Reading a half-hourly consumption profile before buying anything

The data exists and almost nobody looks at it

A meter capable of interval metering records consumption in periods of half an hour or less, and in most jurisdictions the household has a right to that data. It is generally available through the supplier’s portal, through an in-home display, or through a local interface on the meter itself. It is not always easy to extract, and it is worth the trouble.

What it contains is a picture of the house that no annual total can give. An annual figure tells you how much energy was consumed. A profile tells you when, in what size chunks, and with what regularity, and every energy decision worth making depends on those three.

A month of data is enough to be useful. A year is better, because heating and lighting demand are seasonal and a summer profile will mislead you about winter.

Finding the baseload

The first thing to look for is the floor: the lowest consumption the house ever reaches, typically in the small hours when everyone is asleep. That is the baseload, and it is running twenty-four hours a day, every day of the year.

Baseload is worth attention because it is multiplied by a very large number of hours. A steady draw that seems trivial in isolation becomes a substantial annual quantity when it never stops, and unlike a peak it is entirely uncorrelated with anything anyone wanted. It is fridges and freezers, standby electronics, networking equipment, pumps, and in many houses something nobody remembers installing.

Identifying what makes up the baseload is a matter of switching things off and watching the floor move, which is tedious and occasionally revealing. A circulation pump running continuously, an old freezer in a garage, or a piece of equipment left permanently powered are the usual finds.

Finding the shiftable load

The second thing to look for is the large, discrete events: blocks of consumption that appear for an hour or two and then stop. These are the heating elements — the dishwasher, the washing machine, the dryer, the immersion heater, the oven, the vehicle charger.

Each of these is a candidate for shifting to a cheaper period or onto solar generation, and the profile tells you how big each one is and when it currently happens. That is exactly the information needed to decide whether a time-of-use tariff would help, because you can apply the proposed rates to the existing profile and see the answer rather than guessing at it.

The events that cannot move are as informative as the ones that can. If the evening peak is cooking and lighting, it is not going anywhere, and any measure sold on the promise of reducing it should be examined carefully.

Testing a solar or battery proposal against the data

With a profile in hand, a proposal can be checked rather than accepted. For solar, the question is how much of the modelled generation would land in hours when the house is already consuming, since that fraction is worth the import rate and the rest is worth the export rate. A profile that is nearly flat and low during daylight is telling you self-consumption will be poor without deliberate changes.

For a battery, the question is how much energy the house consumes between the end of generation and the start of it again. That quantity is the honest upper bound on useful battery capacity, and it is very often smaller than the capacity being proposed.

This is arithmetic anyone can do in a spreadsheet with data they already own, and it is the most useful hour available to someone considering a system. It will not produce a precise forecast. It will reliably identify a proposal that has been sized for the seller’s convenience rather than the house’s consumption.

The limits of the exercise

A profile describes the house as it is, not as it will be. Buying an electric vehicle, replacing a gas boiler with a heat pump, or a change in who is home during the day will all redraw it substantially, and a system sized against the old profile may be badly matched to the new one.

Interval data is also less precise than it looks. Consumption is averaged across each period, so two short high-power events and one long moderate one can appear identical. For the decisions being made here that resolution is sufficient, but it will not identify individual appliances without a separate monitor.

And weather makes year-on-year comparison harder than expected. A mild winter and a cold one produce noticeably different heating demand from an unchanged house, which is worth remembering before concluding that a measure did or did not work.