Article details
Deciding when to bring a second chiller online, or take one offline, sounds like it should be simple, add capacity when load increases, remove it when load drops. In practice, a plant needs a genuinely reliable signal to actually detect that changing load in real time, and different plants use genuinely different signals to do it. Understanding these different staging approaches, and why each one measures what it measures, is what separates a plant that stages smoothly from one that hunts, overshoots, or stages at the wrong moment entirely.
Why Staging Needs a Real Signal, Not Just a Guess
A chiller plant with multiple chillers has to decide, continuously, whether the currently operating chiller or chillers can handle the actual current load, or whether an additional chiller genuinely needs to come online, and the reverse, whether an operating chiller has become unnecessary and should shut down to avoid running inefficiently at very low, wasteful partial load. Getting this decision wrong in either direction carries a real cost, staging up too late risks losing temperature control entirely, staging up too early or too often wastes energy and adds unnecessary equipment wear, and staging down too aggressively risks immediately needing to stage back up again shortly after.
Load Based Staging
Load based staging uses each chiller's actual percentage of rated capacity as the signal driving staging decisions. The genuinely important detail here is that the trigger points shouldn't be arbitrary round numbers, they should be matched to where each specific chiller actually performs most efficiently. If a chiller's efficient operating range sits between roughly 60 and 80 percent load, a reasonable staging strategy brings the next chiller online once the operating chiller has been running above that efficient range, say around 85 percent load, for a sustained period, rather than reacting to a brief, momentary spike. Staging down follows the same logic in reverse, if the most recently added chiller has been running at a low load, commonly cited around 60 percent or below, for a sustained period, it becomes a genuine candidate to shut down.
ASHRAE Guideline 36 formalizes a version of this using staging part load ratio bounds, keeping the threshold for adding a chiller high enough that doing so doesn't leave both chillers operating at an unreasonably low, inefficient load immediately afterward, while also keeping the threshold from climbing so high that the plant risks genuinely losing the load before staging up. Both stage up and stage down decisions typically require sustained conditions over a defined time period, not an instantaneous crossing, specifically to avoid staging in response to brief, temporary fluctuations that don't reflect a genuine, sustained change in actual demand.
Flow Based Staging
Flow based staging uses the actual flow rate through the secondary loop, the building distribution side, as the signal instead of a calculated load percentage. As covered in primary secondary pumping explained, the decoupler line connecting primary and secondary loops is where this signal often gets measured directly, a flow sensor positioned in the decoupler can detect whether secondary flow demand has grown to the point where it's actually exceeding what the current primary flow can supply, sometimes detected specifically as flow reversal direction through the decoupler itself, secondary flow effectively pulling more from the primary side than the primary side alone is producing.
A genuinely practical sequence of operation built around this approach stages up when secondary flow exceeds a defined threshold, commonly in the range of a specific gallons per minute figure specific to the plant's actual design, sustained for a defined period, often around fifteen minutes, and stages down when flow drops below a separate, generally lower threshold, sustained for a similar period. Building in a deliberate minimum time between any stage up or stage down event, often twenty minutes or more, is a specifically important detail, allowing the system to reach a genuine steady state after each transition before considering another change, rather than repeatedly staging back and forth in response to normal, ordinary flow fluctuation.
Temperature Based Staging
Temperature based staging uses chilled water supply or return temperature directly as the triggering signal, watching for the return water temperature to climb above an acceptable setpoint, or the supply temperature to rise above its target, as the indicator that the currently operating chillers can no longer adequately handle the actual load.
A representative sequence monitors return water temperature against a setpoint, commonly citing figures in the range of six or seven degrees Celsius as a representative setpoint, and if that return temperature exceeds the setpoint by a defined deadband for a sustained delay, often in the range of three to five minutes, the next chiller stages on. Staging down follows a mirrored logic, when return temperature drops comfortably below the setpoint for a sustained period, the most recently added chiller can be shut down, though critically only after that chiller has already satisfied a defined minimum run time, preventing a chiller from starting and stopping in rapid succession purely because temperature happened to dip briefly right after it came online.
A related, more advanced variant uses lift, the difference between condenser water return temperature and chilled water supply temperature, as the actual staging signal for fixed speed chillers specifically, since lift correlates directly with how hard a chiller is actually working. Staging up at a higher lift value and staging down at a meaningfully lower lift value, rather than using the identical threshold for both directions, is a deliberate design choice, building in a buffer specifically to avoid a chiller staging up and immediately staging back down again in a short, wasteful cycle.
Why Every Approach Needs the Same Underlying Safeguards
Regardless of which signal actually drives the staging decision, load, flow, or temperature, the same three protective principles show up consistently across all three approaches. A sustained time delay before acting on any trigger, rather than reacting to an instantaneous reading, prevents staging in response to momentary noise rather than genuine, real demand change. A minimum time between staging transitions, separate from the trigger delay itself, allows the plant to reach genuine steady state after a change before another change is even considered. And using different, deliberately separated thresholds for staging up versus staging down, rather than a single shared trigger point, prevents a chiller from oscillating on and off right at one narrow boundary condition.
Which Approach Actually Fits Which Plant
Load based staging tends to fit plants where each chiller's actual capacity and efficient operating range are well understood and can be reliably calculated or measured in real time. Flow based staging fits primary secondary plants particularly well, since it directly measures the exact condition, secondary demand exceeding primary supply, that the decoupler arrangement exists specifically to accommodate. Temperature based staging is often the most broadly applicable and most directly tied to the plant's actual purpose, delivering water at an acceptable temperature, though it can react somewhat more slowly to a genuinely fast changing load than a flow based signal would, since temperature has to actually drift before it triggers a response, whereas flow can reflect a demand change more immediately.
Many real plants combine more than one of these signals rather than relying on a single approach exclusively, using temperature as a primary safeguard trigger, for instance, while using flow or load percentage as a more responsive, earlier warning signal, precisely because each approach has a genuinely different strength and a genuinely different blind spot.
The Practical Takeaway
Chiller staging isn't simply "add capacity when needed, remove it when not," it's a genuine control engineering decision about which measurable signal, load percentage, flow rate, or temperature, most reliably reflects actual changing demand for a specific plant's particular configuration, combined with deliberate time delays and separated up and down thresholds to prevent the plant from reacting to noise or oscillating unnecessarily. Understanding why each signal works the way it does, not just which threshold number a specific plant happens to use, is what allows genuinely intelligent staging logic rather than simply copying a sequence of operation from a different plant that may have entirely different flow rates, chiller capacities, and load characteristics.
EveryEng. For everyone in Engineering.