Reading load profiles: baseload, peak and load factor

A monthly bill total tells you what a building cost. A load profile tells you why. Plot power draw against time and the shape of the day appears: the floor the building never drops below, the spikes when everything runs at once, and the long stretches in between. Three numbers summarize that shape, and each points at a different lever on the bill.

This guide covers baseload, peak demand and load factor: what they are, how to read them off interval data, and what each one is telling you to do next.

Baseload: the floor the building never leaves

Baseload is the minimum power a building draws even when it is closed and mostly empty, visible as the flat bottom of an overnight or weekend profile. It is refrigeration, servers, standby equipment, exterior lighting and controls that never switch off. Baseload matters because it runs every hour of every day, so a small reduction in the floor compounds across thousands of hours.

A baseload that creeps up over time, or that sits surprisingly high relative to occupied hours, usually means something is running when it should not be. Reading it well is the fastest way to find waste that no one is watching.

Peak demand: the spike you pay dearly for

Peak demand is the highest short-interval power draw in a billing period, and for most commercial and industrial accounts it is billed separately from energy. The charge is set on your highest interval, commonly a 15-minute average), not on how much energy you used overall. Two buildings can consume identical kilowatt-hours in a month and pay very different bills if one concentrates its use into sharp peaks.

That is why the shape matters as much as the total. A single afternoon when heating, cooling and equipment all ramp together can set a peak that drives the demand charge for the whole month, and under a ratchet clause it can raise the billed minimum for months afterward. For the full mechanics of how these charges work, see our guide on demand charges.

Load factor: one number for how efficiently you use the grid

Load factor ties the two together. It is the ratio of average load to peak load over a period, or equivalently your total energy divided by what you would have used had you run at your peak the entire time. A common way to compute it from a bill is total kilowatt-hours divided by peak kilowatts, then divided by the number of days and by 24 hours.

A high load factor means steady, predictable use that spreads cost across many hours. A low one means the bill is being driven by short, sharp peaks. As a rough guide, a load factor above about 80 percent is strong, 50 to 65 percent is average, and below 50 percent is low, though what counts as good depends entirely on the operation.

The table is not a scorecard. A round-the-clock facility earns a high load factor by running constantly, and a restaurant with meal-rush peaks may sit low and be operating exactly as it should. The value is in the trend: a load factor that falls over time signals that peaks are growing faster than usage, which is a warning that demand charges are about to climb.

Reading a profile end to end

Put together, a single profile tells a short story. Start at the bottom to read baseload, and ask whether the overnight floor is higher than it should be. Look at the top for the peaks, and note when they happen and what is likely running. Then compute load factor to see whether the day is steady or spiky. Each answer points somewhere different.

1. High baseload: hunt for equipment running outside occupied hours
2. Sharp isolated peaks: stagger startups and look at demand response or storage
3. Falling load factor over months: peaks are outrunning usage, act before the ratchet sets
4. Flat, high profile with low load factor: check for oversized or constantly cycling equipment

Why clean interval data is the hard part

None of this works on messy data. Interval files arrive in different formats, with gaps, duplicate timestamps, daylight-saving shifts and unit inconsistencies, and a single misaligned day can hide a real peak or invent a false one. Before you can read a profile you have to trust it. MartinAI ingests interval and meter data across formats, resolves the timestamp and unit problems, and produces clean profiles you can act on, so baseload, peak and load factor mean what they say.