HVAC
Chilled water vs DX cooling comparison field guide
The two ways a commercial building makes cold air, what the choice costs you up front and over twenty years, and how to tell which one the building actually wants.
Direct answer
Chilled water and DX are the two ways commercial buildings cool. DX (direct expansion) cools air directly with refrigerant in a coil in the airstream. Chilled water makes cold water at a central chiller and pumps it to coils. DX fits small to mid buildings; chilled water wins at large scale, but the load and the design control the choice.
Key takeaways
- DX cools air directly with refrigerant in a coil in the airstream; chilled water makes cold water at a central chiller and pumps it to coils.
- Use DX on small to mid-size buildings; use chilled water on large buildings and campuses, but load and design set the actual crossover.
- Water-cooled chillers run roughly 20 to 30 percent more efficient than air-cooled because cooling towers reject heat against the wet-bulb temperature.
- DX puts refrigerant throughout the building; chilled water confines refrigerant to the chiller room, simplifying A2L leak detection and ventilation.
- Chillers are rated under AHRI 550/590 (IPLV part-load); minimum equipment efficiencies follow ASHRAE 90.1 and the adopted energy code.
What is the difference between chilled water and DX cooling?
Every commercial building cools one of two ways, and the split is where the refrigerant lives. DX, short for direct expansion, runs refrigerant straight to a coil sitting in the airstream. The refrigerant boils inside that coil, pulls heat out of the air blowing across it, and the cooled air goes to the space. Refrigerant to air, in one step, right at the unit.
Chilled water puts a machine in the middle. A central chiller makes cold water, usually in the low 40s Fahrenheit, and pumps move that water through the building to coils in air handlers and fan coils. The water absorbs heat from the air at those coils and carries it back to the chiller. The refrigerant never leaves the chiller. The thing that travels the building is water, not refrigerant.
That is the whole fork in the road. Refrigerant-to-air at the coil, or water-to-air with a central plant making the cold water. Almost every other difference, cost, efficiency, distribution reach, redundancy, refrigerant safety, follows from that one choice. Get the fork right for the building and the rest of the design has somewhere to stand.
What is direct expansion (DX) cooling?
Direct expansion is cooling where the refrigerant evaporates in a coil placed directly in the air you are conditioning. There is no water in between. The compressor, the condenser, the metering device, and the evaporator coil make a sealed refrigeration circuit, and the evaporator coil sits in the duct or the unit where the supply air passes over it. The refrigerant expands, boils, and that phase change is what cools the air.
DX shows up in three shapes. Packaged units put the whole circuit in one cabinet: the rooftop unit, or RTU, is the common one, covered in depth in the rooftop unit installation and startup guide. Split systems separate the condensing unit outside from the evaporator coil inside, joined by a refrigerant line set. VRF, variable refrigerant flow, is the high end of split DX with one outdoor unit feeding many indoor heads.
The appeal of DX is that it is simple and distributed. Each unit is its own cooling system. One fails, the rest keep running, and a service tech can work on it without touching the building's other zones. First cost is lower and the install is faster because there is no central plant, no chiller room, and no water piping to run. That is why DX owns the small and mid-size market.
What is chilled water cooling?
Chilled water cooling makes cold water at a central chiller and pumps it through the building to coils in air handlers and fan coil units. The chiller is a packaged refrigeration machine, centrifugal, screw, or scroll, that cools a water loop instead of an airstream. Bringing that plant online and proving it makes its tons and its kW per ton is its own job, covered in the chiller plant startup and commissioning guide.
The cold water leaves the chiller, the pumps push it out into the building's piping, and at each air handler or fan coil it runs through a coil. A two-way or three-way control valve throttles the water flow to hold the air temperature the zone asks for. The water warms a few degrees giving up its cold to the air, comes back to the chiller, and the chiller cools it again. The loop runs continuously.
