Four Metrics, Four Answers: Reporting an Electrification Retrofit
A 250,000 ft² mixed-use building (office over ground-floor retail) is gas-heated with condensing boilers and electrically cooled. The owner's RFP states a goal of 'a 30% reduction in energy use.' A proposed all-electric retrofit - VRF heat pumps with heat recovery, LED lighting, and a DOAS with 70% sensible recovery - is modeled against an existing-conditions baseline. Results: site energy −22%, source energy +9%, annual energy cost +14%, and CO2e −38%. The modeler used the state annual average grid factor of 0.25 kg CO2e/kWh and a national electricity source-site ratio of 2.8; cold-climate VRF performance including resistance backup averaged an effective heating COP of 1.7. The LED measure alone cut cooling energy 9% but raised heating energy 6%.
1. The RFP says '30% reduction in energy use.' How do you report these four results, and what do you tell the owner?
Read the worked answer
Report all four in one table and refuse to lead with a single number. Each metric answers a different question:
- Site energy (−22%) - delivered energy at the meter, counting a kWh and a therm as equivalent. This is the metric most codes and programs use, and it is structurally hostile to electrification: replacing 1 kWh of gas heat with ~0.6 kWh of electricity reads as a large site win, even when the grid does more work upstream.
- Source energy (+9%) - site energy times a conversion factor. Here electricity carries a 2.8 factor, so the heat pump's effective source performance (2.8 ÷ 1.7 ≈ 1.65) is worse than the boiler's (1.05 ÷ 0.9 ≈ 1.17). Source energy is the honest answer to 'did we reduce primary fuel demand,' and the answer is no.
- Cost (+14%) - what the owner actually pays. Driven by the local gas-to-electric price ratio.
- CO2e (−38%) - because the grid is comparatively clean, a kWh of heat from this heat pump emits less than a kWh of heat from a 90%-efficient gas boiler.
The honest headline: the project meets a 30% goal only on carbon, which the RFP almost certainly did not mean, and fails it on site energy - the metric the phrase 'energy use' usually denotes. The answer to 'did we hit the goal?' is metric-dependent.
Recommendation to the owner: restate the goal in the metric that actually matters to them before awarding the work, and note that a building with an all-electric or heat-pump baseline would show a completely different (and much more favourable) site energy picture. Reporting only the −22% or only the −38% is advocacy, not analysis.
2. Explain to the owner why cost rises 14% while carbon falls 38%. Show the arithmetic.
Read the worked answer
The two metrics are measuring different things: cost looks at the tariff, carbon looks at the generator.
Per unit of delivered heat (illustrative rates):
| Electricity at $0.14/kWh, COP 1.7 | Gas at $0.90/therm, 90% AFUE | |
|---|---|---|
| Cost per kWh of heat | $0.14 ÷ 1.7 = $0.082 | $0.90/therm ÷ 29.3 kWh/therm ÷ 0.9 = $0.034 |
| CO2e per kWh of heat | 0.25 ÷ 1.7 = 0.147 kg | 0.181 ÷ 0.9 = 0.201 kg |
So a heat pump with an effective seasonal COP of 1.7 - dragged down by cold-climate capacity loss and resistance backup - costs about 2.4× as much per unit of heat as this cheap gas, but emits about 27% less. That single line explains the divergence: gas is cheap and relatively carbon-intensive; electricity here is expensive and relatively clean.
The building-wide +14% cost and −38% carbon come from that heating trade plus the non-heating measures. The lighting and DOAS work cut cooling energy and cooling cost, which offsets part of the heating penalty; carbon benefits more from those measures than cost does because the whole-building carbon total is more dominated by the heating fuel swap.
The interaction to flag explicitly: the LED measure cut cooling 9% but raised heating 6%. Reduced internal gains are a real winter penalty in a heating-dominated building, and a model that reports the cooling savings without the heating penalty is overstating the package. Both effects are already netted in the numbers above; make sure the owner understands they are netted, because the design team will ask why heating went up after an efficiency upgrade.
Sensitivity worth running: the conclusion flips if gas is priced higher, if the heat pump achieves COP 2.3 instead of 1.7, or if demand charges change under an all-electric profile. Report the cost result with those levers identified rather than as a single verdict.
3. Critique the carbon accounting method. What sensitivities would you run before this number leaves the office?
Read the worked answer
Critique of the method as run:
- Annual average factor masks timing. A heat pump runs hardest on winter mornings, when a grid with substantial solar on it is being served by marginal fossil generation. The annual average of 0.25 can understate or overstate the true marginal emissions of the delivered kWh, often by a factor of two at those hours. Ask whether the program permits an hourly or marginal method; if so, use it.
- Average vs. marginal is a policy choice, not a physical one. An average factor answers 'what is the grid's intensity'; a long-run marginal factor answers 'what does one more kWh cause to be built or dispatched.' State which you used and why, because the two give different answers and both get published.
- Refrigerant is missing. A heat-recovery VRF using R-410A (GWP roughly 2,000) with 1–4% annual leakage can add a meaningful fraction of the HVAC carbon over a 10-year horizon. Omitting it flatters the electrification case. Estimate it and disclose it separately.
- Baseline gas factor. Confirm you used full-fuel-cycle or CO2-only factors consistently on both sides; mixing a CO2-only gas factor against a full-fuel-cycle electric factor is a common and silent distortion.
Sensitivities to run and show:
- Grid factor at low and high bounds (e.g., 0.15 and 0.45 kg CO2e/kWh) to test whether 'carbon improves' survives a dirtier grid.
- With and without refrigerant leakage over a 10-year horizon.
- Heat pump effective COP at the modelled value and at a degraded value reflecting real cold-climate performance.
If the carbon conclusion holds across that range, say so - that is a strong result. If it flips at the high grid factor, the owner needs to know before this number goes into a public commitment. Document the factor source, vintage, and method in the assumptions log.