1. Catalogue capacity is an "operating point"
Judging an industrial chiller by its nominal capacity alone means examining only half of the cooling chain. The cooling a chiller delivers depends on how the chilled water (or glycol mixture) leaving the evaporator is distributed through the installation.
The capacity in the catalogue applies to specific conditions:
- Water inlet and outlet temperatures (e.g. 12/7 °C),
- Ambient air temperature for air-cooled units, condenser water temperature for water-cooled units,
- Evaporator water flow rate,
- Fluid type (pure water or glycol mixture),
- The unit's capacity control state (full load / part load).
When any of these changes, the same nominal result cannot be expected. For example, when the outdoor temperature rises above the design value or a lower leaving water temperature is required, the capacity of an air-cooled chiller drops.
One more distinction should be made from the start: the design load of a process and its actual load at a given moment are not the same thing. A process designed for 150 kW may be drawing 100 kW at that moment, depending on production rate. A sound diagnosis begins by separating these two concepts.
2. Heat balance: what capacity are we measuring in the field?
The cooling capacity carried on the water side is calculated with the following relation:
| Symbol | Meaning | Unit |
|---|---|---|
| Q | Cooling capacity | kW |
| ρ | Fluid density (≈ 998–1000 for water) | kg/m³ |
| V̇ | Volumetric flow rate | m³/s (m³/h ÷ 3600) |
| cₚ | Specific heat (≈ 4.18–4.20 for water) | kJ/(kg·K) |
| Treturn − Tsupply | Temperature difference (ΔT) | K (= °C difference) |
In a cooling circuit, water returning from the process is warmer than water supplied to it, so ΔT is positive. When the temperature range is narrow, density and specific heat at the mean temperature are sufficient.
Worked example
| Case | Flow | ΔT | Calculated capacity |
|---|---|---|---|
| Design | 25 m³/h | 5 K | 998 × (25/3600) × 4.18 × 5 ≈ 144.8 kW |
| Reduced flow, assuming the same ΔT | 18 m³/h | 5 K | ≈ 104.3 kW |
At first glance it is tempting to say "flow dropped by 28%, so the unit's capacity also dropped by 28%". From an engineering standpoint this is wrong. The second row of the table is only a comparison assumption. In reality, when flow decreases the system behaves like this:
- If the process load stays the same, less water has to carry the same heat and ΔT increases (e.g. about 7 K instead of 5 K). The heat carried may not change, but the supply water gets colder and the return water warmer; the most remote consumer may not be cooled enough.
- The chiller may approach its leaving water temperature limit; the control system may stage the compressor down or trip on low-flow protection.
Supply–return temperature difference ΔT (K)
ΔT = Q ÷ (1.16 × flow). Even if the heat carried stays the same, enough chilled water may not reach the most remote consumer.
| Flow (m³/h) | 144.8 kW process load |
|---|---|
| 14 | 8.9 |
| 16 | 7.8 |
| 18 | 6.9 |
| 20 | 6.2 |
| 22 | 5.7 |
| 24 | 5.2 |
| 26 | 4.8 |
| 28 | 4.5 |
| 30 | 4.2 |
| 32 | 3.9 |
| 34 | 3.7 |
In other words, a loss of flow shows up less as capacity "disappearing" and more as cooling not reaching the right place.
3. Measurement reliability: an inseparable part of the calculation
A capacity calculation is only as accurate as the measurements you use. Even a small temperature error can significantly affect the result:
- If the combined differential error of two temperature sensors is ±0.4 K, on a 5 K ΔT this alone means ±8% uncertainty.
- With a further ±3% error in flow measurement, the total uncertainty is about ±8.5%. A value calculated as 144.8 kW may in reality be anywhere between 132 and 157 kW.
Therefore:
- Use calibrated, matched temperature sensors,
- Install sensors in the right place (where mixing is complete, with no air pocket, inside a thermowell),
- Read flow and temperatures at the same time and record while the system is running steadily.
Capacity cannot be diagnosed from a single instantaneous temperature reading on the control panel.
4. Don't forget the fluid: the effect of glycol
In low-temperature processes or outdoor piping at risk of freezing, ethylene or propylene glycol is added to the water. Glycol protects the system from freezing but changes the calculations:
- Specific heat drops. For example, in a 30% propylene glycol mixture cₚ is around 3.85 kJ/(kg·K). Despite the slight increase in density, the heat carried at the same flow and ΔT is about 5% lower than with pure water.
