IPMVP Options A through D explained

IPMVP — the International Performance Measurement and Verification Protocol — is how you prove an ECM actually saved what it was modelled to save. Four options, each with a specific use case, each with its own data and cost trade-offs. Picking the wrong option is the most common reason M&V fails in practice.

Why M&V is hard

After an ECM is implemented, you need to prove the savings. The naive approach — compare the next year's energy bill to last year's — fails almost always: production volume changed, weather changed, occupancy changed, operating hours changed. The savings get lost in the noise.

IPMVP gives four formal options for isolating the ECM's effect from everything else that's moving in the plant. Each option is rigorous enough to defend against a regulator (Hydro-Québec, NRCan), an investor (capital reviews), or a sceptical CFO.

The four options at a glance

Option A — Retrofit isolation, key parameter measured

Use it when: the ECM affects one easily-measurable parameter and the rest of the system stays roughly the same.

Example: a VFD retrofit on a single pump. You measure the post-retrofit run-time (the "key parameter") and combine it with a spec-sheet kW for the pump at the new operating point. Pre-retrofit kW comes from nameplate or short-term spot measurement.

Pros: low setup cost, minimal sensors, fast to commission.

Cons: savings depend on a stipulated value (the un-measured kW). Vulnerable if the operating point drifts. Not always defensible against a strict regulator.

Option B — Retrofit isolation, all parameters measured

Use it when: you want individual-ECM rigor and the ECM is bounded enough to instrument (a single motor, a single boiler, a single compressor).

Example: a heat-recovery unit on a kiln exhaust. Install flow meters and temperature sensors before and after the heat exchanger; compute recovered heat directly from m·Cp·ΔT. Compare against baseline (no recovery, all heat lost to stack).

Pros: fully measured, defensible against any reviewer.

Cons: instrumentation cost. Most-defensible Option B installations cost $5K–$25K in sensors, depending on ECM scope.

Option C — Whole-facility regression

Use it when: multiple ECMs land at the same time, or when plant-wide effects (occupancy, production scheduling) make isolated M&V impractical. This is what Opnor EMIS uses for portfolio-level verification.

How it works: build a regression model relating whole-facility energy to independent variables (heating/cooling degree days, production output, operating hours). Calibrate on 12 months of pre-implementation data. Project the model forward into the post-period — the difference between predicted and measured is the savings.

Required: R² ≥ 0.75 on the baseline model. Below that and the model isn't a defensible reference.

Pros: captures cumulative ECM effects without isolating each one. Works on real utility-bill data — no extra sensors.

Cons: can't attribute savings to individual ECMs, only to the program as a whole. Needs a clean "static factors" period between baseline and reporting (no major equipment changes other than the ECMs).

Option D — Calibrated simulation

Use it when: baseline data doesn't exist (new construction), the plant has just gone through a major reconfiguration, or the ECM is a behavioural / operational change with no clean isolation possible.

How it works: build a building-energy or process simulation model, calibrate it to a measured slice of operating data, then run two simulations: one with ECM, one without. The simulation difference is the savings.

Pros: only option when baseline data isn't available.

Cons: most expensive, longest setup. Defensibility depends heavily on calibration quality. Reviewers scrutinize it more than Options A–C.

How to pick the right option

The simplest decision tree we use during scoping:

• Multiple ECMs landing in a 12-month window with no clean isolation? → Option C (whole-facility regression).
• One ECM, easy to instrument, want individual rigor? → Option B.
• One ECM, instrumentation cost matters more than rigor? → Option A.
• No clean baseline period? → Option D.

Most Opnor engagements end up Option C for the program-wide view, with Option B selected for 1–2 specific high-value ECMs that need individual defensibility (typically the most expensive one or the one a sceptical executive wants to see proven).