In a refrigeration system, the highest installed capacity does not necessarily account for the greatest annual energy use. Problems often arise when a plant needs little cooling but its chiller cannot reduce output smoothly. Before deciding to replace the unit, it is worth examining compressor start frequency, minimum cooling capacity, flow rates and the electricity consumption of the entire system.
When does part-load operation become a problem?
Part-load operation means that the process currently requires less cooling than the system’s rated capacity. This is not inherently a fault. A correctly sized and controlled chiller can operate efficiently well below its nominal capacity.
Problems begin when demand falls below the minimum stable capacity the equipment can maintain under the prevailing conditions. For example, if the cooling load is 45 kW but the lowest available capacity of the operating unit is 90 kW, the system must periodically stop the compressor, store cooling energy in the water circuit or use another combination of units. Comparing these two capacities alone does not reveal the number of starts: water volume, control hysteresis and the actual load profile also matter.
Chiller short cycling means repeated, brief on/off cycles of a compressor or its capacity stages. Persistently excessive cycling may impair temperature stability and operating conditions. However, permissible run times, off times and start frequencies depend on the specific model; there is no single safe limit for every chiller. Trane describes the relationship between short cycling in CGAM units and circuit volume, flow rate and temperature differential settings in its compressor short-cycling guidance.
We discuss broader sources of energy losses and system efficiency in our article on industrial refrigeration. Here, we focus on what happens at low load.
How can you identify the cause of frequent starts?
The start counter is a useful clue, but it is not enough for a diagnosis. Two chillers may have similar start counts even though one suffers from low demand and the other from insufficient evaporator flow or an unstable control signal.
| Symptom in the data | Possible explanation | What to check |
|---|---|---|
| Short compressor runs and frequent stops when cooling demand is low | Demand is below the unit’s minimum capacity | Minimum capacity stage, start count, cycle duration and load profile |
| Rapid changes in supply and return water temperatures | Low water-side thermal inertia, poorly set hysteresis or changing flow rates | Effective circuit volume, sensor locations, setpoints and evaporator flow |
| Fluctuating differential pressure or flow alarms | Hydraulic problems, valve operation or incorrect pump control | Actual flow rate, valve positions and manufacturer’s minimum requirements |
| A second chiller starts briefly and is then shut down | Incorrect staging thresholds, delays or sequencing logic | Load trend, staging-in/staging-out conditions and anti-short-cycle timers |
| The process temperature deviates from setpoint despite available capacity | Problems with cooling distribution, heat exchange or end-use controls | Temperatures at critical points, flow rates and load priorities |
First, distinguish a problem at the cooling source from a problem in distribution. A compressor may short-cycle not because it is oversized, but because flow briefly stops or a poorly located sensor repeatedly triggers the controls.
Minimum capacity, water volume and hysteresis
A part-load assessment needs to consider three quantities together: the actual cooling load, the minimum available capacity of the compressor or chiller, and the energy that can safely be stored in the water circuit.
Minimum cooling capacity is not a fixed percentage for every machine
A variable-speed compressor chiller, a bank of staged compressors and a simple on/off unit have different operating characteristics. The modulation range also depends on evaporating and condensing temperatures and control limitations. It should not be assumed that a variable-speed drive will ensure stable operation at arbitrarily low loads.
The analysis requires capacity and power-consumption maps at actual operating temperatures, along with documentation of permitted start frequency, minimum run time and minimum flow. An efficiency figure from a single catalogue rating point cannot replace these data.
A buffer tank lengthens cycles but does not eliminate every cause
Too little water in the circuit means low thermal inertia: the temperature quickly reaches the stop threshold and then the restart threshold. A buffer tank can lengthen cycles if it genuinely increases the circuit’s effective water volume. However, it cannot remedy incorrect pump sequencing, insufficient flow or an incorrectly positioned sensor.
In its documentation for CGAM buffer tanks, Trane links greater circuit inertia with reduced compressor cycling and more stable water temperatures. This illustrates a solution for a particular equipment range, not a universal rule for sizing buffer tanks.