- Viscosity rises. At low temperatures a glycol mixture can be several times more viscous than water; this increases pressure losses in pipes and heat exchangers and reduces the actual flow delivered by the pump.
- Chiller capacity is affected too. Heat transfer in the evaporator weakens; the manufacturer's correction factors must be applied.
5. Why doesn't the pump nameplate tell you the actual flow?
The flow on a pump's nameplate is its rated value at a specific head. The flow a centrifugal pump actually delivers in an installation occurs where the pump curve intersects the system curve.
Put simply: however much pressure the pump produces, the installation resists it. The higher the resistance, the less water the pump moves. The main sources of this resistance are:
- Pipe friction (depending on length and diameter),
- Pressure losses across the evaporator and process heat exchangers,
- Control valves, balancing valves, strainers (filters),
- Elbows, tees, diameter changes and different process branches.
- Pump curve
- System curve (clean circuit)
- System curve (clogged strainer)
Pump head / pressure loss (m)
Illustrative curves. As the strainer clogs, the system curve gets steeper; the same pump moves less water at a higher pressure.
| Flow (m³/h) | Pump curve | System curve (clean circuit) | System curve (clogged strainer) |
|---|---|---|---|
| 0 | 30 | 0 | 0 |
| 1 | 29.95 | 0 | 0.1 |
| 2 | 29.9 | 0.1 | 0.3 |
| 3 | 29.85 | 0.3 | 0.6 |
| 4 | 29.8 | 0.5 | 1.1 |
| 5 | 29.65 | 0.7 | 1.7 |
| 6 | 29.5 | 1.1 | 2.5 |
| 7 | 29.3 | 1.4 | 3.3 |
| 8 | 29.1 | 1.9 | 4.4 |
| 9 | 28.8 | 2.4 | 5.5 |
| 10 | 28.5 | 3 | 6.8 |
| 11 | 28.2 | 3.6 | 8.3 |
| 12 | 27.9 | 4.3 | 9.8 |
| 13 | 27.5 | 5 | 11.5 |
| 14 | 27.1 | 5.8 | 13.4 |
| 15 | 26.65 | 6.7 | 15.4 |
| 16 | 26.2 | 7.6 | 17.5 |
| 17 | 25.7 | 8.6 | 19.7 |
| 18 | 25.2 | 9.6 | 22.1 |
| 19 | 24.65 | 10.7 | 24.7 |
| 20 | 24.1 | 11.8 | 27.3 |
| 21 | 23.45 | 13 | 30.1 |
| 22 | 22.8 | 14.3 | 33.1 |
| 23 | 22.15 | 15.7 | – |
| 24 | 21.5 | 17 | – |
| 25 | 20.75 | 18.5 | – |
| 26 | 20 | 20 | – |
| 27 | 19.2 | 21.6 | – |
| 28 | 18.4 | 23.2 | – |
| 29 | 17.55 | 24.9 | – |
| 30 | 16.7 | 26.6 | – |
| 31 | 15.8 | 28.4 | – |
| 32 | 14.9 | 30.3 | – |
| 33 | 13.9 | 32.2 | – |
| 34 | 12.9 | 34.2 | – |
Pipe friction losses and local losses are calculated with the Darcy–Weisbach approach:
Here f is the friction factor, L the pipe length, D the internal diameter, v the water velocity, g gravitational acceleration and K the loss coefficient of the fitting. The friction factor depends on the Reynolds number and the internal roughness of the pipe.
The important practical consequence: in turbulent flow, pressure loss rises roughly with the square of the flow.
However, throttled valves, a changing flow regime and process branches switching in and out can alter this relationship. Strainers and heat exchangers that foul over time also shift the system curve upwards; a flow that was correct at commissioning may have dropped months later.
Elevation in a closed circuit
In a completely filled, closed circulation circuit, the full elevation difference to a heat exchanger on an upper floor is not added to the pump head. The static head needed for the water to rise on the supply side is balanced by the water descending on the return; the pump only overcomes friction and local losses.
Even so, the following requirements must be met separately and carefully:
- Filling the circuit completely and venting air from the highest points,
- Maintaining minimum static pressure with an expansion vessel and pressurisation,
- Ensuring sufficient pressure at the pump suction (NPSH) and preventing cavitation.
In systems with open tanks or discharging to atmosphere, static head must be taken into account when selecting the pump.
