Heat does not knock buildings down, so it rarely gets a continuity plan of its own. It turns up instead as three unrelated-looking problems on three different desks. A warehouse worker gets dizzy on the third afternoon of a heat wave. The server room's inlet temperature creeps up because a rooftop condenser cannot shed heat into 108-degree air. The utility asks large customers to cut usage between 4 and 9 p.m., and the regional grid operator starts issuing emergency alerts. Safety, IT and facilities each handle their piece. Nobody owns the hazard.
That is a mistake, because the three problems share a cause and a forecast. Heat is one of the few hazards the National Weather Service can see coming several days out with reasonable confidence, and its worst hours are predictable. It has also been the leading weather-related killer in the United States over recent decades, ahead of hurricanes, floods and tornadoes. This playbook treats it as a single hazard with one set of triggers.
The workforce
What regulators expect
There is still no final federal OSHA standard dedicated to heat, but that does not mean there is no obligation. OSHA cites heat hazards under the General Duty Clause, and since 2022 it has run a national emphasis program on indoor and outdoor heat hazards that steers inspections toward hot days and high-risk industries. The agency has kept that program going. In 2024 OSHA published a proposed heat injury and illness prevention standard, built around two heat index triggers (roughly 80°F and 90°F in the proposal) with escalating duties for water, rest, shade, acclimatization and monitoring. Its path to a final rule has been slow, so check the current status on OSHA's heat exposure page rather than relying on any article, this one included. Several states, among them California, Washington, Oregon, Colorado and Minnesota, have their own heat rules, and they differ.
The practical answer does not depend on how the rulemaking ends. A written heat illness prevention program with clear triggers is what a reasonable employer has, and it is what you would want to show an inspector, a plaintiff's lawyer or a grieving family.
The elements that prevent illness
- Acclimatization. A disproportionate share of heat deaths happen in a worker's first few days in hot conditions. NIOSH guidance for new workers is to start at no more than 20 percent of a full heat exposure on day one and add no more than 20 percent a day. Workers returning from a week or more away need a shorter ramp too. Build this into scheduling for new hires, temps and seasonal staff, who are the least likely to speak up.
- Water, rest, shade. Cool water close to the work, scheduled breaks that lengthen as the heat index rises, and shaded or air-conditioned rest areas. "Breaks as needed" is not a schedule.
- Measure at the work, not at the airport. Published heat index values assume shade, and the NWS notes that full sun can add up to 15°F. A loading dock, a kitchen line, a warehouse mezzanine or the back of a box truck can be far hotter than the official reading. Many safety teams use wet bulb globe temperature (WBGT) on site because it captures sun, wind and humidity together.
- Buddy checks and a response plan. Heat stroke is a medical emergency. Supervisors should know to start cooling the person immediately, with cold water, ice and shade, while emergency services are on the way.
- Indoor heat. Warehouses, plants and commercial kitchens without air conditioning are where indoor heat illness concentrates, and they get worse when the grid asks you to cut load.
The grid
Grid stress during heat follows a recognizable sequence. First comes a conservation appeal from the grid operator or utility, asking customers to reduce use over the evening peak. If reserves keep shrinking, the operator moves through Energy Emergency Alert levels defined by the North American Electric Reliability Corporation: EEA1 when all available generation is in use, EEA2 when load management procedures are in effect, and EEA3 when firm load interruption is imminent or under way. EEA3 is when rotating outages happen.
California's grid operator ordered rotating outages during an August 2020 heat wave, the state's first since the 2001 energy crisis. Texas's February 2021 crisis during Winter Storm Uri was a cold event, but it remains the best recent lesson in what grid stress does to continuity plans, and the lessons carry straight over to summer:
- Rolling outages may not roll. The volume of load that had to be shed, plus circuits that could not be cut because they fed critical facilities, meant many Texans lost power for days instead of the short rotations they had been told to expect.
- Dependencies fail together. Some natural gas facilities lost power because they had not been flagged as critical loads, which cut fuel to power plants and deepened the shortfall. Water systems lost pressure and issued boil-water notices. Your suppliers and your employees' homes sit on the same grid you do.
- The margin was thin. ERCOT later said the system came within minutes of an uncontrolled collapse that could have left much of the state dark for a long time. A plan that assumes "a few hours" is a choice, not a forecast.
The 2003 Northeast blackout, which began on a hot August afternoon with heavily loaded lines sagging into trees in Ohio, is the reminder that a grid failure can cross state lines within minutes.
Two things to settle before summer. Ask your utility whether any of your sites qualify for critical-customer or priority-restoration status, and what that status actually delivers. And if you have a demand response contract that pays you to cut load during emergencies, make sure the people who will execute it know which loads they are cutting. A demand response event at 4 p.m. on the hottest day of the year is a bad time to learn that the "nonessential" chiller also cools the server room.
