The Bill That Doubled From Emergency Heat

Why this matters

A heat pump customer with a doubled winter bill is the call most likely to end in a condemned compressor that was never broken. The outdoor unit is sitting silent, the bill is enormous, and every piece of surface evidence points at a failed machine. In this case the machine was fine. A thermostat setting had been changed for a sensible-sounding reason a month earlier and nobody changed it back.

This case runs the suspects in the order the evidence killed them, cheapest first. The one that survived cost nothing to check and was nearly the last thing anyone looked at.

The call and the record

The customer called in early February. Her January bill had come in at roughly twice the previous one, the house had been warm all month, and she wanted the heat pump looked at. The office booked a diagnostic. The tech, before driving out, asked her to photograph last January's bill and this one, and to open the thermostat's app and send the monthly runtime summary. That request is the reason the call took one visit instead of three.

BILL A - last January        31 days   actual read
  Usage 1,450 kWh             46.77 kWh/day
BILL B - this January        31 days   actual read
  Usage 2,920 kWh             94.19 kWh/day
  Energy charge per kWh: up 3% from Bill A
  Same meter number on both

THERMOSTAT RUNTIME, Bill B period (4 Jan - 3 Feb)
  Mode                   Em Heat, all 31 days
  Mode last changed      2 Jan, 6:42 AM
  Compressor heat        0.0 h on every day
  Em heat, daily avg     5.3 h
  Coldest day, low -4 F  9.6 h em heat
  Mildest day, low 34 F  1.8 h em heat

The answer is on that page. The tech did not read it as the answer at first, and the reason is worth following, because it is the reason this call goes wrong so often.

First suspect: the bill itself

Hypothesis: the bill is wrong, or the comparison is unfair.

Both reads are actual, so there is no estimated low month being caught up (see related: Estimated Versus Actual Meter Reads and the Catch-Up Bill). Both periods run 31 days, so use per day doubled exactly as the totals did: 2,920 / 1,450 = 2.01 on the totals, and 94.19 / 46.77 = 2.01 per day. The meter number did not change, so there was no swap.

The rate line moved 3 percent, which accounts for a 3 percent rise on a bill that roughly doubled.

Killed. The house really did use about twice as many kilowatt-hours.

Second suspect: the weather

Hypothesis: this January was much colder.

A heating degree day (HDD) measures how far and how long the outdoor temperature sat below 65 F. The nearest public weather station gave 1,000 HDD for Bill A's period and 1,020 for Bill B's, or 32.26 and 32.90 per day. This January was 2 percent colder.

To size what 2 percent does, take out the baseload, the use per day in a month with no heating or cooling. This house's October and May bills averaged 16.0 kWh/day. Last January's heating was then 46.77 minus 16.0 = 30.77 kWh/day, or 30.77 / 32.26 = 0.954 kWh per degree day. At this January's 32.90 degree days per day, the same house would have used 0.954 x 32.90 = 31.39 kWh/day of heating, plus 16.0 baseload, which is 47.39 kWh/day, or about 1,469 kWh for 31 days. The bill was 2,920.

Killed. Weather explains about 19 kWh of a 1,470 kWh rise. The method is How to Weather-Normalize Two Winters of Bills; here one line of it was enough.

Third suspect: a failed heat pump

Hypothesis: the outdoor unit has failed and the backup strips are carrying the house.

This is where the tech's instinct went, and on arrival the evidence seemed to agree: thermostat calling for heat, warm air at the registers, outdoor unit completely still. No fan, no compressor hum. A heat pump that has lost its compressor will do exactly this, warm the house on strip heat, and run up a bill.

What stopped the tech from opening the outdoor unit was the runtime summary, reread on the porch: compressor heat, 0.0 hours, every day. A failed compressor on a thermostat in Heat mode still gets CALLED. Most thermostats log the call, not whether the equipment answered it, so a dead compressor shows up as compressor hours with no heat behind them. Zero hours of call means the thermostat never asked. And the mode line said why: Em Heat, set on 2 January at 6:42 AM.

Emergency heat (often shown as "Em Heat", "Emer" or "Aux Only") is a mode that locks out the compressor and heats on the electric resistance strips alone. It exists for when the outdoor unit genuinely cannot run. In that mode, a perfectly healthy outdoor unit sits silent, which is what the tech walked up to.

The display said EM HEAT in small letters under the setpoint. It had said so for a month.

Not killed yet, but demoted. A silent outdoor unit in Em Heat proves nothing about whether it works. That gets tested before anyone leaves, below.

What happened on 2 January

The customer remembered it once asked. Freezing drizzle overnight. Early in the morning the outdoor unit's fan stopped, there was a loud whoosh, and a white cloud rose off the unit. She thought it was on fire. A neighbor told her that switching the thermostat to emergency heat "turns the outside part off", which is true, and she did. The cloud stopped. The house stayed warm. Nothing else seemed wrong, so she forgot about it.

What she saw was a defrost cycle: on a cold, wet day, ice builds on the outdoor coil, and the heat pump briefly reverses to warm the coil and shed it. The outdoor fan stops, the reversing valve shifts with an audible whoosh, and melting ice turns to vapor that looks like smoke. It typically lasts a few minutes and is normal.

