Chiller Compressor Distress: Surge, Lift and the Panel Readings That Warn Before a Trip

• Lift is condenser saturation temperature minus evaporator saturation temperature. Compressor power rises with lift, and surge risk peaks at low load combined with high lift.
• A surging chiller is often reporting a problem elsewhere: an underperforming cooling tower, low condenser water flow, scaled tubes or non condensable gases.
• Liquid floodback shows as knocking, frost on the suction line and foam in the oil sight glass. It usually traces back to low suction superheat.
• Restart logic matters as much as the compressor. A seconds long voltage sag can become hours of lost cooling if chillers do not restart before the chilled water loop warms.

A chiller compressor raises refrigerant vapor from evaporator pressure to condenser pressure, so heat taken from the chilled water can be rejected to the condenser water. Lift is the size of that climb.

Lift is the condenser saturation temperature minus the evaporator saturation temperature. It is the pressure difference the compressor must overcome, expressed as temperature.

Compressor power rises with lift. Every degree of lift added by a fouled tube or a warm tower is paid for in motor current. It also adds stress on the compressor itself (ACHR News).

Lift is not set by the chiller alone. The cooling tower, condenser water flow and water treatment all shape the condenser side, which is why compressor distress so often starts outside the chiller.

Surge is an unstable condition in a centrifugal compressor where refrigerant flow repeatedly reverses through the impeller. It is a cycle, not a single event.

At light load and high lift, the impeller cannot build enough pressure to push refrigerant into the condenser. Flow momentarily reverses back through the impeller, recovers, then reverses again.

Each reversal loads and unloads the impeller, shaft and bearings. A compressor can repeat that cycle many times before a protection trips it.

Built in defenses

Inlet guide vanes, variable frequency drive (VFD) speed control and hot gas bypass are the common built in defenses against surge. Each keeps the compressor away from the low flow, high lift corner where surge starts.

Those defenses work within limits. When lift climbs high enough, no amount of vane or speed control keeps the compressor stable.

Hot gas bypass also has a cost. It keeps the compressor stable by adding false load, so the chiller spends energy recirculating refrigerant instead of cooling water. A chiller that leans on it every night is worth a closer look at its lift.

Surge is caused by high lift at low load. Anything that raises condenser pressure, or lowers load while lift stays high, pushes a centrifugal compressor toward it.

The main causes of high lift are:

• High condenser water temperature from an underperforming cooling tower.
• Low condenser water flow, from a throttled valve, a weak pump or a blocked strainer.
• Fouled or scaled condenser tubes, which insulate the refrigerant from the water.
• Non condensable gases in the refrigerant circuit, which raise condenser pressure.

Surge risk is greatest at low load combined with high lift. Light night loads during a hot spell, or a single chiller carrying a small load against a warm tower, are classic conditions.

Surge is often a cooling tower or water treatment problem because most of the causes of high lift sit on the condenser water side. The chiller is where the problem becomes audible, not where it starts.

A tower that cannot reach its design leaving water temperature raises condenser pressure. Scale left behind by poor water treatment does the same from inside the tubes. Low condenser flow does it by carrying heat away too slowly.

That makes a surge alarm a plant level finding. Resetting the chiller treats the symptom, while the tower, the flow or the treatment program keeps causing it.

The Cooling Plant Performance sub-series covers the tower and water side in depth. Here the point is simpler: read a surging chiller as a report about its condenser loop. Loads like these often fall between teams, a pattern covered in cooling loads nobody owns.

Surge sounds like a loud, rhythmic whoosh or thump that repeats as flow reverses and recovers. It is cyclic, which separates it from most other compressor noises.

Two other signs travel with the sound. Motor current swings as the load on the impeller rises and falls, and discharge pressure oscillates. Both show on the chiller display.

Acoustic listening picks up the rhythm, and vibration analysis shows the shaft and bearings taking each cycle. Neither is needed to hear a hard surge; both help catch a mild one.

What repeated surge does

Repeated surge damages bearings and the impeller. It should never just be reset and ignored, because each event adds to damage that does not show until the compressor fails.

Oil free magnetic bearing compressors carry a specific risk. A hard surge can overload the magnetic bearings and drop the shaft onto its backup touchdown bearings.

The surge history

Many centrifugal chiller controllers log surge events. The surge count or surge history on the panel is worth checking on every round. It shows surges that happened when nobody was in the room.

