Skip to content
allanninal.dev/build
Collaborate

Part 4 of 7 · Utility meter reader series ~5 min read

Why gas needs the weather

Heating fuel consumption is dominated by how cold it was, to the point where a year-on-year comparison of raw gas use is mostly a comparison of two winters.

Key takeaways

  • Degree days measure how much heating the weather demanded.
  • Gas per degree day is the comparable number; raw gas is not.
  • A change in the slope means the building or its controls changed.
  • The intercept is the non-heating gas: hot water and catering.
  • Free public weather data is good enough; a nearby station is fine.

What a degree day is

A simple idea: for each day, take how far the average outdoor temperature fell below a base — usually around fifteen and a half degrees, the point at which a typical building needs heating — and add it up. A day averaging ten degrees contributes five and a half degree days; a day averaging eighteen contributes none.

Sum those over a month and you have a single number describing how much heating the weather demanded. Divide the month’s gas by it and you have a figure that can be compared across months and years, because the weather has been divided out.

Two winters

Two Januaries compared raw and normalised for degree daysA bar chart with four bars. Two series: gas used in kilowatt-hours in purple, and kilowatt-hours per degree day in red. Last January raw: forty-eight thousand kilowatt-hours. This January raw: forty-two thousand two hundred. Last January per degree day: one hundred and thirty-two. This January per degree day: one hundred and forty-one. A note says raw use fell twelve per cent while adjusted for weather it rose seven per cent, and the mild winter hid a problem.020000400006000080000~48000Last Jan, raw~42200This Jan, raw~132Last Jan, per DD~141This Jan, per DDGas used, kWhkWh per degree dayRaw use fell 12%. Adjusted for weather it rose 7%. The mild winter hid a problem.
Fig 1. The same two months measured two ways. The raw comparison says the building improved; the normalised one says it got worse and the weather was doing the work.

This is the most common way an energy problem goes unnoticed for a year. A mild winter produces a lower bill, everybody is pleased, and a heating system that has developed a fault or a control that has been overridden is invisible until a cold winter arrives and the bill is startling.

The normalised figure catches it in the mild year, which is the only useful time to catch it.

The slope and the intercept

How the slope and intercept of a gas against degree days line are readA vertical chain of five steps entered by a box labelled Monthly gas and degree days over a year or more. Step one plots one against the other, noting it is close to a line. Step two identifies the slope, kilowatt-hours per degree day, with a side box saying this is heat loss and controls. Step three identifies the intercept, gas at zero degree days, with a side box saying this is hot water and catering. Step four compares both numbers to last year. Step five asks which one moved, since they mean different things. A note says a higher intercept is a hot water problem and a steeper slope is a heating one.AWS ACCOUNTMonthly gas and degree daysa year or morePlot one against the otherit is close to a lineThe slopekWh per degree dayThis isheat loss and controlsThe interceptgas at zero degree daysThis ishot water, cateringCompare to last yearboth numbersWhich one moved?they mean different thingsA higher intercept is a hot water problem. A steeper slope is a heating one.
Fig 2. The two numbers that come out of a degree-day analysis and what each one means. Separating them tells you which system to look at.
  • Compute
  • App integration
  • Management
  • Analytics

Reading the two

A steeper slope means the building is losing more heat per degree of cold, or the heating is running harder than it needs to. Insulation does not get worse quickly, so a steepening slope usually means controls: a thermostat moved, a compensation curve changed, a valve stuck open.

A higher intercept means more gas is being used regardless of the weather, which is hot water, catering, or heating that runs in summer. That is a different investigation and often an easier one.

Where to get the weather

Public weather data from a station within a reasonable distance is more than good enough. The method is not sensitive to a degree of difference, and the alternative — an on-site weather station — adds precision to the smallest source of error in the analysis.

Limits worth stating

Where degree days stop working

  • Buildings with high internal gains — a busy kitchen, a server room — heat themselves, and the base temperature is lower than the standard one.
  • Cooling needs cooling degree days, which are a separate calculation and usually matter for electricity rather than gas.
  • Occupancy changes break the comparison entirely. A building at half capacity is a different building.
  • Process gas — an oven, a furnace — has nothing to do with the weather and should be sub-metered out if it is significant.
  • A year is the minimum for a useful slope. Three months of data produces a line and no confidence.
  • Say the base temperature used, on every report. It is an assumption and people should be able to argue with it.

The occupancy point is worth watching for, because it is the one that produces the wrong conclusion most convincingly. A normalised figure that improved after half the staff started working from home has measured the occupancy change, and reporting it as an efficiency improvement is a mistake that will be repeated at the next comparison.

Next: turning a finding into a saving that is real.

All posts