Chilled Water Low Delta-T Syndrome Case Study

Low delta-T syndrome in a chilled-water plant occurs when the distribution system returns water to the plant colder than expected. The building may still receive cooling, but the same cooling load is carried with excessive water flow. Pumps run harder, control valves lose authority, bypasses hide coil problems, and the plant may stage additional chillers even though the cooling load is below the capacity of one machine.

This case study follows a university laboratory building connected to a central chilled-water plant. Operators report that two chillers and two secondary pumps are running during moderate weather, even when the building automation system shows a load that should fit within one chiller. The engineering question is whether the second chiller is required by real load or forced by low return temperature and overpumping.

The purpose is to show how an engineer can use a water-side heat balance, return-temperature mixing, pump power and BAS trend evidence to separate plant capacity problems from a distribution fault.

Case Context

The building uses variable-air-volume air-handling units with chilled-water cooling coils. A decoupled secondary chilled-water loop serves the building; the central plant sees the secondary return temperature after all coils, bypasses and control valves.

Item	Value or observation
Chilled-water supply temperature to building	6.0^\circ\text{C}
Design return temperature	14.0^\circ\text{C}
Design temperature rise	8.0\ \text{K}
Measured building return temperature	9.8^\circ\text{C}
Measured secondary flow	95\ \text{L/s}
Water density used for screening	1000\ \text{kg/m}^3
Water specific heat	4.186\ \text{kJ/(kg K)}
Secondary differential pressure during event	210\ \text{kPa}
Pump and motor combined efficiency	0.62
One chiller nominal capacity	1600\ \text{kW}
One chiller maximum evaporator flow	55\ \text{L/s}
Affected operating hours	1900\ \text{h/year}
Electricity tariff used for screening	\0.13/\text{kWh}$

The design intent is a high enough water-side temperature rise that coils remove heat efficiently without forcing excessive flow through the plant. The measured 3.8\ \text{K} rise is less than half the design value.

Service Boundary

The analysis boundary is the secondary chilled-water loop entering and leaving the building. That boundary includes cooling coils, control valves, local bypasses, strainers, differential-pressure control, sensors and secondary pumps. It excludes the air-side load calculation except where air-side trends are used as a reasonableness check.

This boundary matters because low delta-T is not automatically a chiller fault. A chiller can be healthy while the distribution loop sends it too much cold return water.

Measured Plant Heat Balance

For liquid water near HVAC temperatures, the cooling load carried by the loop is:

with \dot m in \text{kg/s}, c_p in \text{kJ/(kg K)}, and temperature difference in \text{K}. With 95\ \text{L/s} of water, the mass flow is approximately 95\ \text{kg/s}. The measured temperature rise is:

The cooling load is therefore:

The building is not lightly loaded. It is removing about 1.51\ \text{MW} of heat, which is close to the capacity of one 1600\ \text{kW} chiller. The problem is that it is doing so with far too much water flow.

Flow Required at Design Delta-T

If the same 1511\ \text{kW} load were carried at the design 8.0\ \text{K} temperature rise, the required mass flow would be:

For water at this screening level, that is about 45.1\ \text{L/s}. The measured flow is:

The system is moving a little more than twice the water required for the actual load. This is the central signature of low delta-T syndrome: the plant load looks real, but the flow required to carry it is inflated by a diluted return temperature.

Pump Power Penalty

The measured pump input power can be screened from flow and differential pressure:

where \dot V is volumetric flow in \text{m}^3/\text{s}, \Delta p is pressure in pascal, and \eta is combined pump and motor efficiency. At 95\ \text{L/s}:

At the same differential pressure but with only 45.1\ \text{L/s}:

That is a conservative comparison because a variable-speed distribution pump should not need the same head at half the flow. If the operating point follows the pump affinity relationship approximately, power scales with the cube of flow:

The exact number depends on the control curve and minimum head constraints, but the direction is unambiguous. Low delta-T is converting a cooling-load problem into a pumping-energy problem.

Mixing Evidence

The field investigation did not start by blaming the chiller. The BAS trend package separated productive coil flow from suspected bypass and leakage paths.

Stream	Flow	Return temperature
Cooling coils with real air-side load	83\ \text{L/s}	10.3^\circ\text{C}
Bypass and leaking valve paths	12\ \text{L/s}	6.5^\circ\text{C}
Mixed return to plant	95\ \text{L/s}	unknown before calculation

The mixed return temperature predicted from these streams is:

This matches the measured 9.8^\circ\text{C} return temperature. The agreement is important because it explains the plant symptom without inventing extra load. About 12\ \text{L/s} of cold water is returning with little heat pickup and diluting the productive return stream.

The bypass flow may come from a deliberately open decoupler, a three-way valve left in bypass, a leaking two-way valve, a failed actuator, incorrect differential-pressure reset or control valves forced open by excessive pressure. The calculation does not identify the component by itself, but it makes the fault class testable.

Why the Second Chiller Started

The plant has one 1600\ \text{kW} chiller, which is large enough for the measured 1511\ \text{kW} load at design delta-T. However, one machine can accept only 55\ \text{L/s} of evaporator flow. The building was demanding 95\ \text{L/s} because the return water was too cold.