One big plant, water everywhere it needs to go. That centralization is the point. A chilled-water plant is efficient at scale, lasts decades, and can serve a high-rise or a whole campus from one mechanical room. It costs more to build and it needs water treatment and pumping, but on a large building it is the standard for reasons the rest of this guide works through.
When do you use chilled water vs DX?
Use DX on small and mid-size buildings and chilled water on large buildings and campuses. That is the rule of thumb, and like every rule of thumb it has a fuzzy middle where either one can be made to work and the deciding factors are cost, efficiency goals, and who will run the building.
The reason the line exists is heat transport. Water carries heat far more effectively than refrigerant lines or ducted air can, so a chilled-water loop can run hundreds of feet through a building, up a high-rise, or across a campus to multiple buildings from a single plant. DX is range-limited by refrigerant line length and by the practicality of putting a condensing unit near every zone. Past a certain size and spread, you stop adding DX units and start wanting a plant.
Where exactly the crossover sits depends on the building, not a fixed tonnage. A spread-out single-story building leans DX longer than a tall, dense one of the same area. A 24/7 load with tight efficiency targets leans chilled water sooner. Treat the size crossover as a starting point and let a load calculation and a life-cycle comparison settle the actual call, because the honest answer is that it varies with the design.
What is the difference between air-cooled and water-cooled chillers?
An air-cooled chiller rejects its heat straight to the outdoor air with a condenser coil and fans, the way a big DX condenser does. A water-cooled chiller rejects its heat into a condenser-water loop that runs out to a cooling tower, where the heat leaves the building by evaporating water. Same chilled-water output. Different way of throwing the heat away.
Water-cooled is more efficient, and the reason is wet-bulb versus dry-bulb temperature. A cooling tower rejects heat by evaporation, so it works against the wet-bulb temperature, which on most days sits well below the dry-bulb air temperature an air-cooled condenser has to fight. Lower condensing temperature means the compressor does less work for the same cooling. Published comparisons commonly put water-cooled chillers in the range of 20 to 30 percent more efficient than air-cooled, though the real spread depends on climate and load.
Air-cooled buys simplicity for that efficiency. No tower, no condenser-water pumps, no tower water treatment, no freeze protection on a rooftop loop. It sits outside and runs. Water-cooled wins on operating cost at scale and pays for the tower, the pumps, the chemical treatment, and the maintenance that comes with an open evaporative loop. Small to mid plants often go air-cooled for the simplicity. Large plants go water-cooled for the efficiency. The cooling tower is the piece that decides it.
Which is more efficient, chilled water or DX?
At large scale and at part load, a chilled-water plant, especially a water-cooled one, is more efficient than distributed DX. At small scale, DX is fine, and the efficiency gap is not worth the cost and complexity of a plant. The size of the building is what flips the answer.
Two things drive the central-plant advantage. First, big chillers are simply more efficient machines than small DX compressors, and a water-cooled condenser running against the wet-bulb beats an air-cooled DX condenser running against the dry-bulb on a hot afternoon. Second, part load. Buildings spend most of their hours well below design load, and a central plant with multiple chillers and variable-speed pumps and fans can unload smoothly and stage machines off, holding high efficiency across the part-load range. The chiller industry rates this as IPLV, the integrated part-load value, under AHRI 550/590.
DX is not inefficient, it is just optimized for a different problem. Modern packaged units and VRF carry good full-load and part-load numbers, rated under the AHRI unitary standards, and on a small building they deliver cooling at a sensible operating cost without a plant to maintain. The point is not that DX is bad. It is that the central-plant efficiency edge only pays off when there is enough building to spread the plant cost across.
First cost vs operating and life-cycle cost
DX costs less to buy and install. Chilled water costs more up front and less to run, and lasts longer, on a building big enough to use it. That is the tradeoff, and it is the one owners get wrong most often by looking only at the bid.
DX wins the first-cost line because there is no central plant. No chiller, no tower, no pump room, no building-wide water piping, and a faster install with fewer trades. You buy units and set them. For a small or mid-size building, that lower first cost usually carries the day, and the operating-cost penalty over the equipment life is too small to justify a plant.
