How to Weather-Normalize Two Winters of Bills

Why this matters

A customer holding two winters of bills is asking one question: did my house change, or did the weather? Raw bills cannot answer it, because a colder winter raises use in a healthy house and a mild one hides a failing one. Weather normalization divides out the weather using heating degree days, and leaves a single figure per winter, energy used per degree day, that moves only when something about the house, the equipment or the people in it changed. It takes about twenty minutes with the bills and the station data, and it either clears the shop's install or tells the shop where to look.

What a degree day is, where the numbers come from and why the base temperature matters are owned by the reference card. See related: Degree Days and Season Length. Reading the bill itself, its meter reads and its charges, is owned by the bill-reading cards. This procedure assumes both and does the comparison.

The procedure

1. Collect the bills for both winters and one summer. You need every bill covering the two heating seasons to be compared, plus two or three bills from summer months when the heating system did not run. For each bill, note the read dates, the number of days, the energy used, and whether the read was actual or estimated. Reason: the summer bills give you the house's baseload, and the read dates decide which degree days belong to which bill. An estimated read shifts use between bills, so a winter that contains one should be compared across its whole span, not bill by bill. Skip the summer bills and you cannot separate heating from everything else.

2. Put every bill in one energy unit. Gas bills may state use in hundreds of cubic feet (ccf) and convert to therms with a heat-content factor printed on the bill, which changes slightly from month to month. Convert each bill with its own factor, or use the therms line if the bill prints one. Electric bills are already in kilowatt-hours. Reason: comparing ccf in one winter with therms in the other builds a few percent of error into a comparison whose whole point is a few percent.

3. Sum the heating degree days for each bill's exact dates. From one weather station, chosen for being close to the house in elevation and setting, sum the daily heating degree days from the first day of each billing period to its last, then total them for each winter. Use the same station and the same base, 65 F unless you have a reason to use another, for both winters. Reason: monthly degree-day totals for calendar months do not line up with billing periods that run from the 7th to the 6th. Using calendar months misassigns cold snaps near the read dates.

4. Work out the baseload per day from the summer bills. Add the summer bills' use and divide by their total days. That is the energy the house uses every day for things that do not respond to outdoor temperature: water heating, cooking, a dryer, a gas fireplace pilot. Reason: baseload does not scale with degree days, so it has to come off before you divide. Leave it in and a short mild winter looks worse per degree day than a long cold one, because the same daily baseload is spread over fewer degree days.

Know which way this step errs. Water heating uses more energy in winter than in summer, because the incoming water is colder. A summer baseload therefore underestimates the winter baseload, and the difference gets counted as space heating. That inflates use per degree day in both winters by a similar amount, so the comparison between winters holds even though the absolute figure runs a little high. If the household changed something that only affects one winter's baseload, such as adding a hot tub heater, the step no longer cancels and you have to account for it.

5. Subtract the baseload to get heating use for each winter. Multiply the baseload per day by each winter's days and subtract it from that winter's total. Reason: what remains is the energy that the weather was asking for, which is what the degree days measure.

6. Divide heating use by degree days. Each winter now has one figure: therms or kilowatt-hours of heating per degree day. Reason: this is the house's sensitivity to cold. It should stay roughly constant if nothing changed, whatever the weather did.

7. Make the call with a stated tolerance. Base temperature, billing-date alignment, meter reading and the summer-baseload bias each carry a few percent of error. As a shop default, treat a change in use per degree day within plus or minus 5 percent as no change in the house, and anything beyond it as something changed that deserves a look. Tune the band once you have run a few of these on houses where you know the answer. Reason: without a tolerance, every comparison finds a difference and every difference becomes an accusation.

8. Show the customer the laid-out comparison, not just the conclusion. Write it out (the artifact below is the shape), and walk through the three change lines: what the bill did, what the weather did, what the house did. Reason: a customer who sees that the weather rose more than the bill believes the explanation; one who is told "it was a cold winter" does not.

A worked comparison

A customer had a new gas furnace installed in the autumn before Winter 2 and believes it uses more gas than the old one. The bills are in therms. Three summer bills covering 92 days total 37 therms, which is 0.402 therms a day; the comparison uses 0.40.

