Three Weeks, One Meter: Scoping a Level 2 Audit with a 46% Base Load
A 14,000 m² site (5,000 m² office, 9,000 m² light assembly) built in 1978 consumes 2,100 MWh of electricity and 3,600 MWh of natural gas per year, with a 480 kW peak demand. There is no sub-metering, the BAS retains only 30 days of trend data, and five years of monthly utility bills are available. You have one auditor and one technician on site for three weeks. The client wants an ASHRAE Level 2 audit plus an ISO 50001-ready energy baseline by significant energy use. A 02:00 Sunday reading at the main meter shows 110 kW. The facilities manager insists the site 'hasn't changed in years' and wants the walkthrough to start with lighting because 'that's where the savings are.'
1. How would you sequence the three-week field effort? State what you measure directly, what you sample, and the reasoning behind the split.
Read the worked answer
Do the desk analysis before the site visit - it decides the field plan.
- Base load = 110 kW × 8,760 h ≈ 964 MWh, i.e. 46% of the 2,100 MWh purchased.
- Load factor = 2,100,000 / (480 × 8,760) ≈ 0.50 - a large, persistent base relative to peak.
Field allocation
- Direct measurement (highest uncertainty, highest dollars): install 15-minute loggers on the main switchboard and the 6–8 largest distribution boards on day 1 and leave them in place for the whole three weeks, so the data covers full production weeks plus two weekends. This is the only way to split the 110 kW base load into HVAC, compressed air, server/IT, pumps and exterior lighting.
- Stratified sampling (bounded uncertainty): lighting by fixture type and area, counted from drawings and verified by walkthrough plus a spot illuminance check; motors by nameplate plus operating-hours interviews; envelope by spot inspection.
- Interviews and spot checks: hours of operation, setbacks, weekend practice, control sequences.
Don't spend expensive field days measuring loads whose annual energy is already bounded to within a few percent by nameplate and hours. Spend them where the answer is unknown - the base load.
ISO 50001 baseline: build it from 12 months of billing plus relevant variables (HDD/CDD, production tonnes), not from three weeks of spot data.
2. The facilities manager wants lighting first. Quantify the maximum plausible lighting savings, compare it with the base-load finding, and describe how you redirect the scope without contradicting him.
Read the worked answer
Lighting estimate (order of magnitude, no data needed):
- Installed load ≈ 5,000 m² × 12 W/m² + 9,000 m² × 8 W/m² ≈ 132 kW
- At ~3,000 h/yr → ≈396 MWh, about 19% of electricity
- A 40% LED-plus-controls saving → ≈158 MWh ≈ $19,000/yr at $0.12/kWh
That is real money and it should be in the report. But note its uncertainty is low - a fixture count and a light meter in half a day gives a defensible number.
Compare with the base load: even if only 25 kW of the 110 kW night load is avoidable, that is 219 MWh/yr, plus demand savings, and today nobody knows what it is. Uncertainty there is very high.
Redirecting the client:
- Agree with him publicly - lighting is a genuine measure, and commit to a firm number by day 3 using a sample-based fixture count.
- Show him the 110 kW Sunday reading and the 46% figure as a fact, not an opinion: 'lighting is 19% of your electricity and we'll price it this week; the other 46% is invisible and that's where I need the three weeks.'
- Trade the freed days for logger deployment and weekend surveys.
You are not disagreeing about lighting; you are disagreeing about where the unknown is.
3. The client asks for savings estimates accurate to ±5% of total site energy. How do you respond, and what do you offer instead?
Read the worked answer
Refuse the number, explain why, and offer a defensible alternative.
- With no sub-metering and 30 days of trend data, the end-use split carries roughly ±15–25% uncertainty. Any savings figure inherits that.
- If savings are ~15% of consumption, then ±5% of total site energy means ±33% of the savings - i.e. the client is asking for a precision that is actually looser than it sounds, but the report must say which base it is on.
- Even under IPMVP, savings uncertainty is driven by baseline model uncertainty and by whether the independent variables (production, weather, occupancy) are stable.
What to offer instead
| Deliverable | Achievable accuracy |
|---|---|
| Screening estimate, all measures, Level 2 | ±25–30% of measure savings |
| Top 3 measures by dollar, after 2–4 weeks of targeted metering | ±10–15% |
| Retrofit isolation with permanent sub-metering (IPMVP Option B) | ±10% |
Also flag that an ISO 50001 baseline is a different product from a savings forecast: it needs 12 months of data, defined static factors, and a documented adjustment method. Deliver the audit and the baseline as separate, labelled deliverables so nobody later confuses an estimate with a measurement.