Skip to content
allanninal.dev/build
Collaborate

Part 3 of 7 · Utility meter reader series ~5 min read

What the overnight baseload tells you

The single most useful thing to know about a building’s energy use is what it consumes when nobody is in it, because everything drawing power at three in the morning is drawing it for a reason somebody chose or forgot.

Key takeaways

  • Take the minimum sustained draw across several nights, not one night’s average.
  • Work out what it should be from what genuinely must run.
  • The gap is almost always one or two things, not many small ones.
  • Weekends and shutdowns are the best evidence available.
  • A water baseload above zero is a leak, with no exceptions worth arguing about.

Measuring it properly

How overnight baseload is measured from half-hourly dataA vertical chain of five steps entered by a box labelled Half-hourly data over several weeks. Step one takes the quiet hours, two in the morning to four. Step two looks across many nights rather than one, with a side box explaining that one night can be odd. Step three takes the minimum sustained draw rather than the average. Step four compares against shutdown periods such as Christmas and bank holidays. Step five produces the baseload and the question of what it should be. A note says a shutdown week is a free experiment somebody already ran for you.AWS ACCOUNTHalf-hourly dataseveral weeksTake the quiet hours02:00 to 04:00Across many nightsnot oneWhyone night can be oddThe minimum sustainednot the averageCompare to shutdownChristmas, bank holidaysThe baseloadand what it should beA shutdown week is a free experiment somebody already ran for you.
Fig 1. How baseload is measured. The fourth box is the highest-value comparison available and it costs nothing because the data already exists.
  • Compute
  • Management
  • Analytics

The shutdown week

A building closed for a week over Christmas provides the closest thing to a controlled experiment available: everything discretionary is off, and whatever is still drawing power is either necessary or forgotten.

Comparing the shutdown baseload to the ordinary overnight baseload separates the two categories immediately. Anything drawing power on both is permanent and worth understanding; anything drawing power on ordinary nights and not during shutdown is something that gets switched off when somebody remembers.

What it should be

Working out the expected baseload

  • List what genuinely must run: refrigeration, servers, emergency lighting, security, frost protection.
  • Estimate each from its rating and duty cycle. Rough is fine; the gaps that matter are large.
  • Add ten per cent for standby loads across the building.
  • Compare to measured. A measured figure within a third of expected is unremarkable.
  • A measured figure at twice expected is one or two specific things, not a hundred small ones.
  • This is a half-hour exercise with somebody who knows the building, done once.

The fifth line is the practically important one and it surprises people. Large unexplained baseloads are almost never the accumulation of many small oversights; they are a compressor, a chiller, a pump, an air handling unit, or a heater, running when it should not be.

Which makes the tracing exercise tractable. One afternoon, a clamp meter, and switching things off one at a time is usually enough, and the alternative — a full survey — is rarely necessary.

The shapes worth recognising

Overnight baseload before and after fixing an always-on loadA stacked bar chart with four bars in kilowatts. Two series: genuinely needed load in green, and unexplained overnight draw in red. Expected: four kilowatts needed. Always-on excess: four needed plus fourteen unexplained. Timer wrong: four needed plus nine unexplained. After the fix: five kilowatts, all needed. A note says the fourth bar is measured two weeks later, and without it the saving is a claim.010203040~4Expected~18Always-on excess~13Timer wrong~5After the fixGenuinely needed, kWUnexplained overnight draw, kWThe fourth bar is measured, two weeks later. Without it, the saving is a claim.
Fig 2. Baseload before and after a fix. The fourth bar is the verification step from Part 5 and it is what turns an estimate into a number.

Water baseload is a leak

For water the analysis is simpler and the conclusion is stronger. A building with nobody in it should use no water at all, so any sustained overnight flow is a leak, a running overflow, or a urinal flushing on a timer nobody has adjusted.

There is no benign explanation for continuous overnight water flow, which makes this the clearest finding the whole system produces. A small continuous flow — a trickle — adds up to a startling annual volume and is completely invisible on a quarterly bill.

Gas overnight

Gas baseload is more nuanced because frost protection and hot water recirculation are legitimate. The useful comparison is summer nights: a building using meaningful gas overnight in July has heating running that should not be, or a hot water system with no time control.

Next: why the winter comparison needs the weather.

All posts