The working discipline behind energy auditing is comparability: a savings figure only means something if the baseline and the proposed case describe the same facility, the same weather, the same occupancy, and the same operating hours. This guide builds that discipline across audit levels, bill analysis, demand versus energy, end-use estimation, and documentation, then closes with a four-week practice sequence, two worked scenarios, and a readiness checklist you can score yourself against.
Audit Levels: Matching Depth to the Decision at Hand
Energy audits are commonly described in graduated levels, from a brief walk-through to an investment-grade analysis. Each level answers a different question, and choosing the wrong depth produces either a useless report or wasted effort.
A preliminary walk-through identifies obvious opportunities: visible lighting waste, leaking steam, unmaintained schedules. A general audit adds utility analysis and quantified recommendations with typical assumptions. An investment-grade audit supports a capital decision, so it demands measured end-use data, refined cost estimates, and stated uncertainty. Compare each level against the decision it must support, not against the others.
When you study scenarios, practice naming the level a client actually needs. A plant manager deciding whether to replace a chiller next year needs investment-grade rigor on that measure and little more than identification on the rest. Applying deep analysis everywhere inflates cost; applying shallow analysis to a capital decision undermines trust in the whole report. Depth should be proportional to the financial commitment.
Use this table to sort any scenario description into the right level before you begin calculating.
- Ask: what decision will be made from this report, and by whom?
- Match measurement effort to the size and risk of the recommendation.
| Audit level | Typical scope | Typical output | Best suited decision |
|---|---|---|---|
| Walk-through / preliminary | Brief site survey, bill review | List of visible opportunities | Whether a fuller audit is worth doing |
| General / targeted | Utility analysis, spot measurements | Quantified ECMs with typical assumptions | Budgeting, prioritizing low-cost measures |
| Investment-grade | End-use measurement, refined costs | Defensible savings and cost estimates | Capital approval, performance contracting |
Normalizing Utility Bills for Weather and Usage
Raw bills mix weather, schedules, and baseload into one number. Normalization separates these influences so that a comparison between two periods reflects performance change, not just a hotter summer or a busier month.
Start by splitting consumption into baseload, the weather-insensitive floor visible in mild months, and weather-sensitive load, estimated against heating or cooling degree days through regression. If monthly consumption rises one year, the first question is whether degree days, operating hours, or equipment changed. Only after ruling out those explanations can you attribute the difference to efficiency. This ordering is the core analytical habit to drill.
Worked scenario (bill normalization). A candidate reviews a school's bills and sees July consumption 18% above the prior July. Mistake: reporting 'the school wasted energy this summer.' Better decision: first check cooling degree days, which rose about 15%, then check whether a summer program added occupied weeks. The residual difference is small and plausibly explains itself. Why it matters: an unnormalized comparison can trigger an investigation, or a claimed 'saving,' that the weather account for on its own.
Practice on any two-period bill set: divide the change into weather, schedule, and unexplained components, and refuse to interpret the unexplained piece until the first two are quantified.
Energy Versus Demand: Two Different Dollars on One Bill
Energy (kWh) is total consumption over time; demand (kW) is the peak rate of use in a billing interval. A retrofit can cut one, both, or neither, so every recommendation should state which of the two it moves.
Load factor, roughly average demand divided by peak demand, tells you how flat a load profile is. Lighting controls and equipment shutoffs cut kWh; peak-shaving strategies, load scheduling, and power-factor correction target the kW charge. A variable-frequency drive on a fan running mostly at part load cuts large amounts of energy but only trims demand if it also reduces draw during the facility's peak interval. Confusing the two produces savings claims billed on the wrong line.
Worked mini-example (labeled practice numbers). A facility uses 60,000 kWh in a 720-hour month with a 150 kW peak. Load factor = 60,000 / (150 × 720) ≈ 0.56. If scheduling shifts a 30 kW batch process out of the peak window, kWh may barely change, yet the demand charge drops. If instead occupancy controls cut lighting runtime 20%, kWh falls meaningfully while the coincident peak may barely move. Sketch both effects on a load profile before estimating dollars.
For every efficiency measure you study, label it explicitly as an energy measure, a demand measure, or both, and state which billing line your estimate belongs on.
Quantifying Savings Without Mixing Baselines
A savings estimate is the difference between an adjusted baseline and actual or projected post-case consumption under matched conditions. The widely used IPMVP framework formalizes this: adjust the baseline for routine changes, isolate the effect, and state the method.
In scenario-based reasoning, the baseline is the reference case: what consumption would have been without the measure, adjusted for weather, occupancy, and production. Retrofit isolation can come from metering the affected system alone, from whole-facility regression, or from calibrated simulation. Each approach trades measurement cost against certainty. Recognizing which option a scenario's data supports, and saying so explicitly, is the analytic move to practice.
