Routine Adjustments

• A normal, seasonal fluctuation in temperature between winter and summer.
• A school facility where occupancy predictably decreases during the summer break.

Because these factors are expected to happen, your regression model naturally accounts for them without requiring you to manually intervene or recalculate your baseline.

What is a Non-Routine Adjustment (NRA)?

A non-routine adjustment (NRA) is an unexpected or unpredictable change to an independent variable. These are events that were not pre-defined in the M&V plan.

When an unpredictable change occurs, it must be carefully logged to highlight the specific shift in energy consumption that happened directly due to the non-routine event.

Common Triggers for Non-Routine Adjustments

Examples of non-routine events that require adjustments include:

• An unseasonably extreme warm day occurring in the middle of a winter month.
• A manufacturing facility adding an unexpected third shift to their schedule, significantly increasing building occupancy and operating hours.
• A change in the building’s primary use type, or an expansion where another building is added to the facility footprint.
• Sudden production schedule changes due to an unusually high volume of orders.
• A new ECM is installed that reduced energy consumption. 

How to Handle Unexpected Variables in Your M&V Plan

Essentially, the rule of thumb is this: if the independent variable was not expected to change, then any resulting shifts in energy use must be logged as a non-routine adjustment. Once logged, the baseline or reporting period (depending on exactly when the non-routine event occurred) should be completely recalculated to ensure your energy savings data remains accurate and verifiable.

How to Calculate the Change in Energy Consumption from a non-routine event.

There are two IPMVP approved approaches to handling non-routine adjustments.

• IPMVP Option 1= data omission
• IPMVP Option 3 = additional indicator variable 

Data Omission:

This approach can be used in situations where no consumption is reported or there was an error with the data (i.e. extended power outage, utility meter connectivity issue, etc.). In this approach, the data is simply excluded from the calculation with no change to the adjusted baseline. 

Indicator Variable:

An indicator variable or change variable allows you to utilize an additional regressor to calculate the adjusted baseline. The indicator = 1 when the event occurred and 0 when the event was not occurring. 

For example, let’s say your building was only partially occupied for two weeks. This partial occupancy does not reflect the occupancy seen during the baseline, so you want to adjust the model to reflect this drop in energy consumption. When the regression model was initially built , your energy savings calculation was:

energy = 1,500 kWh + 6(CDDs)

This means that your baseload energy consumption is 1,500 kWh and increases by 6 kWh depending on the number of cooling degree days.

When we include an indicator variable (1=partially occupied and 0= full occupancy), the calculation changes to:

energy = 1,500 kWh + 6(CDDs) – 200 kWh (indicator)

Therefore, when the indicator variable=1, kWh consumption is reduced by 200 kWh per day when the building is partially occupied. When the indicator = 0, then the 200 kWh is multiplied by 0, essentially reverting back to the initial equation.