6. How low flow causes failures
Operating below the evaporator manufacturer's minimum flow:
- Lowers water velocity and reduces the heat transfer coefficient,
- Pulls the refrigerant evaporating temperature and pressure down,
- Under some conditions, causes local freezing inside the heat exchanger.
Especially in low-temperature processes, freezing risk cannot be assessed from the leaving water set point alone. The local wall temperature in the plate heat exchanger channels, the evaporating conditions and whether the water is actually distributed across all channels are decisive. A frozen plate heat exchanger can crack; water can then enter the refrigerant circuit and cause severe damage all the way to the compressor.
The right order of checks when low flow is suspected
- Verify the measurement: Are the sensors and flow measurement reliable?
- Check the strainer: Is the inlet–outlet pressure difference across the strainer above normal?
- Check the pump: Is the direction of rotation correct, and is the speed/frequency at the design value?
- Check the valves: Is any valve closed, half-closed or set incorrectly?
- Check for air: Is there an air lock or air pocket in the circuit?
- Measure heat exchanger pressure loss: What flow does the evaporator inlet–outlet pressure difference correspond to on the manufacturer's curve?
- Ask about changes to the installation: Has a new process branch been added, or have operating conditions changed?
Intervening on the refrigerant side (charge, expansion valve, compressor) before a clear water-side problem has been ruled out both prolongs the diagnosis and creates unnecessary work and leak risk.
7. More flow is not necessarily the answer
When flow looks low, the first solutions that come to mind are upsizing the pump, constantly raising the frequency or opening all balancing valves fully. These can increase flow in one branch while leaving the others even more unbalanced. Points to watch:
- Heat exchangers' maximum flow limit, design pressure and permissible pressure drop must not be exceeded.
- Excessive water velocity leads to noise, erosion in pipes and heat exchangers, degraded control valve authority and higher pump power consumption.
- In plants with several process branches, water takes the path of least resistance. While the total flow reading looks fine, the most remote or most resistive branch may be starved.
What matters is not the total m³/h but delivering the right amount of water to the right point. The way to achieve this is to measure and balance the flow in each branch.
8. Primary–secondary decoupling: when used in the right place
If two-way valves create variable flow in the process but the chiller evaporator requires a minimum flow to be maintained, primary–secondary hydraulic decoupling can be a suitable solution:
- Primary circuit: Passes water through the chiller at constant flow.
- Secondary circuit: Sends water to the process at variable flow according to demand.
Decoupling can be achieved with a common pipe (decoupler), a hydraulic separator (low-loss header) or a buffer tank suited to the design. But simply adding a buffer tank does not make the system right by itself. If the primary and secondary flows differ, mixing occurs in the separator:
- If secondary flow exceeds primary flow, warm water returning from the process mixes into the supply; water going to the process is warmer than expected.
- If primary flow exceeds secondary flow, chilled water returns straight to the chiller; the chiller's return water gets colder, the unit runs inefficiently at low ΔT and may cycle frequently.
The primary flow should therefore be selected to meet the highest secondary demand, and the flows of both circuits should be monitored.
Alternative: variable primary flow. In this design the flow through the chiller is also variable. To apply it, the chiller's minimum and maximum evaporator flow, permissible rate of flow change, control sequence and, if required, a minimum-flow bypass must be verified against manufacturer data. In industrial processes with rapidly changing loads, transient (sudden load change) analysis can be more decisive than nominal-state analysis.
9. How much does a variable-speed pump really save?
For the same centrifugal pump, at similar operating points, the affinity laws apply:
| Quantity | Relation to speed (n) | At 80% speed |
|---|---|---|
| Flow (Q) | Proportional to n | 80% |
| Head (H) | Proportional to n² | 64% |
| Power (P) | Proportional to n³ | 51.2% |
- 100% speed
- 80% speed
Relative to full speed (%)
Theoretical values. Motor and drive losses and constant differential pressure control make the real saving smaller.
| 100% speed | 80% speed | |
|---|---|---|
| Flow | 100 | 80 |
| Head | 100 | 64 |
| Power | 100 | 51.2 |
For example, when a pump with 15 kW shaft power is slowed to 80%, the theoretical power is:
On paper that is a saving of about 49%. But presenting this figure as the saving you will see on the electricity meter is not correct, because:
- There are motor and frequency converter losses; motor efficiency falls at low load.
- In closed circuits control is usually based on a constant differential pressure set point; the system curve does not start at zero, so the actual saving stays below the theoretical value.
