Prepare for the CEM by practicing the full reasoning chain, not isolated facts: convert all fuels to a common unit before comparing, read the utility rate structure before proposing measures, state a baseline before estimating savings, and match each measure to an analysis method whose assumptions you can state out loud. Work every practice problem by writing the unit chain and the assumption list before the arithmetic.
Site versus source energy: why the same fuel never compares directly
Site energy is what crosses the facility meter; source energy adds upstream generation and delivery losses. Compare fuels only after converting everything to a common unit, and label which basis you are using before evaluating any equipment swap.
A recurring calculation trap is adding quantities that were never in the same currency. One therm is 0.1 MMBtu, one kWh is 3,412 Btu or about 0.003412 MMBtu, and a gallon of propane carries roughly 91,500 Btu on a site basis. When a problem gives gas use in therms and electric use in kWh, convert both to MMBtu first, then apply efficiency factors. Writing the conversion in the margin before computing prevents the classic error of comparing 50,000 kWh against 50,000 therms as if they were similar magnitudes.
The site-versus-source distinction also changes equipment comparisons. An electric chiller and a gas-fired absorption chiller can look competitive on site Btu, yet source-based analysis credits the electric unit differently because power-plant conversion losses sit upstream of the meter. Neither basis is universally correct; the useful skill is recognizing which basis a scenario assumes and stating it. Practice rewriting a mixed-unit problem as: annual electric MMBtu (site) equals X, annual gas MMBtu (site) equals Y, then decide whether the question's comparison needs source adjustment.
A useful note: administrative details such as scheduling and eligibility are set by the Association of Energy Engineers and should be confirmed on their site.
Demand charges and load factor: read the bill before proposing the measure
Utility bills contain two separable costs: an energy charge on total kWh and a demand charge on peak kW, sometimes with power-factor adjustments. Analyze the demand pattern first, because a measure aimed at the wrong driver cannot produce the claimed savings.
Worked scenario 1: A plant's bill shows 324,000 kWh in a 720-hour month with a 900 kW peak, and a demand charge of $18 per kW. The load factor is 324,000 divided by (900 times 720), which is 0.5 — the average draw is half the peak. A plausible mistake here is proposing a lighting retrofit as the demand solution: a 20 kW lighting reduction does nothing if the 900 kW peak occurs when chillers start simultaneously on the hottest afternoon. The lighting kWh savings are real, but the demand-charge savings attributed to them are not.
The better decision starts with interval or hourly data: identify when the peaks occur and which equipment runs then. Staggering chiller start-up or pre-cooling might trim 100 kW from the peak, worth about 100 times $18, or $1,800 that month — an order of magnitude larger than the lighting contribution to peak. This matters because a measure-sized-to-the-wrong-driver analysis looks complete on paper but collapses under review, and the same pattern appears in practice problems: always ask what equipment defines the peak, not what equipment uses the most total energy.
Load factor also frames the opposite case: a load factor near 0.9 signals a flat profile where demand charges barely respond to any single measure, and energy charges dominate the economics.
Baselines and degree days: separating weather effects from real savings
A savings claim is the difference between a stated baseline and post-retrofit use. Because heating and cooling track outdoor temperature, normalize energy use to degree days before concluding that a measure worked or failed.
The baseline is the reference consumption the improvement is measured against, and it must reflect conditions — occupancy, schedule, weather — as they actually were, not a single static annual figure. The audit discipline behind this runs from a walk-through survey that builds an equipment and load inventory up to a detailed audit with measurement and end-use breakdowns. In practice problems, check whether the stated baseline includes the same production level or set of operating conditions as the post case; a baseline at 70 percent occupancy makes any comparison to a fully occupied year meaningless.
Degree-day regression is the standard normalization tool: plot monthly fuel use against heating degree days for the same months, and the slope approximates the weather-sensitive load while the intercept approximates the base load. If a boiler retrofit's savings estimate ignores a milder-than-normal winter, the estimated savings will not appear in the bills, and the discrepancy has nothing to do with the equipment. Practice problems often provide just enough monthly data to do this; the disciplined move is computing the regression relationship, or at least comparing a mild month against a matching-degree-day historical month, before quoting a savings percentage.
Motor and pump upgrades: where the cube-law savings model breaks
For centrifugal fans and pumps with no static head, power follows flow cubed, so 80 percent flow suggests roughly half the power. That shortcut fails when static head, system curves, or control methods dominate, and applying it blindly overstates savings.
Worked scenario 2: A 30 kW chilled-water pump runs constantly, and a proposal claims that throttling to 80 percent flow with a variable-frequency drive cuts power to 30 times 0.8 cubed, about 15 kW, saving half the input. The plausible mistake is accepting the cube law without checking the system: if 40 feet of static head must be overcome regardless of flow, the required pressure at 80 percent flow does not fall to 80 percent of the original, and the pump cannot ride down its full curve. Actual power might be 20 kW or more, not 15 kW.
The better decision applies the model only when its assumptions hold: a mostly friction-dominated system with no significant static lift and flow control that lets the pump follow its curve. When static head or constant-pressure control is present, use the pump curve against the system curve, or better, measured kW-versus-flow data. This matters because VFD proposals are frequently built on the cube law alone, and a savings figure that a reviewer can disprove with one system-curve sketch damages the credibility of the entire report. Train yourself to write the assumption — friction-dominated, no static head, flow proportionality — directly above any cubed calculation.