WEATHER-NORMALIZED COMPARISON (gas, therms)
Station: county airport           Base: 65 F
                           WINTER 1    WINTER 2
Billing span                Dec-Feb     Dec-Feb
Days                           90          91
Therms used                   312         356
HDD, same dates             2,520       2,980
Baseload 0.40 x days         36.0        36.4
Heating therms              276.0       319.6
Therms per HDD             0.1095      0.1072
Raw use change                  +14.1 %
Weather change (HDD)            +18.3 %
Use per HDD change               -2.1 %
Call: inside 5 percent, no change in house

Reading it line by line. Winter 2 used 44 therms more than Winter 1 (356 minus 312), a 14.1 percent rise (356 divided by 312 is 1.141). The weather station counted 460 more degree days (2,980 minus 2,520), an 18.3 percent rise (2,980 divided by 2,520 is 1.183). Gas use rose less than the weather did. After the baseload comes off, Winter 1 used 0.1095 therms per degree day (276.0 divided by 2,520) and Winter 2 used 0.1072 (319.6 divided by 2,980). The new furnace winter used 2.1 percent less gas per degree day (0.1072 divided by 0.1095 is 0.979).

The call. Minus 2.1 percent sits inside the 5 percent band, so the honest statement is that the house's use per degree day did not measurably change. It is tempting to tell the customer the new furnace saves 2 percent. Do not: the difference is inside the error of the method, and a claim the next winter's bills can contradict costs more trust than it buys. Say what the numbers support: the bill rose because the winter was colder, and the house responded to the cold the same way it did before.

What would have changed the call. Suppose Winter 2 had used 400 therms instead of 356. Heating use becomes 363.6 therms (400 minus 36.4) and use per degree day 0.1220 (363.6 divided by 2,980), up 11.4 percent on Winter 1 (0.1220 divided by 0.1095 is 1.114). That is outside the band. Before suspecting the furnace, ask what else changed that winter: more people at home in the day, a thermostat setpoint raised, a room added or a zone closed off and reopened, a new gas appliance, a window left cracked in a basement. If none of those changed, the furnace and its distribution become the suspects: short cycling from an oversized or misconfigured unit, duct leakage to an unconditioned space, a zone valve stuck open, or a high-efficiency unit that is not condensing because return water or air runs too warm. That is a diagnostic visit, quoted as one.

When the method does not hold

Heat pumps. A heat pump's energy per degree day rises in colder weather, especially once electric resistance backup starts carrying part of the load. Two winters of different severity will show different use per degree day without any fault. Compare only winters of similar severity, or split the comparison into mild and cold billing periods and compare like with like. A mild winter against a severe one will make a healthy heat pump look worse.

Changes in what the meter serves. A gas meter that started serving a pool heater, or an electric meter that picked up an electric vehicle charger, has a new baseload, and the summer bills from before the change no longer describe it. Use summer bills from after the change, or set the comparison aside.

Partial winters. A house vacant for a month, or a furnace replaced mid-January, breaks the winter into two different houses. Compare matching spans before and after rather than whole seasons.

Too little data. One winter of bills and no summer bills is not enough to normalize. Say so rather than estimating a baseload you did not observe.

How to know you did it right

Two checks. First, the baseload should be plausible for the household: if it implies more gas a day in summer than the water heater and range could burn, a summer bill carried a heating or pool load and the baseload is wrong. Second, recompute each change line from the lines above it before showing the customer, because the change lines are the part they will repeat to a neighbor. Then write the comparison into the customer's record, so the next winter's bills can be added to the same table rather than starting over.

References

• NOAA National Centers for Environmental Information and the local National Weather Service office's monthly climate reports, for daily heating degree days by station
• The utility's bill format and heat-content factor, as printed on the customer's bills
• See related: Degree Days and Season Length; What a Utility Bill Can Tell You That an Inspection Can't
• See related: Reading Gas Meter + Usage Reference; Reading Customer's Electric Bill Reference
• See related: Separating a Usage Habit From an Equipment Fault