IESL Part III: Low Delta-T Syndrome in a Hotel Chiller Plant with Remote DP Control

Low Delta-T Syndrome in a Hotel Chiller Plant with Remote DP Control

A 320-room hotel tower has a chilled water plant of two nominal 500 TR centrifugal chillers with variable-speed drives, primary/secondary pumping and a common header, serving AHUs on 14 floors. Design conditions are 6.5/12.5 degrees C (delta-T 6 K) and 160 kPa differential at the most remote AHU, where the secondary loop differential-pressure sensor is installed. On a design-day afternoon the plant delivers 62% of design capacity, yet the secondary pump runs at 96% speed with the differential held at 160 kPa. Plant delta-T is 3.4 K and chilled water flow is 18% above design. The critical AHU coil valve is 100% open. Lower-floor AHUs are fitted with three-way valves; the rest have two-way valves with characterised ball valves. Chiller 2 cycles on and off every 12 minutes on its leaving-water setpoint. The building operator proposes raising the differential setpoint to 200 kPa to push more water to the top floors.

1. Explain the mechanism producing the 3.4 K plant delta-T. Separate the genuine causes from explanations that the data rule out.

Read the worked answer

Genuine causes, in order of size.

Three-way valve bypass. At part load a three-way valve returns supply-temperature water straight to the return header. The coils keep their design flow while the plant sees a blend, so plant delta-T collapses even though individual coil delta-T may be acceptable. This alone can account for a 6 K design falling to 3 to 4 K.
Remote DP sensor holding worst-case head. The pump is controlling to the pressure required by one remote coil. Every other branch then has excess head that must be destroyed across its control valve. With characterised ball valves of modest authority at part load, the valve characteristic is distorted and the coils receive more flow than they need, so coil delta-T falls too.
The critical coil itself. If that coil is fouled, air-bound or has a failed actuator, the sensor never sees the pressure it needs and the pump runs at 96%, which is exactly the symptom. A 100% open valve at the critical branch is consistent either with a genuine maximum-load condition or with a coil that cannot achieve its duty.
Chiller cycling. A chiller starting every 12 minutes produces leaving-water temperature swings, and valve actuators respond to the swing rather than to load, adding flow noise and further depressing plant delta-T.

Ruled out by the data. The plant is at 62% of capacity, so this is not a shortage of chiller capacity, and the DP is being held at setpoint, so it is not a shortage of pump head. The proposal to raise the setpoint is therefore addressing a problem that does not exist.

2. Evaluate the proposal to raise the differential setpoint to 200 kPa. What would you do instead, and in what order?

Read the worked answer

Assessment of the proposal. Raising the setpoint from 160 to 200 kPa increases flow through the paths of least resistance: three-way bypasses and the coils with the poorest valve authority. The water arriving at the remote coil increases only slightly because the system is already at 96% speed and near the flat part of the pump curve. The consequences are all negative: plant delta-T falls further, chiller loading drops further and short cycling worsens, valve throttling and noise increase, and pumping energy rises steeply because power follows speed roughly as the cube.

Correct sequence.

Survey every coil for air-side and water-side delta-T, valve position and valve authority. Fix fouled coils, air locks, failed actuators and strainers first; no control change will compensate for a coil with 30% of design duty.
Convert three-way valves to two-way on the lower floors. This is the single largest delta-T gain, and it costs little compared with any plant change.
Change from fixed DP to DP reset on valve position. Drive the setpoint so that the critical valve sits at 90 to 95% open, with a floor on the setpoint and a rate limit. This is the standard cure for remote-sensor DP control and it saves pump energy at part load.
Re-site the sensor to the hydraulically most remote coil only after that coil is proven healthy, or use a virtual sensor derived from flow measurement at the header.
Fix chiller staging with equal loading, an adequate deadband and a minimum-load constraint, so the second machine is not hunting on leaving-water temperature.

Measure against targets: plant delta-T at least 5.5 K at 50% load or more, no more than three chiller starts per hour, and pump speed below 80% at 60% load.

3. Write the verification plan: what you would baseline, which metrics prove the fix, and how you would attribute the savings.

Read the worked answer

Baseline (two to three weeks before any change).

15-minute interval logging of plant flow, supply and return temperatures, each chiller's load and power, secondary pump speed and power, and the DP at the sensor.
A one-off coil survey recording, per AHU, air on/off temperature, water in/out temperature, valve commanded and actual position, and design versus actual flow.
Valve authority check: measure the branch pressure drop with the valve open and shut; the control valve should take at least 50% of the branch drop.

Verification metrics.

Plant delta-T at defined load bands (30, 50, 75%), not a single spot reading; target 5.5 K or better.
Chilled water flow as a percentage of design at each load band; should fall below 100%.
Chiller starts per hour and minimum run time.
kW per TR of refrigeration at each load band, normalised for ambient wet bulb.
Pump kW at the same load bands.

Attribution. Use a measurement-and-verification approach such as IPMVP Option B (retrofit isolation with continuous measurement of the chiller and pump electrical power), with weather and occupancy normalisation. Keep independent variables (cooling degree days, occupancy) in a regression, and hold a persistence check at three and twelve months.

Watch for the rebound. If coil cleaning and two-way valve conversion are done but DP reset is not, the plant will drift back to high flow within a season. Log the critical valve position permanently and alarm when it sits outside 85 to 95% for more than an hour.