The machines
Data rooms and closets fail in heat for reasons that have little to do with IT. Cooling units are sized for a design day; a week above that design temperature, with no overnight relief, takes them past it. Rooftop condensers sit in full sun on dark roofs. One failed unit in an N+1 arrangement leaves no margin at all. During the July 2022 heat wave, when the UK recorded temperatures above 40°C for the first time, two major cloud providers reported cooling-related failures at London data centers. If operators with engineering staff on site had trouble, a small server room on a single split system has less headroom than you would like.
- Know your thresholds. ASHRAE's recommended inlet range for most IT equipment tops out at 27°C (80.6°F). Set alarms below the point where equipment throttles or shuts down, and send them to a person, not just a log.
- Check the generator's derating curve. Standby generators lose output at high ambient temperatures. A unit that carries your load on a mild day may not manage it at 105°F.
- Watch the batteries. UPS batteries age faster when kept warm, so a hot summer shortens the runtime you will need in the next outage.
- Reserve portable cooling early. Spot coolers and rental chillers disappear regionally during heat waves, much as generators do before hurricanes.
- Write the shutdown order. Decide in advance what you switch off first to cut heat load if cooling degrades. Test and development systems are the usual first candidates.
The home office
Remote work moves the outage problem into hundreds of homes. When a neighborhood loses power, home internet usually goes with it, mobile networks get congested, and cell sites run on batteries that last hours rather than days. An employee with no air conditioning during a heat wave is a welfare concern as well as a productivity one, and duty of care obligations do not stop at the office door.
Give remote staff a simple rule: if your home loses power or cooling during a heat alert, check in, then go to a designated cooled location (an open office, a coworking space you have arranged, or a public cooling center) rather than trying to work through it. Know which roles can pause for a few hours and which need a backup person in another region.
Freight and suppliers
Heat slows freight. Railroads impose speed restrictions when rail temperatures climb because track can buckle, and pavement buckles too. Hot, thin air reduces aircraft performance, which brings weight restrictions and occasional cancellations at hot airports. Refrigerated loads, pharmaceuticals and food have less margin when reefer units are working harder and delivery windows slip. When heat comes with drought, low river levels can restrict barge traffic, as happened on the Mississippi in 2022 and 2023. Ask critical suppliers how heat and grid alerts affect them, especially those on your regional grid.
Your county or city emergency management office will be running its own heat plan: cooling centers, outreach to vulnerable residents, coordination with the utility. Businesses with working relationships with local emergency management tend to hear about grid conditions and cooling center openings earlier, and can sometimes offer space or transport in return.
Triggers and thresholds
The NWS issues heat advisories, extreme heat watches and extreme heat warnings, with criteria that vary by local forecast office (product definitions). It also publishes HeatRisk, a seven-day, color-coded scale from 0 (green) to 4 (magenta) that reflects how unusual the heat is for that place and time of year, which makes it useful for planning several days ahead. Combine those with grid alerts and your own measurements:
| Trigger | Source | People | Facilities and IT | Business and comms |
|---|---|---|---|---|
| HeatRisk orange (2) or higher, 3 to 7 days out | NWS | Check acclimatization of new and returning staff | Check cooling maintenance, generator fuel, portable cooling contracts | Alert the crisis team; review supplier exposure |
| Heat advisory or extreme heat watch | NWS | Start enhanced water, rest and shade schedule; move heavy work to mornings | Tighten temperature alarm response; defer cooling maintenance | Remind remote staff of the cooled-location rule |
| Extreme heat warning, or HeatRisk red (3) or magenta (4) | NWS | Mandatory work-rest cycles; consider halting strenuous afternoon outdoor work | Cooling staff on site or on call; shed test and dev load if needed | Daily heat status message; customer notice if hours change |
| On-site heat index or WBGT at your high threshold | Your measurements | High-heat procedures; buddy checks every hour | None | Log the readings |
| Conservation appeal | Grid operator or utility | None | Pre-cool spaces before the peak; cut non-essential load | Tell staff why lights and temperatures are changing |
| EEA1 or EEA2 | Grid operator | Plan for outages at sites and homes | Confirm generator readiness, fuel and transfer switch | Brief leaders; review demand response commitments |
| EEA3 or rotating outage notice | Grid operator or utility | Check on remote staff in affected areas | Prepare orderly shutdown where no generator | Activate the continuity plan for affected sites |
| Data room inlet near 27°C (80.6°F), or a cooling unit lost | Your monitoring | None | Shed heat load; deploy portable cooling | Notify application owners |
Decision rules to adopt before next summer
- One owner. Name a single person who owns heat across safety, facilities and IT, with authority to activate the plan on an NWS watch.
- Prepare on the forecast, act on the warning. Use HeatRisk and watches to get ready; do not wait for a warning to order spot coolers or change shifts.
- New people, slower ramp. No new or returning worker does a full shift in heat on day one.
- Assume the outage outlasts the estimate. Plan generator fuel and staffing for days, not hours.
- Test the generator under load in July, not on a cool spring morning.
- Know the first thing you will switch off if cooling degrades, and write it down where the night shift can find it.
- Do not count on remote work as the fallback for an outage that also takes out employees' homes.