The same applies to hysteresis settings: a greater temperature difference between starting and stopping may reduce cycle frequency but can also worsen process conditions. Changes must comply with the equipment instructions, water-temperature limits and user requirements. Factory anti-short-cycle or freeze-protection safeguards must not be disabled.
Chiller staging and pump control
In a multi-chiller system, it is not enough to start another unit whenever temperature exceeds a single threshold. Controls should account for the current load, the minimum capacity of running chillers, process stability, run-time constraints and auxiliary power consumption.
At low demand, it can be better to keep one chiller running steadily rather than operate two at very low loads. There are exceptions: the best combination depends on the characteristics of the chillers, pumps and fans. Adding another unit may improve compressor efficiency while increasing circulation-pump demand. That is why it is important to compare total electrical power across the system at the same cooling load, rather than the power of an individual chiller alone.
Pump capacity control also has limits. Reducing speed can save energy, but evaporator flow must not fall below the manufacturer’s minimum. In variable-flow systems, valve operation, bypasses and hydraulic separation should be checked. The U.S. Department of Energy’s Chilled Water System Analysis Tool assesses chiller capacity and pump operation separately, including under variable-flow conditions.
A practical adjustment sequence is to establish the actual load first, then verify staging thresholds and minimum flow rates, and only then compare alternative operating configurations. Reprogramming the controls without confirming the hydraulics can make matters worse.
What should be measured, and under which operating conditions?
The most revealing data come from several recurring conditions: low demand, normal operation, peak load and transitions between them. Measurements at one steady operating point cannot explain why a chiller restarts every few minutes.
| Data to collect | What the data show |
|---|---|
| Compressor status, capacity stages, starts, run times and alarms | Cycle duration and reasons for shutdowns |
| Chilled-water supply and return temperatures and setpoints | Temperature differential and operating stability |
| Fluid flow through the circuit or a reliable indirect measurement | Calculated cooling load |
| Electrical power of chillers, pumps, fans and auxiliaries | Energy balance for the complete system |
| Outdoor temperature or condenser cooling-medium temperature | Efficiency comparisons under similar conditions |
| Temperatures at critical process points and the production schedule | Whether energy savings compromise cooling performance |
To assess short cycling, record equipment states and temperatures frequently enough to reconstruct individual cycles—often much more frequently than every 15 minutes. Choose the exact sampling interval to match how quickly conditions change and what the logger can capture. A 15-minute average profile may be sufficient for a daily energy balance but conceal brief starts and stops.
Before calculating performance, check that timestamps are aligned, sensors are calibrated and the electricity-metering boundaries are defined. If the chiller reading already includes some pump or fan power, it must not be added a second time.
COP, EER and part-load metrics – what is worth comparing?
The instantaneous cooling efficiency ratio can be expressed as cooling capacity divided by electrical input power. At chiller level, it covers only the unit itself. For a whole-system figure, the relevant pumps, fans and auxiliary equipment must also be included.
If the chilled-water flow rate is known, circuit cooling capacity can be estimated approximately as follows:
Q [kW] ≈ 1.163 × V [m³/h] × ΔT [K], where V is the water volumetric flow rate and ΔT is the difference between return and supply temperatures. The factor 1.163 is an approximation for water; for a glycol solution, use its actual density and specific heat capacity at the relevant temperature.
Metrics based on several load points, including IPLV and NPLV, are available for equipment comparisons. AHRI Standard 551/591-2026 (SI) specifies rating conditions at, among others, 100%, 75%, 50% and 25% load. Such a result helps compare equipment under a defined procedure, but it is not a measurement of a particular plant’s annual energy consumption. Production profiles, process temperatures and pump operation can differ from the rating conditions.
Nor should IPLV, NPLV, EER, SEER or SEPR be compared without checking their definitions. Their applications and test conditions depend on equipment category and the relevant standard. In an operational audit, the key quantities remain electricity consumption measured within consistent system boundaries and the actual amount of heat removed.