- In open circuits, static head, control valve positions and the pump's efficiency at its new operating point change the result.
Most importantly: pump savings must be achieved without compromising the chiller's minimum flow protection. The frequency converter must be limited so that flow never falls below the evaporator minimum.
10. A field assessment example
Consider a plant designed for a 150 kW process load with 26 m³/h flow and a temperature difference of about 5 K. Measurements taken on site:
| Parameter | Design | Measured |
|---|---|---|
| Flow | 26 m³/h | 19 m³/h |
| Temperature difference (ΔT) | ≈ 5 K | 6 K |
| Calculated capacity | ≈ 150 kW | ≈ 132.1 kW |
What to read correctly from this table:
- 132.1 kW does not show that the chiller's capacity is this much. It only shows the heat carried by the water at that moment.
- Nor is it proof that the process continuously draws 150 kW. The plant may be running at a lower load at that moment; there is also measurement uncertainty (about ±8–9%) and transient effects.
- Flow is about 27% below design, and ΔT has risen accordingly. Had the same 132 kW load been carried at design flow, ΔT would have been about 4.4 K. So supply water may be colder and return water warmer than necessary; a temperature problem at the most remote consumer is likely.
For a correct assessment, the following should be recorded together over the same time period: the chiller's compressor loading and stages, set points, process production rate, strainer and evaporator pressure differences, and measurement accuracy.
No pump or chiller replacement should be recommended until the reason for the gap between 26 m³/h and 19 m³/h is clear. In most cases the cause is a clogged strainer, a mis-set valve, an air lock or a process branch added later — a problem that can be fixed at low cost.
11. Records to hand over at commissioning
A professional hydraulic commissioning report should include at least:
- Design flow and measured flow (total and per branch),
- The actual operating point on the pump curve,
- Evaporator inlet–outlet pressure difference,
- Strainer pressure loss,
- Fluid type and glycol concentration,
- Supply and return water temperatures,
- How the minimum flow protection was tested,
- Points where air was vented,
- Positions of control and balancing valves,
- Trend logs taken at different load levels.
Finally, it should be confirmed by measurement that the required temperature is reached at the most remote consumer point of the process. These records serve as a "reference snapshot" to compare against in any future performance complaint.
That is why, in Carnot's process-focused engineering approach, defining the boundary conditions of the unit and the installation together is decisive: chiller selection, hydraulic design and commissioning records are treated as a single whole.
Conclusion
Nominal chiller capacity is a necessary but not sufficient parameter for correct selection. A process is truly cooled only when heat balance and hydraulic balance are established together. When a performance problem appears in the field, the answer is not to upsize the capacity or the pump, but to measure first, prove the source of the problem, and then design the solution accordingly.
Frequently asked questions
My chiller looks big enough in the catalogue but the process isn't cooling. What should I check first?
The water side first: measure the actual flow, supply and return temperatures, and strainer and evaporator pressure differences. A large share of such complaints stem from flow and distribution problems.
How do I calculate chiller capacity in the field?
For pure water you can use Q (kW) ≈ 1.16 × flow (m³/h) × ΔT (°C). Measurements must be taken simultaneously, during steady operation and with calibrated sensors. In glycol systems, use the actual fluid properties.
If flow is low, is upsizing the pump the right solution?
Usually not. First find what is reducing the flow (clogged strainer, valve, air, an added branch). An oversized pump leads to erosion, noise, unbalanced distribution and high energy consumption.
The evaporator flow switch trips often. Can I disable it?
No. This protection prevents the evaporator from freezing and cracking at low flow. If it trips often, that is not a fault but a symptom of a problem that needs to be solved on the water side.
Does every plant need a primary–secondary system?
No. It is a suitable solution when the process has variable flow and the chiller requires a constant minimum flow. In simple constant-flow systems it can add unnecessary complexity and mixing losses.
Glossary
- Flow rate: The amount of water passing through a pipe per unit time (m³/h).
- ΔT (temperature difference): The difference between water returning from the process and water supplied to it.
- Evaporator: The heat exchanger in a chiller where the water is cooled.
- System curve: A graph of the resistance (pressure loss) the installation presents at different flows.
- Head: The pressure the pump adds to the water, expressed in metres of water column.
- Cavitation: Vapour bubbles forming at the pump suction when pressure drops too low, then collapsing and damaging the pump.
- Hydraulic separator: A low-resistance fitting that lets the primary and secondary circuits operate independently.