Chilled water spends more at construction and earns it back through efficiency and longevity at scale. A centrifugal chiller and a well-run plant can serve a building for 25 to 30 years, where packaged DX units are often planned for replacement on a shorter cycle. Run the life-cycle cost, not the first cost, on any building near the crossover. On a large building the plant usually wins on total cost of ownership even though it lost the bid-day number, and that is exactly the comparison a careful owner asks for.
Distribution: refrigerant lines vs water pipes vs packaged at the zone
How the cooling gets to the space is one of the hardest practical limits between the two systems. Chilled water moves heat in pipes that can run almost anywhere in a large building. DX either packages the whole circuit at the zone or moves heat in refrigerant lines that have real length limits.
Refrigerant line length is the constraint people underestimate on split DX and VRF. Manufacturers cap the line length and the vertical separation between indoor and outdoor units, with VRF reaching several hundred feet on the longest single run, total piping into the thousands of feet, and vertical lift over a hundred feet, the exact numbers varying by system, and pushing past those limits costs capacity and reliability. Long lines hold more refrigerant, complicate oil return, and drop performance. A packaged RTU sidesteps this by keeping the whole circuit in one cabinet at the zone, but then the air has to be ducted, and duct has its own reach limit.
Water has no such ceiling for a commercial building. A chilled-water loop runs the height of a high-rise and the length of a campus, because pumping water is efficient and pipe is easy to route. That distribution reach is the structural reason chilled water owns big and spread-out buildings. When the cooling has to travel far, water travels better than refrigerant or ducted air.
Packaged vs split DX
Within DX, the first split is packaged versus split. A packaged unit holds the entire refrigeration circuit in one cabinet, factory-charged and sealed. The rooftop unit is the classic example, and its install and startup is its own discipline, covered in the rooftop unit installation and startup guide. A split system separates the condensing unit from the evaporator coil and joins them with a field-installed refrigerant line set.
Packaged is simpler to commission because the circuit is sealed and proven at the factory. You set it, connect duct, power, and condensate, check rotation, and verify a factory charge rather than building and evacuating a field circuit. The tradeoff is that the unit has to sit where it can serve the space, usually on the roof, and the air is ducted from there.
Split DX buys flexibility and pays for it with field refrigerant work. Brazing the line set, pulling a deep vacuum, and charging the circuit are jobsite steps where workmanship determines reliability, and a sloppy braze or a vacuum that never hit the target leaves moisture and non-condensables in the system. The split lets you put the noisy condensing unit outside and the coil where the air is, but every field joint is a future leak you signed for.
VRF as advanced DX
VRF, variable refrigerant flow, is split DX taken to its most capable form. One outdoor condensing unit feeds many indoor units through a shared refrigerant piping network, and the system varies the refrigerant flow to each indoor head to match its load. A single outdoor unit can serve dozens of indoor units, with some systems handling 60 or more, each controlled to its own zone.
The headline feature is heat recovery. A heat-recovery VRF system can cool some zones while heating others at the same time, moving heat from the zones that have too much to the zones that need it instead of throwing it away. In a building with a hot south side and a cold north side at the same hour, that simultaneous heating and cooling is real energy saved, and it is something a simple DX or a basic chilled-water system does not do without extra equipment.
VRF earns its place as distributed cooling for mid-size and zone-heavy buildings that want individual control without a central plant. It is still DX, though, so it carries DX's constraints in sharper form. The refrigerant piping network is extensive, the total refrigerant charge in the building is large, and the line-length and refrigerant-safety limits matter more, not less. VRF is powerful and it is not a free pass on the refrigerant questions later in this guide.
Air handlers and fan coils on chilled water
On the chilled-water side, the terminal equipment is the air handler and the fan coil. Both do the same basic thing: blow air across a chilled-water coil and deliver cooled air to the space. The difference is scale and where they sit.