Example: the same cooling load, two different efficiency figures
Suppose that, during a period of reduced cooling demand, 12 m³/h of chilled water flows through the circuit and return water is 4 K warmer than supply water. The calculated cooling capacity is approximately 55.8 kW (1.163 × 12 × 4). For an illustrative water-cooled chiller system, assume the following electrical input powers. Pumps and cooling-tower fans are counted separately and are not included in the chiller input:
| Power-consuming component | Illustrative value |
|---|---|
| Chiller | 15 kW |
| Circulation pumps | 3 kW |
| Cooling-tower fans | 2 kW |
| Entire system within the defined balance boundary | 20 kW |
The ratio for the chiller alone is 55.8 / 15 ≈ 3.72, while for the entire system it is 55.8 / 20 ≈ 2.79. Both figures describe the same moment, but use different system boundaries. This is why assessing an upgrade solely on catalogue COP or instantaneous compressor power can be misleading.
This is a theoretical example using assumed figures, not measurements from an actual system. Real-world assessments must also account for measurement uncertainty and changing loads. This energy balance alone does not assess short cycling: compressor starts and run times also need to be logged. Annual savings cannot be extrapolated from a single operating point without an hourly load profile or another justified baseline method. The U.S. Department of Energy likewise highlights the importance of assessing part-load equipment efficiency and sizing pumps for typical rather than only maximum operating conditions; see section 5.3 of the Best Practices Guide for Energy-Efficient Data Center Design.
How should modernization measures be prioritized?
Buying a smaller chiller is not necessarily the first step. Some problems can be reduced by adjusting control logic, while others arise from genuine equipment or hydraulic limitations.
| Possible measure | When to consider it | Condition or risk |
|---|---|---|
| Adjust start/stop setpoints and delays | Cycles are short and process requirements permit steadier control | Follow manufacturer instructions, safety functions and temperature tolerances |
| Change the staging sequence of multiple chillers | Additional units start too early or run only briefly | Compare compressor and auxiliary energy use at the same load |
| Verify flow rates and pump control | Alarms, sharp ΔT changes or unstable pressure occur | Maintain minimum flow, evaporator operating limits and process requirements |
| Increase the effective water volume | Circuit inertia is too low relative to the minimum capacity stage | Size it based on equipment characteristics, hydraulics and permissible temperature deviations |
| Change the cooling source or modulation range | Minimum capacity remains significantly higher than sustained demand | Check load distribution, reserve capacity and whole-system economics |
If the analysis points to replacing the unit, selecting an industrial chiller requires a separate assessment of capacity, operating parameters and reserve. Simply reducing rated capacity does not guarantee good performance at peak production loads.
How can savings be verified without compromising process conditions?
A before-and-after comparison should cover broadly similar cooling demand and comparable condensing conditions, supply temperatures and production types. Ideally, divide the data into several load ranges and account for the hours spent in each. A change in energy use after modernization is not automatically attributable to it if production volume or required temperatures have also changed.
In practice, I would track four outcomes in parallel: energy use by the entire system, the number and duration of cycles, temperature deviations at critical points, and alarms and cooling availability. Lower electricity use is not a success if process stability deteriorates or equipment repeatedly approaches its operating limits.
An initial diagnosis does not require a complete measurement set. Start with temperature trends, compressor states and available energy readings. If these are insufficient, define what additional logging is needed before considering possible changes.
What information should be prepared for a system assessment?
If chillers start frequently or energy consumption rises despite a low cooling load, describe the operating conditions and share any available data: equipment types and quantities, load and temperature trends, compressor states, flow rates and electrical power. Not every measurement needs to be available at the outset.
Energy Trend can help define the diagnostic scope and assess options as part of an energy efficiency audit. The aim is to identify measures supported by measurement and evaluate their impact on energy use and process conditions—not to assume from the outset that a new machine is required.
Technical references
- AHRI Standard 551/591-2026 (SI) – chiller ratings and part-load points.
- Trane, Compressor Short Cycling CGAM – symptoms, flow and control settings.
- Trane, Buffer Tank CGAM – the role of water volume.
- U.S. Department of Energy, Chilled Water System Analysis Tool – User Manual.
- U.S. Department of Energy, Best Practices Guide for Energy-Efficient Data Center Design.
