Low Desert Almanac

Keeping a house cool at 115 degrees

115 °F is not a normal day here: it is a rare afternoon with a long record behind it. What the Phoenix normals and the station record say a house is actually up against.

Short answer

A Phoenix house is not cooled against an average; it is cooled against the days the average hides. The station record has 102 days at or above 115 °F since 1933, and in a normal July there is no hour of the day below 86 °F, so the equipment runs through a night that never cools.

Plenty of places are hotter than Phoenix on a given afternoon. Very few carry the load that Phoenix does, because the number that decides how a house is built and how its cooling system is run is not the afternoon high: it is the low, and here the low in July is 84.5 °F.

This page is about the physical problem: what the house is up against, and which parts of the day decide it. It does not estimate anyone’s bill. If you want to know what the summer does to a statement, that is the subject of understanding a summer electric bill.

What 115 °F means in the record

The title number is worth putting in its place before anything else is said about it. The Sky Harbor station record runs from 1 June 1933 to the present, which is 32,201 daily observations across ninety-three years. Here is how often it has reached each threshold.

ThresholdDays in the recordEarliest it has happenedLatest it has happened
100 °F9,16218 March27 October
105 °F5,27719 March16 October
110 °F1,12210 May6 October
115 °F10217 June28 September

Read the 115 °F row twice. One hundred and two days across ninety-three years, and not one of them in May or October: the days at or above 115 °F fall entirely between mid-June and the end of September, and most of them in July. Those 102 days are spread over 56 separate calendar dates, which is another way of saying that only a few days of the calendar have ever been that hot at all. The station’s own hottest reading is 122 °F, on 26 June, and it has been 120 °F or above on three days in the record.

By contrast, 100 °F is not rare at all. The station has reached it on 9,162 days, the normals give it 111.3 days a year, and the normal high sits at or above 100 °F from 31 May to 18 September. The design case for a Phoenix house is therefore not 115 °F; it is a long run of century days with heat-soaked nights, with 115 °F as the occasional evening that the equipment has to survive rather than the shape of the season.

The night decides the summer

The chart below is the same day in two months, hour by hour, from the hourly normals.

A normal July day against a normal January day, hour by hour

Each line is the normal temperature for that hour at Phoenix Sky Harbor. The dashed line at 80 °F is a reference, not a threshold in any guidance. Hours are local standard time: Arizona does not move its clocks.

Normal temperature by hour of day in July and January at Phoenix Sky Harbor, 1991 to 2020 In a normal July the temperature starts the day near 93 degrees at midnight, falls only to 86 degrees at five and six in the morning, and climbs to its peak of 105 degrees at four and five in the afternoon, still at 100 degrees at eight in the evening and above 90 degrees at midnight. In a normal January the same day runs from 53 degrees at midnight down to 48 degrees around dawn and up to 67 degrees at four in the afternoon. The July line never falls below 80 degrees at any hour, while the January line never reaches it. 110° 100° 90° 80° 70° 60° 50° 40° 80 °F July January 12am 6am noon 6pm 11pm The July line is above the 80 °F reference for all twenty-four hours. The January line never reaches it.

NOAA NCEI U.S. Climate Normals 1991–2020, hourly, Phoenix Sky Harbor (USW00023183), rounded to whole degrees as the site's hour strip prints them. The July minimum of 86 °F falls at 5 and 6 a.m.; the peak of 104.8 °F falls at 4 p.m. in the unrounded series. A normal is a thirty-year average and is the wrong tool for a specific afternoon.

Two features of the July line are the whole problem.

The first is where the floor sits. A normal July day bottoms out at 86 °F between five and six in the morning, which is warmer than the air most people keep their living rooms. There is no hour in which the house can be opened and flushed with cool air, and no hour in which the building envelope gets to shed the heat it collected yesterday. A house that gains heat all day and is given no cool hour to lose it is running a deficit that only the cooling system can pay.

The second is how late the peak arrives. The normal July high is still 100 °F at eight in the evening and above 90 °F at midnight. The evening that people in temperate cities use for cooking, laundry and yard work is, here, one of the hottest parts of the day, which is why the timing question in the peak-hours planner is really a heat question wearing an electricity question’s clothes.

Compare the January line to see how unusual this is. The same city, six months earlier, swings from 48 °F before dawn to 67 °F in the afternoon: a range that makes opening a house at night genuinely useful and makes a January cooling problem mostly about a clear, cold sky.

What the house is actually doing

A house in this climate gains heat in four ways, and they do not respond to the same remedies.

Conduction through the roof and walls is the largest of them in most single-storey homes, which is why attic insulation and radiant barriers are the first things a building scientist looks at. Radiation through glass is next, and it is the one with a daily rhythm, because west-facing glass takes the full force of the afternoon sun from about two o’clock onwards. Infiltration (hot outdoor air drawn in through every gap in the envelope) is the third, and duct leakage is the fourth and most commonly underestimated, because a duct running through a 140 °F attic can lose a large share of the cool air it is carrying before it reaches the room it is meant for.

