The refrigerant never leaves the rooftop in one system, and never leaves the chiller plant in the other — the thing that actually travels through the building in a large system is just water.
Every mechanical cooling system runs the same refrigeration cycle underneath — compress, condense, expand, evaporate. What actually differs between direct expansion (DX) and chilled water systems is where that cycle physically happens and what medium carries the cooling effect the rest of the way to the occupied space. Get that one distinction straight and the rest — why homes use one approach and hospitals use the other — falls out almost automatically.
In a direct expansion system, refrigerant itself evaporates inside a coil that air is blown directly across — the evaporator sits right at (or very near) the point where the air is actually being cooled. In a chilled water system, refrigerant never leaves a centralized chiller at all; the chiller uses the same refrigeration cycle to cool water instead of air, and that chilled water is pumped through piping to coils scattered throughout the building, where the actual air-to-water heat exchange happens. Same physics, completely different architecture for how it reaches the occupant.
DX keeps things simple by locating the entire refrigeration cycle right where cooling is needed — one self-contained circuit, one zone. That works great until a building has many zones, at which point DX means many separate compressors, many separate refrigerant circuits, and many separate maintenance points, each independently sized. Chilled water solves the scaling problem by moving the refrigeration cycle into one place — a central chiller plant — and distributing the effect of that cycle (cold water) rather than the cycle itself. Water is cheap to pipe, easy to control, and carries no refrigerant charge, so a single large, efficient chiller (or a handful of them) can economically serve air handlers scattered across an entire building. The trade is centralized plant complexity and standing water infrastructure in exchange for simple, refrigerant-free equipment at every zone — a trade that pays off once a building is large enough to have many zones to serve.
Not quite. Cost matters, but it's downstream of a genuine architectural difference, not the cause of it. DX keeps the refrigeration cycle simple and localized — right at the point of cooling — but that approach doesn't scale efficiently to many separate zones without proliferating many independent refrigeration circuits, each with its own compressor and maintenance burden. Chilled water centralizes the entire refrigeration cycle into one efficient chiller plant and distributes only simple, non-refrigerant water to serve many zones throughout the building. That genuine difference in architecture — not just sticker price — is why chilled water is the standard approach for large, multi-zone commercial and institutional buildings, while DX remains the standard for smaller, simpler applicationslike homes, small commercial spaces, and standalone rooftop units. Picking one for a 40-story hospital and the other for a single-family home isn't a coin flip on price — it's the direct consequence of how each architecture scales.
Explains the real architectural difference between direct expansion (DX) and chilled water cooling systems: in DX, refrigerant evaporates directly inside a coil at the point of cooling; in chilled water, refrigerant never leaves a centralized chiller, which instead cools water that is piped throughout the building to serve many zones.
A DX system runs the entire refrigeration cycle — compressor, condenser, expansion device, and evaporator — inside (or immediately adjacent to) the air-handling equipment itself. Air is blown directly across the evaporator coil, where refrigerant is evaporating and absorbing heat, cooling the air in one step. This is the architecture behind residential split systems, packaged rooftop units, and most small commercial equipment: one refrigeration circuit, one air stream, one zone (or a small, ducted group of zones).
A chilled water system confines the entire refrigeration cycle to a centralized chiller. Instead of cooling air directly, the chiller cools water — typically down to somewhere around 42-48°F supply temperature — and a pumped piping loop carries that chilled water throughout the building to air-handling units (AHUs) and fan coil units (FCUs) at each zone. At every one of those units, the actual cooling happens through an ordinary water coil: air is blown across tubes carrying chilled water, no refrigerant involved at that point at all. The refrigerant never travels past the chiller plant's own skin.
DX is simple and inexpensive per unit, but each unit is generally sized to one zone or air-handling area — serving many zones means many separate refrigeration circuits, each with its own compressor, refrigerant charge, and maintenance schedule. Chilled water trades that proliferation for a single (or small number of) large, efficient central chiller plant, with cheap and simple water piping doing the work of reaching every zone. That centralization has real fixed costs — the plant, the pumps, the piping infrastructure — which only pay for themselves once a building has enough zones and enough square footage to spread them across. That is exactly why small buildings default to DX and large commercial or institutional buildings default to chilled water: it is an economy-of-scale decision built into the architecture, not a coin flip on unit price.
No. In a properly designed chilled water system, the entire refrigeration cycle — compressor, condenser, expansion device, evaporator — is contained within the chiller package itself. Only chilled water leaves the chiller plant and circulates through the building's piping distribution system.
It can, technically, with enough refrigerant piping run to every zone — but doing so means running long refrigerant lines throughout the building, which raises concerns around refrigerant charge, leak risk, oil return, and line sizing that become impractical at scale. In practice, large buildings that need many independently controlled zones adopt chilled water specifically to avoid these issues, using simple water piping instead of extensive refrigerant piping.
Not inherently — both rely on the same refrigeration cycle physics. Chilled water's efficiency advantage at large scale comes from economies of scale in the central chiller plant (larger compressors and heat exchangers tend to be more efficient per ton) and from the ability to optimize one plant rather than many independent DX units. At small scale, a well-selected DX unit can be just as efficient as a comparably sized chiller.
Cooling towers reject heat from the chiller's condenser to the outdoor air using evaporative cooling, typically for water-cooled chillers. They are part of the central plant's heat-rejection loop, separate from the chilled water loop that serves the building's air handlers — a chiller can also be air-cooled and skip the tower entirely, rejecting heat directly to outdoor air instead.
Yes — this is common. A large building on a central chilled water plant might still use small standalone DX units for spaces added later, IT rooms needing dedicated redundant cooling, or areas run on a different schedule than the main plant, where running new chilled water piping isn't practical.
Try our HVAC Studio
More calculators, simulators, and guides for this discipline.