Chiller Staging Logic: Threshold Selection, Anti-Recycle Timers, and Part-Load Efficiency

Most chiller plants operate on one of two extremes. Operators who stage up too late, riding one chiller toward surge because startup is disruptive. Or plants with aggressive automation that stage too early, running two chillers at 35% load each, deep in the part-load zone where kW/TR degrades.

Both are expensive. The first risks compressor damage. The second wastes energy running two machines when one would serve the load at a better operating point.

Correct staging requires thresholds derived from chiller thermodynamics: where the part-load efficiency curve inflects, where the compressor approaches its physical limits, and what safeguards protect the equipment from the consequences of rapid load changes. This post covers the threshold engineering, the operational safeguards, and the sequencing logic that determines which chiller loads first.

Stage-Up and Stage-Down Thresholds: What They Protect and How to Set Them

Staging thresholds answer two questions. Stage-up: at what aggregate load should a second chiller start? Stage-down: at what aggregate load should a running chiller shut down? The answers are not arbitrary conventions. They are derived from the compressor's operating envelope and its part-load efficiency curve.

Stage-up at approximately 85% aggregate capacity

The stage-up threshold protects against compressor overload. For centrifugal chillers with variable geometry vanes (VGV), the critical constraint is surge. As the cooling load increases, the compressor must maintain a sufficient pressure ratio across the impeller. The VGV opens progressively to increase gas flow.

Above approximately 85 to 90% of rated capacity, the compressor operates near the limit of its stable operating envelope. The exact surge boundary depends on condenser water temperature: at elevated CWS (fouled condenser or high wet-bulb), the surge line shifts inward and the chiller may approach surge at 75 to 80% capacity. At low CWS (cool day, clean condenser), it can safely run to 90 to 95%.

The 85% threshold is set conservatively to provide margin across the expected range of condenser conditions. If load continues to rise beyond the surge boundary, the impeller cannot maintain head, gas flow reverses momentarily, and the compressor surges. Surge events cause violent vibration, bearing damage, and impeller erosion. A single severe surge can take a chiller offline for weeks.

The 85% threshold provides headroom. It ensures a second chiller begins its startup sequence while the running machine still has margin before the surge boundary. Centrifugal chillers typically require 5 to 10 minutes from start command to full capacity (oil pressure verification, soft-start ramp, VGV opening sequence).

During this ramp-up, the running chiller must hold the full building load. The stage-up threshold must account for both the staging delay and the ramp-up time: if the building can gain 5% load in 10 minutes, an 85% trigger with a 5-minute delay means the running chiller may reach 90% before the second machine is carrying load.

The stage-up delay (typically 5 minutes) prevents false triggers from transient load spikes, but once the load sustains above 85% for the delay period, the second chiller must start.

For screw chillers with slide valve unloading, surge is not the failure mode. At high load, the slide valve is at full extension, oil discharge temperature rises, and the compressor approaches its thermal limit. The stage-up threshold is still in the 80 to 90% range, but the rationale is thermal protection rather than aerodynamic instability.

An additional constraint in Indian commercial installations: utility demand charges mean that starting a second chiller creates an inrush spike that may set a new peak kVA for the billing period, adding a financial penalty to the staging decision.

Stage-down at approximately 40% aggregate capacity

The stage-down threshold protects against inefficient operation. Every chiller has a part-load factor (PLF) curve that describes how its kW/TR varies with the fraction of rated capacity being delivered. At full load, kW/TR is at its rated value. As the load drops, kW/TR initially improves: the compressor unloads and the power consumption drops faster than the cooling output.

Most centrifugal and screw chillers reach their best kW/TR at approximately 70 to 80% PLR.

Below this sweet spot, the curve inverts. kW/TR begins to degrade as the compressor operates less efficiently. At 40% PLR, kW/TR has returned to approximately full-load levels or slightly worse, with VGV nearly closed (centrifugal) or slide valve deeply retracted (screw).

Below 30% PLR, kW/TR degrades sharply and centrifugals approach their minimum stable operating point, where even small load fluctuations risk triggering surge protection.

The 40% threshold is set at the inflection point where continued operation becomes thermodynamically expensive. De-staging one chiller and letting the remaining machine(s) absorb the load at a higher PLR (closer to the 70 to 80% sweet spot) reduces total chiller kW.