Worked scenario (baseline adjustment). A candidate compares a warehouse's raw bills before and after an LED retrofit and reports a 25% saving. Mistake: the second year added a refrigerated section and more operating hours, so raw bills are not comparable. Better decision: adjust the baseline upward for the added load and extended schedule, then compare; the lighting-specific saving, metered on the lighting circuit or estimated from fixture counts and runtime, is isolated and smaller but defensible. Why it matters: mixing baselines makes the number unreproducible, and an overstated saving destroys the credibility of the entire report.
Build the habit of writing one sentence per savings claim naming the baseline, the adjustments made, and the isolation method used.
End-Use Estimation: Where Rated Power Betrays You
End-use estimates combine connected power, duty cycle, diversity, and runtime. The recurring trap is treating nameplate ratings as actual draw, and assuming every installed fixture or motor runs whenever the building is open.
For lighting, connected load equals fixture count times wattage, but consumption requires measured or scheduled runtime and controls. For motors, nameplate horsepower is a rating, not a load; actual draw depends on the driven load and typically must be measured or estimated from load factor. Compressed air leaks are a continuous load that persists outside production hours. Steam trap failures waste energy even when production is idle. Each system needs its own treatment of when, and how hard, it actually runs.
Worked mini-example (labeled practice numbers). Fifty 60 W fixtures replaced by 20 W fixtures, running 10 hours a day: saving = 50 × 40 W × 10 h = 20 kWh/day before controls effects. If occupancy sensors halve effective runtime, the saving roughly doubles. Compare that with a 30 kW motor at nameplate: if the driven load holds it at 70% of rated input, the true draw is about 21 kW, and using the nameplate figure overstates the baseline by 9 kW. Sketch the duty cycle first; the arithmetic is secondary.
Practice estimating from the three questions in order: what is connected, when does it actually run, and how close to rating does it operate.
Documentation: Making Every Number Traceable
An audit report must let a reader reconstruct each savings figure. That means stated assumptions, identified sources, named methods, and acknowledged uncertainty, organized so the numbers can be checked without asking the author.
Compare two report styles. Style one lists measures with savings totals and no visible assumptions. Style two gives, per measure, the baseline condition, the data source, the calculation path, the assumptions on runtime and rates, and the estimate's confidence level. Only the second can survive review, negotiate a retrofit contract, or support a measurement and verification plan. When you study case scenarios, grade them on whether the numbers are reproducible, not whether the recommendations look right.
Documentation also extends to the audit process itself: dated observations, photographs tied to locations, instrument readings with the instrument identified, and utility data with the billing periods clearly labeled. In scenario questions, watch for plausible reports that omit the billing period or the assumptions. Those omissions are the substantive content, because a number without its reference conditions cannot be compared, verified, or trusted later.
Rebuild one example report from a practice case and check whether a second person could reproduce each figure from your stated inputs alone.
A Four-Week Preparation Sequence and Readiness Checks
Prepare in layered passes: concepts first, calculation discipline second, scenario synthesis third. Self-check scores are learning milestones, not predictions of any exam outcome.
A realistic adaptable sequence: week one, audit levels and the baseline concept, summarizing each level and when it applies. Week two, bill analysis and normalization, working degree-day regressions and load-factor calculations by hand. Week three, end-use estimation for lighting, motors, compressed air, and steam, always with duty cycles sketched first. Week four, full case scenarios: choose an audit level, normalize the baseline, estimate savings, and write the traceability sentence for each claim. Adjust the pacing to your prior exposure.
Practical exercise with expected observations: take one building's twelve monthly bills (public facility datasets or a colleague's anonymized bills work) and produce (a) a baseload estimate from the mildest months, (b) a degree-day regression, (c) a load-factor calculation per the labeled method above, and (d) one normalized year-over-year comparison. Expected observations: the baseload appears as a flat floor; regression residuals grow in months with schedule changes; the normalized difference is smaller and more defensible than the raw difference. Self-check rubric, scored 0 to 3 each: baseline adjustment named, units consistent, demand versus energy correctly assigned, assumptions stated, arithmetic reproducible. Reaching roughly 12 of 15 signals solid analytical fluency; lower subscores point to the specific layer to revisit.
Readiness checks before test day: you can place a scenario in the correct audit level; you can decompose a bill change into weather, schedule, and unexplained parts; you can label every measure as energy, demand, or both; you can write a one-sentence traceability statement for a savings estimate; you can explain why nameplate power is not measured draw. If any check fails, return to the matching section rather than rereading broadly. For administrative details about the credential itself, including current eligibility and scheduling information, consult the issuer directly at the Association of Energy Engineers website; this guide covers study content only.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.