The central difficulty in demand-side management is that demand and energy are priced and managed differently: a measure can cut consumption substantially while leaving the billed peak intact. The study approach that fits this exam is to practice attaching every proposed saving to a specific interval of the day, using load shapes rather than annualized totals. This guide gives you the diagnostic tools, a decision table, two worked scenarios with common mistakes, and a self-check exercise to make that habit automatic before test day.
Demand vs. Energy: Why a kWh Saving Can Leave the kW Peak Untouched
Demand-side management targets how much power a facility draws at specific times, not just how much energy it consumes. Train yourself to attach every saving to an hour of the day before judging its value.
Start with the definitions and keep them separate. Demand is power — the rate at which a facility draws electricity at an instant or over a short interval, measured in kilowatts. Energy is demand sustained over time, in kilowatt-hours. Many utility bills price the two separately: energy charges accumulate across the month, while a demand charge keys off the single highest interval. A measure that reduces consumption evenly can still leave that highest interval intact, which is why energy-only reasoning misprices demand-side work.
Build a three-question habit for every measure you evaluate. First, which interval does the facility's peak actually occur in, according to the interval data? Second, is the affected equipment running during that interval, and at what output? Third, what happens in the hours after the measure acts — does storage recharge, does a shifted load land somewhere new? A saving that never touches the top interval contributes nothing to demand reduction, no matter how large its kWh total.
Reading Interval Data: Load Factor as a Diagnostic, Not a Score
Load factor — average demand divided by peak demand — reveals the shape of consumption. Low load factor means concentrated peaks with real demand exposure; near-unity load factor means energy measures matter more than peak control.
Load factor is average demand divided by peak demand over a period. As a worked example: a meter records 1,000 kWh in a day, so average demand is roughly 41.7 kW across 24 hours. If the highest 15-minute interval peaked at 120 kW, load factor is about 0.35 — a peaky profile with concentrated, scheduled spikes. A flat 24-hour production profile with the same energy would show a load factor near 1.0, where demand control has little room to act.
Run this exercise on one month of 15-minute interval data from any meter you can access, or on a published sample dataset. Compute load factor for the month, list the ten highest intervals with their timestamps, and note which equipment plausibly runs at those times. Expected observations: if load factor sits below roughly 0.4, the top intervals usually cluster at the same hour on similar days, pointing to one schedulable driver; if it sits above roughly 0.7, savings will come mostly from energy measures instead.
- Self-check rubric: you can compute load factor by hand, without notes.
- You can name a plausible equipment driver for each of the top three intervals.
- You can state whether a proposed measure operates during those intervals.
- You can explain what a rising or falling month-over-month load factor implies about scheduling.
Choosing Between Shedding, Shifting, and Efficiency Measures
The three measure families act on different parts of the load shape: efficiency lowers the whole curve, shedding clips peaks, shifting moves blocks of load in time. Match the measure to the shape you diagnosed in the interval data.
Efficiency upgrades shrink the entire load curve, shedding clips the top intervals while leaving the rest alone, and shifting relocates a block of consumption from high-cost hours to low-cost hours. Storage dispatch is a hybrid: it behaves like shedding while discharging and like shifting when it recharges. Power factor correction reduces kVA, which matters only where a utility bills demand in kVA rather than kW — a billing detail to confirm, never assume.
The families also interact, and the interactions create double-counting traps. A curtailment plan that sheds a specific chiller depends on that chiller still existing; a parallel efficiency project that replaces the chiller erases the shed. Shifting can manufacture a new peak when several loads move into the same off-peak window at once. Before combining savings from two measures, check whether each one's baseline still exists after the other is installed, and whether the shifted load lands outside every top interval.
| Measure family | Effect on load shape | Works best when | Watch-outs |
|---|---|---|---|
| Efficiency upgrade | Lowers the whole curve, including most intervals | Consumption is broad and steady across operating hours | May barely touch the peak if the equipment is idle at the peak interval |
| Peak shedding / curtailment | Clips the top intervals only | One or two identifiable drivers dominate the peak | Requires tolerance for the load loss, or a flexible or stored alternative |
| Load shifting | Moves a block of kWh from peak to off-peak hours | The load is deferrable without disrupting operations | Can create a new peak if loads pile into the same off-peak window |
| Storage dispatch | Sheds the peak while discharging; adds load while recharging | The peak window is short, predictable, and data-verified | Recharge load and depleted state of charge can undo the benefit |
| Power factor correction | Reduces kVA, not necessarily kW | Demand is billed in kVA and reactive load is significant | Irrelevant where demand is billed in kW; confirm the tariff first |
Scenario A: An LED Retrofit That Did Not Touch the Demand Charge
A retrofit justified with demand savings can fail because the peak occurs when the retrofitted equipment is off or lightly loaded. Judge demand savings interval by interval, not by multiplying connected load across all operating hours.
