Part-load chiller trips and a fresh-air cutback in a Grade A office tower
A 28-storey Grade A office tower in Quarry Bay is served by two 500 RT water-cooled centrifugal chillers with variable-speed drives, designed for 6.5 °C/13 °C chilled water with variable primary flow. The lead chiller runs alone until it reaches 85% of design current before the second chiller is enabled. During the first summer the lead chiller tripped repeatedly on low evaporator pressure between 14:00 and 16:00, when building load was 35–40% and only the lead chiller and one primary pump were running with the pump at 40% speed. Tenants on the west façade complained of warm air. Separately, the O&M team, under pressure to cut energy, has closed the fresh-air dampers to roughly 10% of design; design fresh-air rate is 1,200 L/s for 120 occupants; the BMS now logs 1,450 ppm CO2 in occupied zones at mid-morning and staff report drowsiness.
1. Diagnose the repeated low-evaporator-pressure trips and set out how you would confirm the cause before changing anything.
Read the worked answer
Most likely cause: operation below the chiller's minimum evaporator flow.
At 35–40% load with one chiller and the primary pump at 40% speed, evaporator flow is about 40% of design - likely below the minimum, commonly 45–60% for a centrifugal. With load still present, evaporator ΔT rises and the refrigerant must run colder to transfer the heat; evaporator pressure falls until the freeze-protection or low-pressure cut-out trips. The pump had been slowed to hold a DP setpoint that was satisfied because the AHU valves throttled at part load. The control loop was doing its job; nobody had told it about the chiller's minimum flow.
Confirm before acting. Log evaporator flow (or pump speed plus the pump curve), entering and leaving chilled-water temperatures, saturated suction and discharge pressures, approach temperatures, compressor current as %FLA and VSD speed. Verify the flow switch is proving flow and has not been bypassed. Rule out refrigerant loss (subcooling and superheat), evaporator fouling (approach), and faulty sensors - but the correlation with pump speed points strongly to flow.
Fix. Maintain minimum evaporator flow with a bypass valve or by holding the primary pump at minimum speed; stage the second chiller or reset the DP setpoint so the loop cannot starve the running machine. Recommission the chiller sequence, review the 85% staging point, and add a low-flow alarm that is trended.
2. Assess the fresh-air cutback using a CO2 mass balance, and quantify whether the energy saving justifies it.
Read the worked answer
Mass balance. Steady state above outdoor: C = C₀ + G/Q. Take 120 occupants × about 0.005 L/s CO2 each = 0.6 L/s, outdoor 420 ppm, target 1,000 ppm:
Q = 0.6 / (1,000 − 420) × 10⁶ ≈ 1,030 L/s.
The 1,200 L/s design is about 10 L/s per person - adequate with margin. At 10%, Q ≈ 120 L/s, which predicts about 5,400 ppm, not the 1,450 ppm logged. So either the dampers are not actually at 10%, the CO2 sensors or their locations are wrong, or mixing and occupancy differ from the assumption. Investigate before accepting either number - the inconsistency is itself a finding.
Judgement. Cutting fresh air to 10% is not an acceptable energy measure. It breaches the design basis and the indoor air quality objective, exposes the owner to tenant complaints and possible lease or certification consequences, and may compromise any zone where the same air handling supports pressurisation or smoke-control modes.
Quantify the saving. 1,200 L/s ≈ 1.44 kg/s; at roughly 10 kJ/kg enthalpy difference the fresh-air load is only about 14 kW against a 500 RT (≈1,760 kW) chiller. The saving is trivial compared with the risk. If energy reduction is wanted, use CO2-based demand-controlled ventilation with a hard minimum, occupancy sensing and proper economiser control.
3. Recommend the plant sequencing and control changes, and state how any energy saving should be demonstrated.
Read the worked answer
Sequence and controls
- Establish and enforce chiller minimum evaporator flow, and interlock pump speed with chiller loading.
- Stage chillers on load (current or kW) with a defined add/subtract deadband and time delay; do not let one machine run down to 35% and wait.
- Reset condenser-water temperature within the chiller manufacturer's envelope to improve part-load efficiency, but not so low that the VSD machine surges or the tower fans hunt.
- Reset chilled-water supply temperature at part load, subject to the most demanding terminal - the west façade solar gain.
- Keep fresh-air dampers under BMS control with a minimum-position limit; trend CO2 and damper position.
- Prove savings by measurement and verification against an agreed baseline, not by closing dampers.
Professional point. The O&M instruction to close the dampers, and any pressure to delay fixing it, should be recorded in writing. You cannot certify comfort or indoor air quality that the system as operated cannot deliver, and the engineer who signs the handover or the energy report carries that statement.