That creates a staging trap:

1. The load is below one-chiller capacity.
2. The required flow at the measured delta-T exceeds one chiller’s evaporator flow limit.
3. The plant stages a second chiller to provide flow path capacity, not because the building truly needs two chillers of cooling capacity.

This is why low delta-T is operationally expensive. The plant can be capacity-rich and still behave as if it is constrained.

Energy Cost Screening

The trend data show a measured operating COP of 4.4 with two chillers and high pumping flow. A corrected single-chiller operating state is expected to reach about 5.2 under the same weather and load after bypass correction and pump reset. For a 1511\ \text{kW} load:

The compressor, tower and auxiliary penalty is approximately:

Adding the variable-speed pump penalty estimate of 32.2 - 3.4 = 28.8\ \text{kW} gives:

For 1900\ \text{h/year}:

At \0.13/\text{kWh}$, the screening cost is:

This is not a utility-grade measurement and verification result. It is an engineering screening estimate used to justify fault isolation, valve repair and post-correction trending.

Root Cause Findings

The investigation found three interacting faults:

Finding	Engineering effect
A normally closed bypass valve leaked about 7\ \text{L/s}	Cold supply water returned to the plant with almost no heat pickup.
Two large AHU cooling-coil valves leaked when commanded below 10\%	Coil flow remained high even when air-side load was low.
The differential-pressure reset sequence held 210\ \text{kPa} during part-load operation	Control valves lost authority and leakage paths carried more flow.

None of these faults alone explains the entire plant symptom, but together they reproduce the measured mixed return temperature and excessive flow.

Measurement Uncertainty

The temperature sensors used for the heat balance were checked against a calibrated reference over the operating range. The remaining uncertainty after correction was estimated at \pm 0.15\ \text{K} for each temperature measurement. The uncertainty in measured delta-T is:

Relative to the measured 3.8\ \text{K} temperature rise:

The ultrasonic flow meter uncertainty was estimated at \pm 3\%. Combining flow and delta-T uncertainty for the heat balance:

This uncertainty is not large enough to explain the difference between 3.8\ \text{K} measured delta-T and 8.0\ \text{K} design delta-T. The low delta-T diagnosis is therefore robust.

Corrective Action

The corrective work package avoided replacing equipment before proving the distribution fault:

1. Repair the leaking bypass valve and add a temporary clamp-on flow check after repair.
2. Stroke the two suspect coil valves under differential-pressure conditions matching occupied operation.
3. Replace worn valve seats and recalibrate valve position feedback.
4. Restore differential-pressure reset using the most-open-valve logic, with a minimum pressure limit for remote coils.
5. Add a BAS alarm when secondary flow exceeds the design-delta-T flow by more than 25\% for a sustained load band.
6. Trend supply temperature, return temperature, secondary flow, chiller staging, pump speed and valve positions for two weeks after correction.

The key engineering decision is to treat the symptom as a system fault, not as a chiller-capacity shortage. The acceptance evidence must show load, flow and return temperature together.

RPN Screen

Failure mode	Severity	Occurrence	Detection	RPN	Comment
Bypass valve leakage	7	5	6	210	High energy penalty and weak automatic detection before trending changes.
Coil valve leakage	6	6	5	180	Common in aging HVAC valves and visible only when flow is measured or inferred.
Fixed high differential pressure	5	6	4	120	Increases leakage and control instability, but can be caught by BAS review.
Chiller over-staging	7	4	5	140	Plant symptom that follows the distribution fault.

After repair and added alarms, the target is to reduce detection scores to 2 or 3 because the BAS should identify abnormal flow-per-load and low return-temperature events before they persist for a season.

Validation Criteria

The plant correction is accepted only if the same load band no longer requires excessive flow. During a post-repair test with cooling load between 1.4 and 1.6\ \text{MW}:

Criterion	Acceptance target
Building return temperature	at least 13.2^\circ\text{C} after stable operation
Secondary flow	45 to 55\ \text{L/s} for the load band
Chiller staging	one chiller online when load is below 1600\ \text{kW} and flow is below one-machine limit
Bypass leakage	less than 1\ \text{L/s} inferred or measured
Pump speed	reduced relative to the faulted case at comparable load
Heat-balance agreement	air-side and water-side load within stated uncertainty

The acceptance test also checks comfort and humidity. Raising return temperature by reducing wasteful flow is acceptable only if zones remain within temperature and latent-load requirements.

Engineering Lessons

Low delta-T syndrome is a mass-balance and control problem before it is an equipment-capacity problem. The decisive evidence is not a single low return-temperature trend. It is the combination of load, flow, supply temperature, return temperature, bypass evidence, valve position, pump pressure and chiller staging.

The second lesson is that design delta-T is not just a sizing number. It protects plant capacity, pump energy and chiller staging logic. When distribution faults collapse delta-T, the plant may operate as if it is short of capacity even while real cooling load is within the installed equipment rating.

The final lesson is to validate corrections under comparable load. A repaired valve tested during a low-load morning does not prove that the plant will avoid low delta-T during a warm afternoon. The useful proof is a stable trend package showing restored return temperature, lower flow, correct staging and acceptable zone conditions at the same cooling load band that previously caused the fault.