An air handling unit, or AHU, is the large piece. It handles big airflows, often with outside-air intake, filtration, and sometimes a heating coil, and it serves a floor or a large zone through ductwork. A fan coil unit, or FCU, is small and local, serving a single room or a small zone, common in hotels and apartments where each room wants its own control. Both have a chilled-water coil and a control valve, and that valve is the heart of the controls.
The valve throttles water flow to hold the zone's air temperature, and how it is piped matters. A two-way valve varies flow and lets the plant run variable-flow pumping, which is where a lot of the part-load pumping savings live. A three-way valve bypasses water around the coil and keeps flow constant, simpler but less efficient. The coil, the valve, and the way the loop is pumped are where chilled-water performance is won or lost at the zone.
Redundancy, maintenance, space, and weight
The failure mode is opposite between the two systems, and so is where the equipment sits. Distributed DX fails one zone at a time and spreads its weight across the roof. A central plant, if it has no spare capacity, fails the whole building at once and concentrates its weight in one room.
DX is inherently distributed redundancy. With a dozen rooftop units, one compressor failure takes out one zone while the other eleven keep the building running, and the service call is a single unit. The flip side is a dozen units means a dozen of everything to maintain, on the roof, in the weather, each with its own filters, belts, coils, and refrigerant circuit. It also loads the roof: packaged units are heavy, and the structure has to carry the units plus snow and maintenance access, with curbs and penetrations everywhere. The win is no chiller room eating rentable floor area.
A central plant concentrates the risk, the maintenance, and the weight. Lose the only chiller and you lose all the cooling, which is why serious plants are built N+1: enough chillers, pumps, and tower cells that one can be down for service or failure and the plant still makes design capacity. That redundancy is non-negotiable on a building that cannot go warm, like a data center or a hospital. The plant wants a real mechanical room with floor space, structure for heavy equipment, and a rigging path to get the chiller in and back out at end of life. A water-cooled plant adds the cooling tower outdoors, with its own weight and access. Fewer machines in one place is easier to service well, but it has to be planned into the building from the start, not found later, and the plant is only as reliable as the spare capacity the owner paid for.
Controls and zoning
Zoning is where DX and chilled water feel different to the people in the building. DX zones in chunks of equipment. Chilled water zones in valves and air terminals, which can be finer.
A packaged DX unit is one zone, or a few if it has zoning dampers, and adding zones means adding units. That is coarse but dead simple, and a small building with a handful of RTUs is easy to control and easy to understand. VRF changes this for DX by giving every indoor head its own control, which is why VRF competes directly with chilled water on zone-heavy buildings like hotels and offices that want room-by-room setpoints.
Chilled water zones at every coil and valve, and the air side adds VAV boxes that throttle airflow per zone off a single air handler. That gives fine control across a large building from central equipment, but it leans harder on the controls system to stage the plant, reset the water and air temperatures, and keep the pumps and chillers running at their efficient points. The more central the cooling, the more the building rides on the controls being commissioned right. A plant with bad controls runs but runs expensive, and nobody notices for years.
Refrigerant charge, leaks, and A2L safety
Where the refrigerant lives is a real difference, not a detail. DX puts refrigerant throughout the building, in every unit and every line set. Chilled water keeps the refrigerant in the chiller, in the mechanical room, and sends only water out to the building.
That changes the leak and safety picture. A DX building has many refrigerant circuits and many joints, and on split and VRF systems the field-brazed connections are the usual leak points. More refrigerant spread across more of the building means more places to lose charge and more to consider for occupant safety. A chilled-water building confines the refrigerant to the chiller and its room, where it can be monitored, ventilated, and contained in one place.
The industry shift to A2L refrigerants, the mildly flammable low-GWP class replacing older high-GWP refrigerants, raises the stakes on this and is worth understanding by topic before specifying either system. A2Ls bring charge limits, leak detection, and ventilation requirements that hit a distributed DX or VRF building differently than a contained chiller room. The adopted refrigerant codes and the equipment listing govern the specifics, and they are tightening every cycle, so confirm the current requirements for the refrigerant the equipment actually uses rather than assuming last cycle's rules.