The instinct to act on smoke was right, and nobody should talk a customer out of it. The line the shop now draws for customers: vapor from a defrosting unit is white, has no smell, and stops within minutes. Dark smoke, a burning smell, or anything coming from the indoor unit or the panel means turn the system off at the thermostat, leave the house if the smell is strong, and call the shop or 911. When in doubt, they call. What she was missing was only the second half: call the shop the same day and ask about it.

Does the strip runtime account for the bill?

A mode setting is a plausible story. Before telling a customer it explains a doubled bill, size it. The procedure for this belongs to How to Confirm Backup Heat Runtime From the Thermostat and a Clamp; here are this house's numbers.

What the strips actually draw. The heater kit nameplate read 15 kW at 240 V, on two circuits. With the thermostat calling, the tech (qualified for energized measurement, electrical compartment cover only, the blower door interlock left alone) clamped one conductor of each circuit: 31.1 A and 30.9 A, together 62.0 A, at a measured 238 V. Power is 238 x 62.0 = 14,756 W, about 14.76 kW. Cross-check: the elements' resistance at rating is 240 squared / 15,000 = 3.84 ohms, which at 238 V should draw 238 / 3.84 = 62.0 A. Every element was energizing.

What the runtime adds up to. The log's 5.3 hours a day of em heat, times 14.76 kW, is 78.2 kWh/day. Over 31 days that is 5.3 x 31 = 164.3 hours, times 14.76 kW, about 2,425 kWh.

What the bill says heating used. Bill B's 94.19 kWh/day minus the 16.0 baseload leaves 78.19 kWh/day, or about 2,424 kWh. The strips account for essentially all of it. That agreement is closer than this method deserves: the baseload estimate and the blower fan's energy both carry more error than a few kWh, so the honest reading is "the strips explain the heating on this bill", not a precise match.

Does the runtime behave like heat? On the coldest day, a low of -4 F, the strips ran 9.6 hours. On the mildest, a low of 34 F, 1.8 hours. Runtime tracking cold is what heating does, which says the strips were doing the house's heating rather than running from a stuck relay.

What the heat pump would have spent. Last January the house used 0.954 kWh of electricity per degree day for heating. This January, all on strips, it used 78.19 / 32.90 = 2.38 kWh per degree day. Strips turn one unit of electricity into one unit of heat, so this house's own bills say the heat pump was delivering about 2.38 / 0.954 = 2.5 units of heat per unit of electricity last January. The same heat by heat pump would have taken about 2,425 / 2.5 = 970 kWh. The difference, about 1,455 kWh, matches the 1,451 kWh left over after weather (2,920 minus 1,469). That ratio is an average over a month that includes defrost and some backup on the coldest mornings, not a rating for the unit.

Proving the heat pump before switching it back

The mode was the cause of the bill. That does not mean the outdoor unit was fine; it means nobody had asked it to run for a month. Two checks before handing the system back:

The unit itself. A month of freezing drizzle and snow had packed one side of the outdoor coil and frozen snow around the fan guard. The tech switched off the outdoor disconnect, confirmed the unit was dead, and cleared snow from the top and sides by hand, leaving ice on the coil alone: chipping it bends fins and can puncture tubing, and the first defrost cycle will clear it. The disconnect had stayed on all month, so the compressor's crankcase heater had stayed powered. Had the customer also pulled the disconnect, the compressor would not have been started until the crankcase heater had warmed it for the time the manufacturer's instructions call for, because starting a cold compressor with refrigerant pooled in the oil can damage it.

The first run. Disconnect on, thermostat to Heat, setpoint raised 2 F. The tech stood beside the unit, clear of the fan discharge, and watched: outdoor fan running, compressor running, supply air at the registers warm and steady after a few minutes, the thermostat display reading HEAT with no EM text. The runtime summary the next morning showed compressor hours and a little aux on the coldest stretch before dawn. Had the compressor not started, the job would have stopped there and become a repair diagnosis, and the customer would have been told that emergency heat is the right mode until it is fixed.

What changed afterward

For the customer. The tech showed her where the mode appears on the display, what EM HEAT looks like next to HEAT, and what a defrost cycle sounds like, and wrote the shop's number on the thermostat's wall plate with "call before switching to Em Heat". She was not told what the next bill would be. She was told what the next runtime summary should show: compressor hours on most days, little or no em heat, some aux on the coldest mornings.

For the office. The high-bill intake script on heat pump accounts gained one question, asked before anything is scheduled: "Can you look at the thermostat and tell me exactly what it says next to the temperature: Heat, Auto, Em Heat, or something else?" On this call it would have taken under a minute and saved the truck roll. The tech still would have come out, because a mode left in Em Heat after a scare means the unit needs to be seen running, but the visit would have been booked as a check, not a diagnosis.

For the techs. Read the runtime summary's compressor line before touching a silent outdoor unit. Zero compressor hours on a heating call is a setting until proved otherwise.

References

• See related: How to Confirm Backup Heat Runtime From the Thermostat and a Clamp
• See related: How to Weather-Normalize Two Winters of Bills
• See related: How to Diagnose a High-Bill Call, in Order
• See related: COP, AFUE and Delivered Heat: Putting Every Fuel on One Scale
• Heat pump and thermostat manufacturer installation and operating instructions for mode, defrost and crankcase heater behavior