This 40% threshold applies to fixed-speed chillers with VGV or slide valve capacity control. Variable-speed drive (VSD) centrifugal chillers have a different profile: compressor speed tracks load directly, and kW/TR continues to improve at low PLR. For VSD chillers, the stage-down threshold can be 25 to 30%. Plants mixing VSD and fixed-speed machines require different thresholds for each type.

The two-chiller decision at the boundary

Consider a 600TR plant with two identical 300TR chillers at 50% aggregate load (300TR total). Run both at 50% PLR each, or run one at 100% PLR?

The chiller-side comparison is close to break-even (at 50% PLR, kW/TR is near rated levels for most machines), but the second chiller brings auxiliary equipment that tips the balance.

In a plant with 22 kW CW pumps and 15 kW primary pumps, the second chiller adds 37 kW of auxiliary load (one CW pump plus one primary pump). Even if the two-chiller option saves 5 to 10 kW on the chiller side from better part-load efficiency, the 37 kW auxiliary penalty makes the single-chiller option materially cheaper in total plant kW.

The stage-down at 40% reflects this: below 40% aggregate (120TR on a 600TR plant), running two chillers at 60TR each (20% PLR) is clearly worse than running one at 120TR (40% PLR).

These thresholds are plant-specific. They depend on the chiller manufacturer's PLF curve, the condenser water temperature (which shifts the surge boundary), and the auxiliary equipment configuration. Applying generic thresholds without examining the chiller's actual performance curve is a controls engineering shortcut that costs energy.

Anti-Recycle, Minimum Run Time, and Lead Rotation: The Operational Safeguards

Staging thresholds determine when to add or remove a chiller. The operational safeguards determine what happens between those decisions.

Anti-recycle timer

The anti-recycle timer prevents a chiller that has just been shut down from restarting immediately. After shutdown, high-pressure refrigerant migrates through the circuit as pressures equalize. Oil entrained in the refrigerant needs time to drain back to the compressor sump.

Restarting before equalization risks liquid slugging (incompressible liquid refrigerant damaging valves and rods) and oil starvation (insufficient lubrication on restart).

A typical anti-recycle timer for water-cooled centrifugal and screw chillers is 15 to 20 minutes. This is configurable per site because different chiller manufacturers and refrigerant types have different equalization times. An R-134a centrifugal may need a different lockout than an R-410A screw.

Minimum run time

Once a chiller starts, it must run for a minimum period before it can be shut down. On startup, the compressor circulates refrigerant carrying oil through the circuit. The oil must complete one or more full circuits and return to the sump before the compressor can safely stop.

Short-cycling starves bearings of lubrication and subjects the compressor to thermal shock from differential expansion.

A typical minimum run time is 10 to 15 minutes. This timer overrides any stage-down request: if the load drops immediately after a stage-up (a common scenario when a conference room empties or cloud cover reduces solar gain), the newly started chiller continues running until the minimum run time expires, even if the staging logic says it should shut down.

The scenario where both bite: Load oscillation near a staging boundary

Consider a plant oscillating around 85% aggregate capacity. Without safeguards: load exceeds 85%, stage-up delay expires, second chiller starts. Load drops below 40% (the second chiller absorbs the excess), stage-down triggers, second chiller shuts down.

Load rises above 85% again, stage-up triggers, but the second chiller cannot restart because the anti-recycle timer has not expired. Comfort fails until the timer clears.

This oscillation is the most common algorithmic failure mode in staging logic. The engineering response is to set the stage-up and stage-down thresholds far enough apart (85% and 40%, a 45-percentage-point gap) that transient load fluctuations do not cross both boundaries within one timer cycle.

The asymmetry in delay timers is deliberate. The stage-up delay (typically 5 minutes) is shorter than the stage-down delay (typically 10 minutes). The reasoning reflects the cost of errors: staging up too late risks comfort complaints and compressor surge (high consequence). Staging down too late wastes energy but causes no equipment damage (lower consequence). The system is biased toward starting chillers faster than it stops them.