The water side: pumps, treatment, and the loop
Chilled water has a water side. DX does not, and that single fact carries a lot of the cost, the efficiency, and the maintenance difference between them. The water is an asset and a liability at once.
The chilled-water loop needs pumps to move it, and pumping is a real energy load that a good design minimizes with variable-flow pumping and two-way valves. The loop needs to be filled, air-vented, balanced, and chemically treated, because untreated water scales, corrodes, and grows biological fouling that coats the inside of the coils and the chiller tubes and quietly destroys efficiency. Hydronic loop design and balancing is a discipline of its own, worth reading by topic, and it is where a plant's real performance is set.
A water-cooled plant has a second water side, the condenser-water loop and the cooling tower, which is an open evaporative system and the harder one to keep clean. Tower water concentrates minerals as it evaporates, so it needs blowdown, makeup, and treatment, and an open tower is where Legionella risk and chemical management live. Skip the water treatment on either loop and you do not see it for a year. Then the approach temperature drifts, the kW per ton climbs, and the tubes are fouled. DX has none of this, which is a genuine simplicity advantage where it applies.
How do you choose between chilled water and DX?
Choose on building size first, then efficiency goals, redundancy needs, first-cost budget, and who will maintain it. No single factor decides it. The honest selection runs a load calculation and a life-cycle cost, and lets the building's own numbers settle the call rather than a preference.
Lean DX when the building is small to mid-size, the budget is first-cost sensitive, the zones are few or the layout is spread out, and the people maintaining it are comfortable with packaged equipment and refrigerant service. Lean chilled water when the building is large or tall, runs long hours, has aggressive efficiency targets, needs the redundancy of an N+1 plant, and has facilities staff or a service contract that can run a plant with a water side. VRF sits in the middle for zone-heavy mid-size buildings that want fine control and heat recovery without a plant.
The matrix below is a starting frame, not a verdict. Real selection weighs these factors against each other for the specific building and lets the design carry the decision.
| Factor | Leans DX | Leans chilled water |
|---|---|---|
| Building size | Small to mid-size | Large building or campus |
| First cost | Lower, faster install | Higher, central plant |
| Operating efficiency at scale | Good, optimized small | Better, especially water-cooled at part load |
| Distribution reach | Limited by line length or duct | Water runs almost anywhere |
| Redundancy | Distributed, one zone at a time | N+1 plant, paid for deliberately |
| Maintenance | Many units, spread out | Fewer machines plus a water side |
| Refrigerant location | Throughout the building | Contained in the chiller room |
| Equipment life | Shorter replacement cycle | 25 to 30 years for the plant |
Data center cooling: CRAC vs CRAH
The data center world runs this exact split under different names. A CRAC unit is DX. A CRAH unit is chilled water. Same fork, same tradeoffs, just with the precision and uptime stakes that a server room demands.
A CRAC, computer room air conditioner, is a DX unit with its own compressor and refrigerant coil, cooling the room air directly and rejecting heat outside. It is self-contained and works without a central plant, which suits smaller computer rooms and lower-availability loads, commonly cited as a fit below a few hundred kW. A CRAH, computer room air handler, has no compressor. It runs room air over a chilled-water coil fed by a separate chiller plant and modulates a control valve, the same as any chilled-water air handler.
CRAH wins the large data center for the same reasons chilled water wins large buildings generally: better efficiency, variable-speed fans without an onboard compressor, scalable heat removal, and the maintenance moved to the central plant. CRAC wins the smaller or standalone room on simplicity and independence. A hyperscale facility is almost always a chilled-water plant feeding CRAHs. A single server closet is almost always a CRAC. The decision is the same one this whole guide is about, scaled to a load that cannot go warm.
What to document in the comparison
When you put the two systems side by side for an owner, document the comparison factor by factor so the decision is on paper and defensible later. The table below is the frame we use, and the value in it is that it forces the life-cycle conversation instead of letting the first-cost number win by default.
