Measurement & Verification with IPMVP: Options and the Measurement Boundary

Measurement and Verification (M&V) is the process of calculating the actual savings achieved by an energy improvement measure in a defensible way. The International Performance Measurement and Verification Protocol (IPMVP) is the reference framework for this process. Without a written M&V plan, reported savings remain just a claim that can't hold up against a challenge from management or an auditor.

The four IPMVP options

IPMVP offers four main options for designing an M&V project, distinguished by the measurement boundary and how directly the parameters involved are measured.

• Option A - Retrofit Isolation (partial measurement): some relevant parameters are measured while others are estimated using engineering judgment
• Option B - Retrofit Isolation (full measurement): all parameters affecting the retrofitted equipment or system are measured directly
• Option C - Whole Facility: based on main site or production-line meter data and a regression model at the overall facility level
• Option D - Calibrated Simulation: used when a simple before/after comparison isn't possible, relying on a simulation model calibrated with real data

Defining the measurement boundary

The measurement boundary needs to be set in writing before the improvement project is implemented: will only the retrofitted equipment be measured (as in Options A and B), or will the whole site be treated as one unit (Option C)? This decision directly affects metering cost, result accuracy, and sensitivity to changes unrelated to the project.

Avoided energy use versus normalised savings

Avoided energy use is the difference between actual consumption in the reporting period and what would have occurred without the improvement measure, based on a baseline adjusted to that period's conditions. Normalised savings make the same comparison, but against a fixed set of conditions (such as a typical weather year), so savings can be meaningfully compared across different years.

Non-routine adjustments

A non-routine adjustment is needed when a change occurs at the site that isn't explained by the model's routine variables (such as production or weather) — for example, adding new equipment mid-reporting-period or changing operating hours. The method for calculating this adjustment needs to be documented and agreed by both parties in the M&V plan upfront, so it isn't challenged in the final report.

• The measurement boundary and the chosen IPMVP option
• The length and conditions of the baseline period and the reporting period
• The list of independent variables and how they're adjusted
• The method and acceptance criteria for non-routine adjustments
• The acceptable level of uncertainty for the final result

Uncertainty in the M&V result, and a worked example

Every M&V result carries a level of uncertainty stemming from several sources: the accuracy of the measurement equipment used, the residual error of the baseline model, sampling error when only part of the system is measured, and estimation error in the non-routine adjustment method. The result is usually reported as a confidence interval (such as 90%); the greater the financial value of the savings to the organisation, the lower the acceptable uncertainty level should be, even if more precise measurement costs more. For example, in a compressor motor upgrade project, if only that motor is measured (Option B), uncertainty is low and metering cost is limited; but if the upgrade also affects consumption elsewhere — for example, reducing thermal load or affecting the cooling system — Option C may be more suitable for capturing the net effect on the whole site, even with the added uncertainty from fluctuations in other loads.

Choosing the right option

The choice of option depends on the project's complexity, the budget available for metering, interactive effects between systems (such as a lighting measure that also changes the cooling load), and how financially material the savings are to the organisation. Small, isolated projects are usually assessed with Option A or B, while broad site-wide improvement programmes are usually assessed with Option C. In large projects with several simultaneous measures, a mix of options is sometimes used across different parts of the site, preserving the accuracy needed for high-value measures while keeping metering costs justified for smaller ones.

M&V is only as accurate as the agreement, reached before the project starts, on both the measurement boundary and the adjustment method.