The same discipline extends to fan retrofits: variable-speed savings on a system with fixed-inlet dampers, or where flow barely varies in practice, follow measured data, not the theoretical curve.
Payback, SIR, and life-cycle cost: choosing the right screening metric
Simple payback, net present value, savings-to-investment ratio, and life-cycle cost answer different questions and rank projects differently. Match the metric to the decision at hand rather than defaulting to payback for everything.
Simple payback divides installed cost by first-year savings and ignores both project life and the time value of money. A controls project with a two-year payback and a five-year life, compared against a long-lived measure with a seven-year payback, can rank in either order once full lifetimes and escalation are considered. Life-cycle costing and net present value capture those dimensions; the savings-to-investment ratio expresses benefit per dollar invested and scales cleanly across measures of different sizes.
Practice by re-ranking the same project set under different metrics and noticing where the ordering flips. In scenario work, a useful check is whether the recommended metric matches the measure's lifetime and whether maintenance, energy escalation, or replacement costs appear where the metric requires them.
A short comparison to anchor your choices:
Note that study-scope administrative specifics live with AEE; the analysis methods above are general energy-management practice and the framework to rehearse.
| Metric | What it answers | Best suited for | Blind spot |
|---|---|---|---|
| Simple payback | How fast is the cash recovered? | Quick screening of low-cost measures | Ignores life, escalation, time value of money |
| Net present value | What is total value in today's dollars? | Ranking projects with different lives | Absolute dollars favor large projects over efficient ones |
| Savings-to-investment ratio | Benefit per dollar invested? | Comparing measures across size ranges | Needs an agreed analysis period and discount rate |
| Life-cycle cost | What is the lowest total ownership cost? | Choosing between equipment options | Requires full cost streams: energy, maintenance, replacement |
M&V options: deciding what to measure and what to stipulate
Verification approaches range from measuring a single device's performance with an agreed routine (Option A) to continuous whole-facility billing analysis (Option C). Choose based on how much savings matter, how isolable the measure is, and what data collection costs.
The conceptual core is isolation. A lighting retrofit in one wing, with hours and fixture wattage easy to specify, fits an approach where key parameters are measured and others stipulated by agreement. A plant-wide controls upgrade that shifts loads across many systems fits a whole-facility approach using utility billing data with regression adjustments, because the savings cannot be isolated at any single device. A retrofit-isolation approach measuring the affected system fits something in between; a calibrated simulation serves complex cases where measurement alone cannot attribute the effect.
The practical mistake is mismatching rigor to the measure: billing-based analysis on a small retrofit drowns the signal in weather noise, while stipulating parameters on a large interactive project invites disputes. In scenario practice, state for each measure which option isolates its effect most cheaply, and name the routine or measurement that supports any stipulated value. Savings confidence, cost of verification, and the interaction between measures all belong in that justification.
Tie this back to the baseline section: whole-facility verification is precisely where degree-day normalization stops being optional and becomes the adjustment mechanism in the savings equation.
A six-week sequence with a bill-analysis self-check you can grade
Structure preparation in three passes: concepts and unit fluency, applied scenarios on rates and systems, then full audit-style case work with financial screening and verification choices. Grade yourself with a rubric, not a feeling.
Weeks one and two: unit conversions, load factor, demand-versus-energy charges, and site-versus-source reasoning; compute every value with an explicit unit chain. Weeks three and four: system-level scenarios — HVAC, motors, lighting, steam — each time writing the assumption list above the calculation and screening the result with two different financial metrics to see whether the recommendation survives both. Weeks five and six: end-to-end case analysis, from baseline and degree-day normalization through a measure list, savings estimate, screening, and a verification approach chosen and justified per measure.
Practical exercise: obtain twelve months of bills for a building you can observe, or construct a plausible twelve-month set. Compute annual energy use, load factor, and the top three demand months; regress energy against heating degree days; propose one measure; estimate its savings with a stated unit chain and assumption list; and choose a screening metric and verification approach with justification. Expected observations: load factors between roughly 0.4 and 0.7 for a typical commercial profile, a visible weather slope in winter months for heated buildings, and demand peaks concentrated in a small number of hours that total-energy reasoning would miss.
Self-check rubric — every item should be verifiable on your own work:
Readiness checks before you stop: compute load factor from a raw bill without notes; state out loud the two assumptions the cube law needs and one real system that violates them; convert a mixed fuel bill to MMBtu and explain when a source-basis adjustment is warranted; and for three different measures, name the verification option you would choose and the alternative you rejected. These are learning milestones, not predictions of any score.
- Units: every result carries correct, consistent units with conversions shown
- Baseline: stated explicitly, with weather or operating conditions addressed
- Peak analysis: the measure targets the equipment that defines the demand peak
- Assumptions: model limitations, such as cube-law conditions, written above the math
- Metric fit: the financial metric matches the measure's lifetime and the decision being made
- Verification: the approach isolates the measure at a cost proportional to its savings
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.