The most common failure mode is human, not algorithmic. Night-shift operators who switch to manual because the staging "keeps cycling" are responding to a real symptom (thresholds too close together or timers too short), but the fix is to correct the parameters, not disable the automation.

The operator who always runs both chillers "to be safe" adds approximately 37 kW of unnecessary auxiliary power. Over a year, that is over ₹10 lakh in pump energy that delivers no additional cooling. A well-configured staging sequence that the operator trusts is worth more than a sophisticated algorithm overridden to manual by the second week.

Lead rotation

In a multi-chiller plant, lead rotation determines which chiller starts first. The simplest approach rotates the lead on a fixed daily or weekly schedule. This equalizes starts but not run hours, and it ignores efficiency differences between machines.

Runtime-based rotation improves on this by selecting the chiller with the lowest accumulated operating hours as lead, equalizing wear across the fleet and distributing maintenance intervals. However, it can force the less efficient machine into the lead position during its rotation, increasing total plant kW during those hours.

A fixed lead is appropriate when one machine is materially more efficient than the others and the maintenance team schedules the standby chiller independently.

The tension between rotation and efficiency is real. Runtime equalization says: load the underused chiller. Efficiency-based sequencing says: load the most efficient chiller. In a fleet of identical machines, these converge. In a mixed fleet (different ages, different types, different COP), they diverge.

The resolution is a weighted approach: prioritize efficiency-based sequencing during normal operation, but shift to runtime-based rotation when the run-hour delta between machines exceeds a configurable threshold (e.g., 500 hours).

Efficiency-Based Sequencing: Loading the Most Efficient First, Unloading the Least Efficient First

In a plant with identical chillers of the same age, staging order makes no difference to total plant kW. In a plant with a 5-year-old centrifugal (COP 5.4) and a 12-year-old screw (COP 4.3), it determines whether you spend ₹3 to 5 lakh more per year than you need to.

The logic: when staging up, load the chiller with the best kW/TR at the current operating point. When staging down, unload the worst. The most efficient machine carries the base load during the majority of operating hours (50 to 80% load band), and the less efficient machine runs only during peak periods.

Determining "most efficient" requires measurement, not nameplate comparison. A chiller's actual kW/TR depends on current PLR, condenser water temperature, chilled water setpoint, and mechanical condition (refrigerant charge, compressor wear, fouling).

A chiller rated at COP 6.0 but with 20% charge loss may operate at COP 4.8, making it less efficient than an older machine in better condition.

This connects to continuous chiller performance monitoring. The rated-vs-actual kW/TR scatter plot provides the data that feeds efficiency-based sequencing: if CH-1 consistently operates above its rated curve while CH-2 is on or below it, the staging logic should prefer CH-2 as lead regardless of nameplate ratings.

The mixed-fleet problem adds a capacity dimension. Consider a 500TR centrifugal (COP 5.4) and a 250TR screw (COP 4.8) at 400TR building load.

Option A: centrifugal alone at 80% PLR, chiller kW = (400 × 3.517) / 5.4 = 260 kW, one set of auxiliary equipment.

Option B: both machines (200TR each), chiller kW = 130 + 147 = 277 kW, two sets of auxiliary equipment.

Option A wins by 17 kW on chillers plus the auxiliary savings. The decision reverses at higher loads where the centrifugal alone would exceed 85% PLR. The staging logic must evaluate total plant kW for each combination at the current load, not just per-chiller COP.

Chiller Staging Is a Thermodynamic Decision

Staging is the most consequential automated decision a BMS makes in a chiller plant. Every staging event changes total plant kW by 30 to 50%, affects compressor life, and determines comfort response time.

Tor Shield's Chiller Plant Manager automates staging with configurable thresholds (85% stage-up, 40% stage-down), anti-recycle timers (default 20 minutes), minimum run times (default 15 minutes), and three lead rotation strategies (Round Robin, Runtime-based, and Fixed).

The sequencing configuration is visible on the CPM settings page alongside the plant schedule and setpoint controls. Efficiency-based sequencing uses the same real-time kW/TR data that feeds the rated-vs-actual performance scatter plot, ensuring the staging logic responds to the chiller's current operating efficiency, not just its nameplate rating.

Chiller sequencing is not "turn on the next one when it gets hot." It is a thermodynamic decision made dozens of times per day.