Fill it in with the building's real numbers, the load calculation, the routed distribution lengths, the local energy rates, and the efficiency targets the project is held to, not generic figures. The owner who sees the operating cost and the redundancy laid out next to the first cost makes a different and better decision than the one who only sees the bid.
| Factor | DX (direct expansion) | Chilled water |
|---|---|---|
| Core method | Refrigerant boils in a coil in the airstream | Central chiller cools water pumped to coils |
| Refrigerant location | Throughout the building and line sets | Contained in the chiller and its room |
| Best building size | Small to mid-size, spread-out | Large, tall, or campus |
| First cost | Lower, no central plant | Higher, plant plus distribution |
| Efficiency at part load | Good with modern units and VRF | Higher, especially water-cooled |
| Redundancy model | Distributed, one zone fails at a time | N+1 central plant |
| Water side | None | Pumps, treatment, and a loop, plus a tower if water-cooled |
| Data center name | CRAC | CRAH |
Common mistakes
- Putting a chilled-water plant on a small building, where the first cost and the water-side maintenance are overkill the efficiency never pays back.
- Pushing DX too big and too spread out, adding units past the point where a central plant would have cost less to run and been easier to maintain.
- Running split DX or VRF refrigerant lines past the manufacturer's length and lift limits, losing capacity and reliability to save a plant.
- Ignoring part-load efficiency on a large DX scheme, comparing only full-load numbers when the building spends most hours at part load where a plant pulls ahead.
- Specifying chilled water and then skipping water treatment, letting scale and fouling drive the kW per ton up while nobody is watching.
- Choosing air-cooled where a water-cooled plant and tower would have paid back, or water-cooled where the building was too small to justify the tower, pumps, and treatment.
- Selling the owner on first cost alone and never running the life-cycle comparison that the building near the crossover actually needs.
Selection field checklist
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Standards and references
ASHRAE Standard 90.1 sets the minimum equipment efficiencies that both systems have to meet, for chillers and for unitary DX equipment, and the adopted energy code in the jurisdiction is what makes those minimums enforceable. Confirm the adopted edition and any local or state amendments, because the efficiency floors tighten with most cycles and the number in an old reference is not the number you are held to.
Equipment is rated under AHRI standards, and naming the right one matters. AHRI 550/590 covers water-chilling packages and standardizes both full-load efficiency and the integrated part-load value, the IPLV, that captures part-load performance. Unitary DX equipment is rated under the AHRI unitary standards. Use the rated numbers to compare machines, but treat any efficiency spread between the two systems, including the common 20 to 30 percent water-cooled advantage, as design-dependent and worth confirming for the actual climate, load, and equipment rather than carrying it as a fixed fact.
The manufacturer's data and the project specification govern the specifics. Refrigerant safety, charge limits, and ventilation follow the adopted refrigerant and mechanical codes, which are in flux as the industry moves to A2L refrigerants, so verify the current requirements for the refrigerant the equipment uses. Where the spec or the equipment listing is stricter than a code minimum or a rule of thumb, the stricter requirement controls. Size the crossover and the efficiency claims to the building, not to a brochure.
Units, terms, and synonyms
The two systems carry a pile of names and units, and the same idea reads differently across a drawing set, a chiller schedule, and a manufacturer cut sheet.
DX is direct expansion, the refrigerant-at-the-coil approach. Chilled water is sometimes written CHW. Cooling capacity is given in tons, where one ton is 12,000 BTU per hour, and chiller efficiency is given in kW per ton, where lower is better, or sometimes as a coefficient of performance. DX and unitary efficiency shows up as SEER2, EER2, or IEER for part load. Part-load chiller performance is the IPLV or NPLV. Chilled-water supply temperature is commonly in the low 40s Fahrenheit, and the data-center versions of the two systems are the CRAC, which is DX, and the CRAH, which is chilled water.
