Thermal Zoning & the Model

A thermal zone is a volume of air the simulation solves as a single, well-mixed node - one temperature, one humidity, one heat balance. It is emphatically not the same as an architectural room. Several rooms with similar conditions and a shared HVAC setpoint can be lumped into one zone; conversely, one large hall with a sunny glazed end and a shaded back might deserve to be split into two.

The test for whether two spaces belong in the same zone is simple: would they behave the same way in the heat balance, and are they controlled together? If two offices face the same direction, have the same use and are served by the same thermostat, merging them changes almost nothing and saves computing time. If they differ in orientation, internal load or control, merging them averages away exactly the differences you are trying to study. A zone, then, is a modelling judgement about which spaces share a fate - and it is the unit at which the engine reports temperature, load and comfort.

It helps to remember why the 'well-mixed' assumption is made at all: solving a full three-dimensional temperature field for every room, every hour of the year, would be impossibly slow, so energy models trade spatial detail for speed by collapsing each zone to one representative temperature. That trade is almost always worth it for whole-building energy - but it is also the reason a single zone cannot, by construction, tell you about a draught in one corner or a hot spot by the glass. Those questions belong to finer tools; the zone is deliberately coarse.

Practising energy modellers split a building along three axes. Orientation first: perimeter spaces get very different solar and conduction loads depending on which way they face, so the standard move is to peel a perimeter zone off each facade - south, east, west, north - typically to a depth of about 4-5 m where daylight and solar reach, and leave the deep interior as a core zone with no exterior wall. The classic 'five-zone-per-floor' model (four perimeters plus a core) comes straight from this logic.

Second, use: spaces with different schedules or internal loads deserve their own zone even on the same facade - a 24-hour server room, a densely-occupied conference room, a naturally-ventilated corridor. Third, HVAC: any spaces controlled by a separate thermostat, setpoint or system must be separate zones, because the model assigns conditioning per zone. A perimeter office on VAV and a core on a different air-handler cannot share a zone even if they face the same way. Put together, these three axes tell you where a split earns its keep - and, just as importantly, where it does not.

How finely you zone depends on the question. Early in design, a shoebox model - a single zone, or a handful, standing in for the whole building - is the right tool. It runs in seconds and is perfect for the big, cheap-to-change decisions: orientation, window-to-wall ratio, glazing type, the value of thermal mass or shading. You are comparing options, and a coarse model captures the direction of each change reliably even if the absolute number is rough.

Later, a detailed multi-zone model - dozens of zones matching real rooms and HVAC layout - is needed for per-room comfort, plant sizing, and code or rating compliance, where a certifier expects the geometry to reflect the building. But detail has a cost: more zones mean more inputs to get wrong, longer runs, and more places for errors to hide. The skill is matching resolution to purpose. A common mistake is building a heroic 200-zone model to answer a question a 3-zone shoebox would have answered in a tenth of the time - and with fewer ways to be wrong.

A quiet benefit of starting with a shoebox is that it forces you to name the few things that truly matter before detail can hide them - orientation, glazing ratio, envelope, mass. When the detailed model comes later, you already know what its answer should roughly be, so a wild disagreement flags an input error rather than a revelation. Working coarse-then-fine, in that order, is one of the most reliable habits an energy modeller can build.

Energy modellers routinely simplify geometry, and this is legitimate craft, not cheating. Small architectural jogs, mullions, thin partitions and minor bays that barely affect the heat balance are cleaned up so the model stays fast and robust. What you must preserve are the things the balance is sensitive to: total conditioned floor area and volume, glazing area and orientation per facade, shading, and the party walls between conditioned and unconditioned space. A good simplification keeps the physics and drops the drafting.

But every zoning and simplification choice moves the results, so you must make them consciously. Merge a hot west perimeter into a mild core and you will underestimate the west peak while overestimating the core - the average hides the problem you were hired to find. Split a naturally-ventilated space from its air-conditioned neighbour, or forget to, and the ventilation and cooling energy shift. Treat an unconditioned car park as conditioned and the EUI inflates. None of these are software bugs; they are modelling decisions. The honest habit is to document your zoning logic, so that anyone reading the result knows what lumps of air produced it - and can judge whether they were the right lumps.

A complete zone map is not only about the occupied rooms. Real models also carry unconditioned zones - car parks, stairwells, service shafts, an unconditioned warehouse - which have a heat balance but no HVAC holding a setpoint; they float, and they matter because they buffer the conditioned spaces they touch. Treating an unconditioned car park as conditioned is a classic error that silently inflates the EUI, because the model then spends imaginary energy cooling a space nobody cools.

Then there are plenum zones - the void above a suspended ceiling or below a raised floor through which return air and services run. Whether you model the plenum explicitly changes how heat from lights and the roof reaches the occupied zone below, and detailed models often include it. There are also thermal-bridge and ground-contact subtleties: a slab-on-grade floor exchanges heat with the earth quite differently from a wall exposed to air, and tools handle this with special ground-contact objects. You do not need all of this for an early shoebox - but you do need to know it exists, so that when a compliance model demands it you understand what those extra zones and objects are doing, and why leaving them out or getting them wrong shifts the answer.

The practical takeaway is to build the zone map your question needs and no finer, but to know what you left out. A shoebox that omits the plenum and lumps the car park into the conditioned volume is fine for an orientation study and misleading for a compliance submission - not because one is right and the other wrong, but because they answer different questions at different resolutions. Writing down what each zone represents, and what you deliberately simplified, is what lets the next person trust the map.