How Is a Central Chilled Water Plant Properly Designed?

Chiller selection starts with the load profile, not the peak number. I build an annual cooling load profile from the building simulation or campus metering data, then test chiller lineups against it — counting run hours in each staging combination and computing annual energy, not just design-day efficiency. This is where unequal machine sizes earn their keep: a plant with a small, a medium, and a large chiller spends most of its hours running one or two machines near full load instead of short-cycling a single giant. Staging controls decide which machines run and when. I sequence on measured load and chiller efficiency curves rather than simple timers, with rotation to equalize wear, automatic standby start on failure, and condenser water temperature reset that lets chillers unload as the towers produce colder water. Magnetic-bearing centrifugal and high-efficiency screw machines have made part-load performance so good that the old rule of thumb about avoiding part load is obsolete — but only if the staging sequence actually uses that capability instead of fighting it.

The distribution system carries the plant's efficiency to the building, or squanders it. In a primary/secondary plant I size the decoupler for low velocity and near-zero pressure drop so the primary and secondary loops truly decouple, and I instrument it — flow direction and temperature in the decoupler tells the operator instantly whether the plant is keeping up or the building is misbehaving. Secondary pumps get VFDs and differential-pressure control reset by valve position, so pump speed follows the most demanding coil rather than a fixed setpoint. In a variable primary plant, the design centers on chiller protection: minimum flow bypass with fast-acting control, flow meters on each chiller, and sequencing that never strands a running chiller below its minimum. Both architectures live or die on coil delta-T, so I specify pressure-independent control valves at air handlers, design coils for realistic entering conditions, and put system delta-T on the operator's dashboard with an alarm. The piping itself gets reverse-return or carefully balanced direct-return layouts, air elimination, and expansion handling sized for the full system volume — details that decide whether the plant is maintainable or miserable.

Cooling towers are selected for the design wet-bulb with an approach that balances tower cost against chiller efficiency — colder condenser water makes chillers dramatically more efficient, up to the manufacturer's limits. I arrange multiple cells so any cell can be isolated for maintenance or winter operation without losing the plant, and I place towers considering plume, drift, noise, and distance from intakes and occupied outdoor spaces. Makeup water, blowdown, and basin heating for freeze protection are designed as systems, not accessories. Water treatment protects the entire investment. Scale on a condenser tube costs more in a year of wasted energy than treatment costs in several, so I design filtration, chemical or non-chemical treatment, and automated blowdown control from day one — and I meter makeup and blowdown so the operator can see the treatment working. Legionella risk management follows the industry guidance for tower operation and maintenance, with the design providing the drains, access, and monitoring points the water management plan needs. • Size chillers to the block load with N+1 redundancy and unequal machines for load-following • Choose variable primary flow for pump savings or primary/secondary for simplicity — deliberately • Fight low delta-T syndrome with pressure-independent valves and BAS delta-T alarming • Select towers for design wet-bulb with cell redundancy and proper placement • Design water treatment, blowdown, and legionella management as core systems • Model waterside economizer hours with local weather data on every plant