- DX / direct expansion
- Cooling where refrigerant evaporates in a coil directly in the airstream, with no water loop in between
- Chilled water / CHW
- Cooling where a central chiller makes cold water that is pumped to coils in air handlers and fan coils
- Air-cooled vs water-cooled chiller
- Heat rejected to outdoor air with a condenser coil, versus to a cooling tower through a condenser-water loop
- Ton / kW per ton
- One ton equals 12,000 BTU per hour of cooling; kW per ton is chiller input power per ton, lower is more efficient
- IPLV / NPLV
- Integrated and non-standard part-load value, the AHRI 550/590 metrics for chiller efficiency across the load range
- VRF
- Variable refrigerant flow, advanced split DX with one outdoor unit feeding many indoor units, often with heat recovery
- CRAC / CRAH
- Computer room air conditioner (DX) and computer room air handler (chilled water), the data-center forms of the two systems
FAQ
What is the difference between DX and chilled water cooling?
DX, or direct expansion, cools air directly with refrigerant boiling in a coil in the airstream. Chilled water makes cold water at a central chiller and pumps it to coils in air handlers and fan coils. The split is where the refrigerant lives: at the unit for DX, in the chiller room for chilled water.
What is direct expansion cooling?
Direct expansion cooling is cooling where the refrigerant evaporates in a coil placed directly in the air being conditioned, with no water loop between. The phase change pulls heat from the air. It comes as packaged rooftop units, split systems, and VRF, and it is simple, distributed, and lower in first cost.
When do you use chilled water vs DX?
Use DX on small to mid-size buildings for lower first cost and simplicity, and chilled water on large buildings and campuses where its efficiency and distribution reach pay off. The crossover is not a fixed tonnage. It depends on size, height, spread, efficiency goals, and a life-cycle cost the design should run.
What is the difference between air-cooled and water-cooled chillers?
An air-cooled chiller rejects heat to outdoor air with a condenser coil and fans. A water-cooled chiller rejects heat to a cooling tower through a condenser-water loop. Water-cooled is more efficient, often cited 20 to 30 percent, because the tower works against the wet-bulb temperature, but it adds a tower, pumps, and water treatment.
Is chilled water more efficient than DX?
At large scale and at part load, a chilled-water plant, especially water-cooled, is more efficient than distributed DX, because big chillers and tower condensing beat small DX compressors. At small scale the gap is not worth a plant. The efficiency spread is design-dependent, so confirm it for the actual climate and load.
Why is chilled water used in large buildings?
Water carries heat far better than refrigerant lines or ducted air, so a chilled-water loop runs hundreds of feet through a high-rise or across a campus from one plant. DX is limited by refrigerant line length and by putting a unit near every zone. Past a certain size and spread, a central plant wins on efficiency and reach.
What is the difference between a CRAC and a CRAH unit?
A CRAC, computer room air conditioner, is a DX unit with its own compressor cooling room air directly. A CRAH, computer room air handler, has no compressor and runs room air over a chilled-water coil fed by a central plant. CRAC suits smaller rooms; CRAH suits large, efficiency-focused data centers.
What are the refrigerant line length limits on a split DX or VRF system?
Manufacturers cap refrigerant line length and the vertical lift between indoor and outdoor units, with VRF often allowing several hundred feet on the longest run, more than a thousand feet of total pipe, and over a hundred feet of vertical lift, the limits varying by system. Exceeding them costs capacity, complicates oil return, and hurts reliability. Confirm the limit for the specific equipment, not a generic figure.
Does DX or chilled water cost more?
DX costs less to buy and install because there is no central plant. Chilled water costs more up front and less to run, and the plant lasts longer, on a building big enough to use it. On a building near the crossover, run the life-cycle cost, because the plant often wins on total cost even after losing the bid.
What maintenance does chilled water need that DX does not?
Chilled water has a water side DX does not. The loop needs pumps, filling, air venting, balancing, and chemical treatment, and a water-cooled plant adds a condenser-water loop and a cooling tower with blowdown, makeup, and Legionella management. Skip the treatment and fouling drives efficiency down quietly over a year.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.