Sizing a system that can handle all four is a real calculation, done from a room-by-room load estimate rather than from floor area. The trade works from ACCA’s published Manual J and the site’s home cooling task scheduler works from the maintenance side of the same problem rather than attempting to reproduce a load calculation. Anything that claims to size a system from square footage alone is guessing.

Shade before cooling

The cheapest cooling decision on a Phoenix property is not a thermostat setting. It is what is standing between the sun and the west wall.

A tree, a ramada or a shade screen on the west and south sides changes the load before the equipment ever sees it, and it does so at every hour it matters most. Where the shade falls, and which wall takes the afternoon sun, is a geometry question with a precise answer for a given date and orientation: the sun path and shade finder works it out from the sun’s altitude and bearing at Phoenix’s latitude, and the reading the sun across a yard article covers the method.

The same logic applies inside the envelope: the Department of Energy’s guidance puts window coverings, reflective films and exterior shading ahead of buying more cooling capacity, and exterior shading ahead of interior blinds, for the obvious reason that the heat is stopped outside the glass rather than after it has crossed.

The equipment, and what fails in July

The parts of a cooling system that fail in a Phoenix July are usually the parts that were allowed to degrade through a mild spring.

A filter that is loaded raises the pressure drop across the coil, which reduces airflow and lets the coil run colder and wetter than it should. A condenser coil with a season of dust and cottonwood on it cannot reject heat into 110 °F air as designed. A condensate drain line that has never been flushed is the source of most mid-summer water damage calls. And a system that is a little low on refrigerant will run all summer without complaint and fail on the first day it is asked to hold a temperature it could previously manage.

Filters, coils, drain lines and the outdoor unit’s clearance are homeowner tasks and belong on a calendar. Refrigerant handling, electrical work and anything inside the sealed system are not, and this site does not walk anyone through them. The cooling system maintenance calendar sets the household side of that schedule out month by month.

On-peak hours, plainly

Both Valley utilities publish time-of-use plans, and those plans have on-peak windows that fall on weekday afternoons and evenings, when a Phoenix house is also at its hottest.

The hours themselves differ between plans and change from time to time, so this page does not print them. What it will say is that the two facts (the system’s hardest working hours and the grid’s most expensive hours) overlap almost exactly, which is the whole reason the peak-hours habit planner links out to the utilities’ own current plan pages rather than describing a window of its own.

This is not a financial recommendation, and no assumption here about what anything costs. Where a task can move without touching comfort (a dishwasher, a laundry cycle, an electric vehicle charge) it can move; where it cannot, moving it is not worth the discomfort. The sibling article on on-peak hours goes through that trade without putting a number on it.

What varies by household

Four things make one household’s answer different from the next, and all four are worth knowing before comparing notes with a neighbour.

The envelope: when the house was built, how much insulation is in the attic, and whether the ducts run through a conditioned space or through the attic.

The glass: how much of it faces west, and whether it is single, double or shaded.

The equipment: a single-stage unit, a variable-speed unit and an evaporative cooler respond to the same 110 °F afternoon in three different ways, which is why evaporative coolers in dry heat is a separate article rather than a paragraph here.

And the people: how many are home in the afternoon, what temperature the household finds tolerable, and how much of the day the doors are open. A thermostat setting is a household decision before it is an engineering one.

None of that adds up to a figure this site can print for your house, and it would be dishonest to pretend otherwise. What it does add up to is a schedule: a system maintained before the season, a house shaded where the sun lands, and the hardest tasks moved out of the hottest hours.

Sources

Where this came from

Questions

Short answers

Questions about keeping a house cool at 115 degrees

Is 115 °F a normal Phoenix temperature?
No. It is a rare afternoon with a long record. The Sky Harbor record has 102 days at or above 115 °F since 1933, concentrated in June, July and August, and it has never happened in May or October. What is normal is 111.3 days a year at or above 100 °F, which is a different and much larger problem.
Can a house be cooled by opening the windows at night?
Not in July or August. The normal July low is 84.5 °F and the coolest hour of a normal July day is 86 °F, so there is no hour to flush the house with. From roughly late June to mid-September the outdoor air is warmer than the indoor air the house is trying to hold, and ventilation makes it worse.
What thermostat setting should a Phoenix household use?
This site does not print one, and it is worth knowing why. The Department of Energy publishes scheduling and setback guidance rather than a single number, because the right setting depends on the equipment, the ductwork, the house and how people are in it. What can be said plainly is that a system that is set back hard and then asked to recover at 4 p.m. on a 110 °F afternoon is being asked to do its hardest job at its worst hour.