IPMVP Option C When You Replace CHP With a Heat Pump: How to Adjust the Baseline

Replacing a cogeneration plant with a heat pump is one of the harder things you can do to a whole-facility measurement and verification baseline. Option C compares metered energy before and after, and this project moves consumption from one fuel to another rather than simply reducing it. Electricity rises. Gas falls. The building may end up using less energy at source and more at the meter, or the reverse, depending on the plant it replaced and how the grid is supplied. A baseline that was built around a gas-fired plant producing both heat and power no longer describes the building you now operate.

This is a solvable problem, and the protocol already tells you how to think about it. What it does not do is hand you a formula, because the right treatment depends on what you are trying to report. Below is the reasoning an M&V practitioner goes through, the two legitimate paths, and the data you need in place before either one works.

Start with what Option C actually measures

Option C determines savings at the whole-facility meter. The IPMVP generally accepted M&V principles published by the Efficiency Valuation Organization state the framework plainly: savings cannot be measured directly, because savings are the absence of consumption. They are determined by comparing measured consumption before and after, "making suitable adjustments for changes in conditions".

The protocol splits those adjustments in two. Routine adjustments handle factors expected to change during the reporting period, such as weather or production volume. Non-routine adjustments handle factors that are not usually expected to change. The document names examples of the second group: "the facility size, the design and operation of installed equipment, the number of weekly production shifts, or the type or number of occupants".

The protocol is explicit that static factors have to be watched: for factors not usually expected to change, "the associated static factors must be monitored for change throughout the reporting period". If nobody was tracking the plant configuration as a static factor, the adjustment becomes a reconstruction exercise after the fact, which is far harder and far less defensible.

Why fuel switching breaks a single-meter comparison

A gas-fired cogeneration plant does two jobs: it makes heat and it makes electricity on site. Remove it and both jobs move. The heat load transfers to the heat pump, which draws electricity. The electrical load the plant used to serve is now imported. Two things happen to your meters at once.

• Site gas consumption falls sharply, because the plant was the largest gas consumer in the building.
• Site electricity consumption rises on two counts: the heat pump's own draw, and the on-site generation you no longer produce.

If you report savings in raw site energy, the result depends almost entirely on the coefficient of performance you achieve and on how efficient the old plant really was, not on whether the project was a good idea. Report in source or primary energy and you get a different answer. Report in cost and you get a third answer, driven by the spark spread between your electricity and gas rates. Report in emissions and you get a fourth, driven by the carbon intensity of the grid you are now leaning on.

None of those four answers is wrong. They answer different questions. The failure mode is picking one implicitly, by defaulting to whatever the spreadsheet already had, and then defending it later.

Path one: keep Option C and make a non-routine adjustment

This path keeps the whole-facility boundary and adjusts the baseline to reflect the post-project plant configuration. In practice you are answering a counterfactual: what would this building have consumed, on each fuel, during the reporting period, if the cogeneration plant were still running under reporting-period conditions?

Answering it needs an engineering model of the plant you removed, and that model needs to be built from real operating data rather than nameplate figures. At minimum:

1. The plant's measured electrical output and fuel input across at least a full year, at a resolution fine enough to show part-load behaviour. A cogeneration unit at 40 percent load has a very different heat rate from the same unit at full load.
2. The building's thermal demand over the same period, separated from the plant's own parasitic loads.
3. The plant's actual availability, including forced outages and maintenance windows, because the baseline year almost certainly included hours when the plant was down and the building was importing power anyway.
4. The reporting-period weather and occupancy, so the counterfactual is evaluated under the conditions that actually occurred.

IPMVP calls this reporting basis avoided energy consumption, and it is explicit about the mechanics: when savings are reported under the conditions of the reporting period, "baseline period energy needs to be adjusted to the reporting period's conditions". The document even names the practice, noting that "the term forecasting is sometimes used to describe the adjustment of baseline period energy to reporting period conditions".

The strength of this path is that it preserves continuity with whatever you have already reported. Its weakness is that the counterfactual carries every assumption you made about a machine that is no longer there to check. Reviewers know this, which is why the protocol's conservative principle matters: where judgments are made about uncertain quantities, the method should be designed so savings "are not overstated".

Path two: treat it as a new baseline and measure the retrofit isolated

The second path accepts that the building changed identity and stops pretending otherwise. You close out the old reporting stream, establish a new baseline for the post-CHP building, and measure the heat pump's performance with an isolation option rather than at the whole-facility meter.

This is usually the honest choice when the plant removal was not an efficiency measure at all but a decision driven by end of asset life, emissions targets, or a maintenance bill nobody wanted to pay again. In those cases there is no efficiency saving to claim from the swap itself. What you want to know going forward is whether the heat pump is delivering the coefficient of performance it was specified at, and that question is answered by sub-metering the unit and its thermal output, not by watching the main meter.

If a contract or an incentive program requires whole-facility reporting, you may not have this option. Read the agreement before you choose. Some performance contracts define the measurement boundary and the adjustment method in terms that a fuel switch was never written to accommodate, and the time to renegotiate that is before the plant comes out, not after.

The mistake that invalidates both paths

Mixing bases inside one report. A surprising number of post-retrofit reports compare baseline site gas against reporting-period site electricity converted at a different factor than the baseline used, or compare a weather-normalized baseline against un-normalized reporting data. IPMVP's consistent principle asks that reporting be comparable "across different periods of time for the same project", and a quiet change of units or factors between periods breaks exactly that.

The practical defence is boring and effective: state the basis, the conversion factors with their publication dates, the measurement boundary, and the adjustment method on the front page of every report, and do not change any of them without saying so. IPMVP's transparent principle asks for all M&V activities to be "clearly documented and fully disclosed" so an outside reviewer can follow what was done. A reviewer who can reproduce your arithmetic rarely argues with your conclusion.

If you are working through the wider mechanics of Option C baselines, we cover the general case in IPMVP Option C baseline adjustments and the setup work in building an Option C baseline.

What you need in place before any of this works

Every path above assumes one thing: that you can produce complete, reconciled consumption history for each fuel, aligned to the same periods, for the baseline year and the reporting period. In our experience that assumption is where fuel-switching M&V actually fails. The gas account was billed on a different cycle than the electricity account. A month is missing because an invoice was paid but never filed. The plant had its own sub-meter that stopped logging two years before it was decommissioned. One year of data sits in a spreadsheet with a column somebody widened and then re-sorted.

The engineering is the interesting part of this work, but the data assembly is what takes the weeks. Getting utility bills and interval data into one clean, period-aligned record, with the units and the rate components intact, is what makes the adjustment defensible rather than approximate. For the general shape of that work, see how a utility data pipeline should be put together.

Wider context on how M&V fits alongside monitoring and reporting is set out by the Efficiency Valuation Organization, and the US Federal Energy Management Program publishes its own M&V guidelines that follow the same options structure for federal performance contracts.