Scenario: a warehouse shows a monthly peak of 800 kW at 17:15 on hot afternoons, driven by cooling, refrigeration, and forklift charging. Lighting runs mainly from 06:00 to 14:00 and contributes about 40 kW at the peak hour. A proposed LED retrofit cuts lighting power by 60 percent. The plausible mistake: the analyst multiplies the full lighting reduction across every operating hour and reports a large demand saving. At the peak interval, lighting drops only 24 kW, so the new peak is roughly 776 kW and the billed demand barely moves.
The better decision is to model the measure against the actual peak interval before claiming anything. Once the model shows lighting touches little of the peak, pair the retrofit with a measure aimed at the real drivers — for example, staggering forklift charger starts or pre-cooling the space before the afternoon window. Why it matters: the retrofit still saves energy and remains worthwhile, but attributing demand savings to it would misstate the project's value and would collapse under any check against post-retrofit peak data.
Scenario B: Dispatching Storage Too Early and Missing the Peak
Storage and curtailment measures depend on timing: discharging before the true peak window exhausts the resource, and the interval data then sets a new peak afterward. Define the dispatch window from the data, not from the clock.
Scenario: a site with a 500 kWh battery — about 250 kW for two hours — targets a peak window that interval data places between 16:45 and 18:30 on hot weekdays. The plausible mistake: an operator starts a two-hour discharge at 14:00 to be safe. By 16:45 the battery is empty, the original peak recurs, and the site pays the demand charge anyway while adding recharge energy to the bill. The early dispatch delivers essentially no demand benefit at all.
The better decision: anchor dispatch to the observed peak window, hold state of charge until it opens, and plan recharge deliberately — a fast recharge immediately after discharge can create a brand-new peak. Spread recharge across low-demand hours or cap its rate. Why it matters: firing inside the window converts a wasted dispatch into the full demand reduction the hardware was sized for, and the recharge load is part of the measure's true cost, not a footnote to be ignored.
Documenting Demand Savings So They Survive Scrutiny
Demand savings are only defensible when documented against a defined baseline load shape, a stated peak-window definition, and the post-period's actual peak interval. Write the measurement method down before the measure is installed.
Monthly peaks are noisy: they move between hours and seasons as weather and schedules change, so one comparison month proves little. A defensible package contains the baseline interval data, the rule used to define the peak — for example, the single highest 15-minute interval per billing period — any adjustments for changed operations, and post-period interval data showing what the new peak interval looks like. Persistence matters too: check whether the peak stays suppressed across several months rather than in one favorable one.
Also record which peak you are claiming. A facility peak is the site's own maximum; a coincident or system peak is when the grid itself tops out, and demonstrating reduction of the two requires different evidence. Keeping these distinct in your documentation is a habit worth building now, because both exam scenarios and real programs hinge on which peak a measure is supposed to affect.
A Preparation Sequence and Concrete Readiness Checks
Structure preparation around load shapes and decisions rather than isolated facts: re-derive the arithmetic, annotate real data, classify measures with the table, then rehearse written decisions against scenario datasets under time pressure.
A sequence you can adapt: spend the first study block re-deriving the kW, kWh, and load-factor relationships by hand until the arithmetic is automatic. Next, plot and annotate one full month of interval data, marking peaks, drivers, and windows. Then classify a list of candidate measures using the shedding, shifting, and efficiency table. After that, write two full scenario decisions — one retrofit, one dispatch — naming the mistake you are avoiding. Finish with timed practice questions from the free CDSM practice page, tagging every error to a concept.
You are ready when you can do all of the following without notes: compute a load factor and name its drivers; state which intervals a proposed measure touches; explain why a large kWh saving can leave a demand charge unchanged; define a storage dispatch window from interval data; and keep facility and coincident peaks separate in an explanation. For administrative details of the credential itself — eligibility, scheduling, and current requirements — rely on the Association of Energy Engineers directly rather than secondary summaries, including this one.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.