Heat load calculations, done properly

Most oversized chillers and short-cycling heat pumps started life the same way: somebody skipped the load calculation and reached for a number that "usually works".

This guide covers what a proper heat load calculation includes, where the common shortcuts fail, and what to check when someone hands you a calc and asks you to build from it.

What the calculation is actually for

A load calculation answers one question. On the design day, how much heat does this space gain or lose, and therefore what duty does the plant need? Everything downstream inherits its accuracy: equipment selection, pipe and duct sizing, electrical supplies, plant space. Get the loads wrong and no amount of good engineering afterwards fixes it. You just build a well-executed mistake.

The components of a cooling load

A room-by-room cooling load has five parts, and all five need doing.

Fabric gains. Heat conducted through walls, roof, floor, and glazing. U-value times area times temperature difference, applied surface by surface. Modern, well-insulated fabric has shrunk this component dramatically, which is exactly why old rules of thumb now oversize.

Solar gains. Usually the dominant gain in a glazed space, and the one rules of thumb handle worst, because it depends on orientation, shading, and glass specification. A south-west-facing meeting room and an identical north-facing one are different design problems.

People. Sensible and latent heat per occupant varies with activity. Design occupancy matters more than actual occupancy: a meeting room is designed full, not at its Tuesday-afternoon average.

Equipment and lighting. Nameplate ratings overstate real heat output for most office equipment, but dense IT loads are real and continuous. Comms rooms hiding inside "office" floorplates catch people out constantly.

Ventilation and infiltration. Outside air brought in deliberately, plus air leaking in around the fabric. On humid design days the latent portion of the fresh-air load is substantial, and it's the component most often quietly dropped from quick calcs.

Heating loads run the same logic in reverse: fabric losses plus ventilation and infiltration, minus nothing. You don't count internal gains you can't rely on at 6 a.m. on the coldest morning of the year.

Design conditions

Before a single gain is calculated, the design conditions set the frame, and they should be written on the calculation rather than implied. That means outdoor design temperatures for the actual location, from recognised climate data rather than the hottest day anyone remembers. It means indoor setpoints agreed with the client, because the difference between designing to 22 °C and 24 °C is real money in plant. And it means stated assumptions for anything unknown: fabric build-ups, occupancy, equipment. An assumption written down is a design decision. An assumption in someone's head is a future dispute.

Diversity

Sum every room's peak and you'll buy a system for a building that never exists. The south rooms peak in the afternoon, the east rooms peaked at 10 a.m., and the meeting rooms are never all full at once. Room loads size the terminal units; a diversified building load sizes the central plant. Applying diversity twice, or not at all, is one of the most common structural mistakes in calcs I review.

"Add a bit to be safe"

Oversizing feels prudent and isn't. An oversized system costs more to buy and install, then spends its life cycling, which means worse comfort, worse humidity control, worse efficiency, and more wear. Margins have their place, but they should be chosen, stated, and applied once. What actually happens on many jobs is that everyone in the chain quietly adds their own, and the plant ends up half as big again as the building needs.

Checking someone else's calculation

When a calc lands on my desk for review, the first pass is always the same five questions.

What design conditions were used, and are they stated? If you can't find them, that's the finding.
Does the W/m² pass the sniff test? Experience gives you sector ranges, and a number well outside them needs a reason.
Where's the fresh-air load? Look for the latent component specifically.
Has diversity been applied once, deliberately?
Do the room loads reconcile with the plant selection, or did a margin creep in between the two?

Twenty minutes of this catches most of the errors that matter, because the errors that matter are rarely arithmetic. They're assumptions nobody wrote down.

Where software fits

Load software is genuinely useful. It does the repetitive surface-by-surface arithmetic without transcription errors and makes "what if" comparisons cheap. What it can't do is notice that the occupancy is wrong, the glass specification is optimistic, or the comms room got labelled as a store. The judgement about conditions, assumptions, diversity, and margin is still a human job, and it's the part that decides whether the building works.

I produce room-by-room load calculations with the design basis stated and defensible, either as part of a full HVAC design package or as an independent check on an existing design. If you've got a calc you're not sure about, send it over.