Article details
Almost every cooling system, regardless of size or application, ultimately relies on the same refrigeration cycle. What differs — and what genuinely matters for design decisions — is how that cooling actually gets delivered to the space being conditioned. That's the real distinction between chilled water and direct expansion (DX) systems, and it's one of the more consequential early decisions in cooling system design.
The Core Difference
In a DX system, refrigerant itself does the cooling directly at the point of use. Refrigerant circulates from a compressor/condenser unit straight to an evaporator coil positioned in or near the space being cooled — air passes directly over that coil, refrigerant absorbs the heat, and that's the cooling effect. The refrigeration cycle happens close to where the cooling is actually needed.
In a chilled water system, the refrigeration cycle happens somewhere else entirely — inside a chiller. The chiller cools water (or a water-glycol mixture) instead of cooling air directly, and that chilled water is then pumped through a piping network to air handling units or fan coil units distributed throughout the building. Cooling is delivered indirectly: refrigerant cools water, water cools air.
That one architectural difference — refrigerant reaching the space directly, versus refrigerant staying contained in one central location and water carrying the cooling outward — is the root of nearly every practical difference between the two approaches.
Where Each Approach Tends to Make Sense
DX systems tend to fit smaller and mid-sized applications well — individual homes, small commercial spaces, standalone retail units. The refrigerant piping runs are naturally short since the compressor and the space being cooled aren't far apart, which keeps installation relatively straightforward and keeps refrigerant charge (and therefore leak risk and refrigerant cost) comparatively contained.
Chilled water systems tend to fit large buildings and campuses well — hospitals, universities, high-rise commercial buildings, large industrial facilities. Once a building is large enough, or spread out enough, running refrigerant piping directly to dozens or hundreds of separate zones becomes impractical and carries substantial refrigerant charge and leak-risk exposure. Water is a much more practical medium to distribute that far and to that many terminal units — a chilled water plant can serve an enormous, sprawling building from one centralized source.
Refrigerant Exposure: A Genuinely Important Practical Difference
This is one of the more consequential differences that doesn't always get enough attention early in a design decision: a DX system has refrigerant present at every terminal unit throughout the building — every zone with its own evaporator coil is a location where a refrigerant leak can occur. A chilled water system, by contrast, keeps essentially all of a building's refrigerant charge contained within the central chiller plant. Everywhere past the chiller, only water is circulating.
As refrigerant regulations tighten and low-GWP alternatives (often with different flammability and charge-limit characteristics) become more common, this containment difference is becoming a more actively weighed factor in system selection, not just a minor technical footnote.
Maintenance and Redundancy Considerations
A DX system's simplicity cuts both ways. With fewer major components and a more direct refrigerant path, individual DX units are often simpler to service — a technician can diagnose and repair one unit without necessarily affecting the rest of the building. But that same decentralization means a building with many DX units has many separate pieces of equipment to maintain over time, each with its own service life and failure points.
A chilled water system concentrates the mechanical complexity into fewer, larger, more centralized pieces of equipment — the chiller plant itself. This makes centralized monitoring and major maintenance more consolidated, and well-designed chiller plants often incorporate genuine redundancy (multiple chillers, so one can be serviced while others carry the load). But it also means the central plant is a more consequential single point of failure if redundancy isn't properly designed in — losing the chiller plant affects cooling everywhere downstream simultaneously, unlike a DX failure that's typically isolated to one zone.
Efficiency: It Depends More Than It Might Seem
Neither approach is universally more efficient — it depends heavily on scale, load diversity, and how well the system is designed and controlled. Large chilled water plants can take advantage of load diversity across a big building (not every zone peaks at the same time) and can be designed with high-efficiency centrifugal or screw chillers that outperform many smaller individual DX units on a straight efficiency basis at that scale. But a chilled water system also introduces additional energy losses that DX systems avoid entirely — pumping energy to move water throughout the building, and an extra heat transfer step (refrigerant-to-water, then water-to-air) that a direct refrigerant-to-air DX system doesn't have.
At smaller scale, a well-selected DX system often wins on straightforward efficiency and simplicity, precisely because it avoids the pumping and extra heat-exchange losses that don't pay for themselves without a large enough building to justify the central plant's efficiency advantages.
How to Actually Think About the Choice ?
Rather than treating this as "which system is better," the more useful framing is: at what point does a building's scale and layout justify the added complexity of a central chilled water plant?
Small, compact buildings rarely reach that point — DX usually wins on simplicity and cost. Large, sprawling, or high-rise buildings usually justify it — the practical challenges of running refrigerant piping to dozens of zones, combined with load diversity benefits and centralized efficiency potential, tip the balance toward chilled water. The genuinely interesting design decisions happen in the middle ground, where building size, layout, redundancy requirements, and refrigerant strategy all have to be weighed together rather than defaulting to either approach out of habit.
EveryEng. For everyone in Engineering.