A retrofit needs a measurement plan before a savings claim

The single most important decision in measuring a retrofit is taken before the works begin: the baseline.

A baseline is the description of how the building performed before the intervention, normalised for the conditions under which it performed. It requires metered data over a period long enough to represent the building's range of conditions, together with the weather and occupancy data needed to interpret it. Once the works are done, the building's performance has changed and the pre-intervention period cannot be recovered — the data may still exist, but the building that produced it does not.

This is why a baseline established after the works is not a baseline. It is typically a short period of data selected because it is available, normalised with assumptions about conditions that were never recorded. The resulting comparison can be presented as a saving, and it does not distinguish the intervention from the weather, the occupancy, or the simple passage of time.

The practical implication is that the measurement plan is part of the project's preparation, not its reporting. Metering, data collection and the recording of operating conditions are activities that must be in place before the works, and their cost belongs in the project's budget.

Two variables move a building's consumption independently of any intervention, and both must be handled explicitly.

Weather is the first. Cooling and heating demand track outdoor conditions, so a period that is warmer or cooler than the baseline period will show a difference in consumption that has nothing to do with the retrofit. Cooling degree days — a measure of how far and for how long outdoor temperature exceeds a reference — is the standard normalising variable for cooling-dominated buildings, and it should be used to adjust both periods to a common condition. In a Gulf context, where cooling dominates the load, this adjustment is usually the largest single correction.

Occupancy and use are the second. A building that is fuller, open longer, or used more intensively consumes more, and a difference in the level of use between the two periods will appear as a change in performance. The normalising variable depends on the building — occupied hours, occupancy count, throughput, production volume — and should be one the building actually records reliably.

Where a normalising variable is not recorded consistently, the honest course is to state that the comparison has not been normalised for it and to identify the direction and rough magnitude of the possible distortion. Presenting an unnormalised difference as a saving is the error the plan exists to prevent.

A measurement plan that will support a savings claim carries the following, agreed before the works.

The following is illustrative and entirely hypothetical. It demonstrates the arithmetic of a savings calculation; it is not a benchmark and it asserts no actual tariff, consumption or result.

Suppose an assumed baseline period shows consumption of 100,000 kWh, and a post-retrofit period of the same length shows 84,000 kWh. The simple difference is 100,000 − 84,000 = 16,000 kWh, or a 16% reduction.

That figure is not yet a saving, because the two periods were not necessarily comparable. Suppose the assumed cooling degree days were 1,200 in the baseline period and 1,080 in the post-retrofit period — 10% lower. If the building's consumption is assumed to vary proportionally with cooling degree days for the weather-driven element, then part of the 16% reduction is attributable to the milder period rather than to the retrofit.

The correction requires knowing what share of consumption is weather-driven, which is established from the building's own data rather than assumed. If, purely for illustration, 60% of baseline consumption were assumed weather-driven, the weather-attributable reduction would be approximately 60% of the 10% milder conditions applied to the baseline — roughly 6,000 kWh — leaving approximately 10,000 kWh attributable to the retrofit and other factors, pending the occupancy normalisation.

At an assumed tariff of 0.40 per kWh, an assumed saving of 10,000 kWh would be worth approximately 4,000 over the period. The tariff is assumed, not asserted, and the actual figure depends on the building's own contract structure, including any standing or capacity charges that a consumption reduction does not affect.

The purpose of the illustration is to show that the headline difference — 16% — is not the saving. The saving is what remains after the drivers that are not the retrofit have been accounted for, and establishing it requires the baseline and normalisation data that the measurement plan was written to capture.

The final element of the plan is a rule about what to do with the result.

A measured variance from the expected saving is information. It may indicate that the measure is not performing as designed, that commissioning was incomplete, that maintenance has lapsed, that the baseline was unrepresentative, or that a change in use has occurred. All of these are worth investigating, and the decision rule sets the threshold at which the investigation is triggered so that it happens consistently rather than when someone happens to notice.

The discipline the rule enforces is a distinction between investigating a variance and claiming a cause for it. An investigation that finds a measure performing below expectation has identified a question to resolve; it has not established why the shortfall occurred, and reporting it as though it had would substitute a plausible explanation for an actual finding. The rule keeps the reporting to what the data supports — the variance, the investigation triggered, and the questions outstanding — which is the honest output and the one that leads to the measure actually being corrected.