Liquid floodback, also called slugging, is liquid refrigerant returning to a positive displacement compressor. Screw, scroll and reciprocating compressors are built to compress gas, and liquid does not compress.

The signs are mostly visible or audible:

• Knocking from the compressor as liquid reaches the compression parts.
• Frost on the suction line or the compressor body.
• Foaming in the oil sight glass.
• Oil dilution, as refrigerant mixes into the oil and thins it.

Floodback usually traces back to low suction superheat. Superheat is how far the suction gas sits above its boiling point, and too little means liquid is reaching the compressor.

Diluted oil lubricates poorly, so floodback damages bearings as well as valves. A frosted suction line is worth recording every time, even when the compressor sounds normal.

Reverse rotation in scroll compressors

Three phase scroll compressors run backward if two supply phases are swapped. The result is loud noise, no pumping and rapid damage, often right after electrical work.

High discharge temperature matters because it breaks down compressor oil and stresses valves and bearings. Oil that has cooked loses its ability to protect the parts it is meant to separate.

Manufacturers publish discharge temperature limits for their compressors. Use the original equipment manufacturer (OEM) limit for the machine in front of you, not a generic number.

High lift and low suction pressure both push discharge temperature up, because the compressor squeezes the gas through a bigger pressure rise. Thermal imaging can show a discharge line running hotter than its peers, which is a reason to check the panel reading against the OEM limit.

Motor heat is a separate question. A hermetic or semi hermetic compressor motor shows overheating as a motor temperature alarm, covered in Part 6.

The chiller panel shows compressor health through temperatures, pressures, current and alarms read together. No single number tells the story, but the set does.

These are the readings worth recording each round:

• Evaporator and condenser saturation temperatures or pressures, which give lift.
• Leaving chilled water and condenser water temperatures.
• Evaporator and condenser approach temperatures.
• Oil pressure differential.
• Motor current as a percentage of rated load amps (RLA).
• Inlet guide vane position.
• Active and historical alarms, including surge history where recorded.

What approach temperatures tell you

An approach is the gap between a leaving water temperature and the refrigerant saturation temperature in the same heat exchanger. It shows how well heat is crossing the tubes.

A condenser approach that creeps up over weeks points to fouled or scaled tubes, or to non condensable gases. An evaporator approach that rises points to fouling or a refrigerant problem on the chilled water side. Both raise lift before anything alarms.

Oil pressure differential belongs on the same list. A falling differential is an early sign of oil trouble, including the dilution that floodback causes.

How to use them

Compare each chiller with its peers and with its own history. A chiller whose lift or approaches sit above design, or above an identical machine on the same plant, is telling you something before it trips.

A photograph of the panel on every round turns those readings into a trend. That is the habit described in reading the gauges nobody monitors, applied to the most expensive machine in the room.

Chillers fail to restart after a power dip when their restart logic, interlocks or supporting pumps do not bring them back in sequence. Once the chilled water loop warms too far, some chillers cannot restart at all.

What happened in Azure Australia East

On August 30, 2023, a lightning strike about 18 miles (29 km) from the affected availability zone caused the voltage sag (DatacenterDynamics). The two affected data halls had seven chillers, five running and two on standby.

All five running chillers faulted, and only one standby started. Microsoft said the five did not restart because their pumps did not get the run signal from the chillers (DatacenterDynamics).

Microsoft also said night staffing was insufficient to restart the chillers in time. Some customers saw impact for more than 24 hours.

The lesson for any plant

The power event lasted seconds. Restart logic, pump signals and chilled water temperature turned it into hours of lost cooling.

Every chiller's automatic restart should be tested, not assumed. Every site should also know how long a manual restart of every chiller takes with the people actually on shift.

Computer room air conditioner (CRAC) compressors fail mostly from high head pressure and short cycling. CRACs are direct expansion units with their own compressors, so each one is a small refrigeration plant.

Dirty condenser coils raise head pressure, which raises lift and discharge temperature. The compressor works harder for the same cooling.

Short cycling, frequent starts and stops, wears contactors and compressors. Each start is a stress event, and a unit that cycles constantly spends its life quickly.

Air cooled coils are easy to check on a walk past. Dirt, debris and blocked airflow are visible, and they belong on the preventive maintenance schedule.

In both, a damaged compressor means lost cooling capacity and an expensive repair. The difference is how quickly lost capacity becomes a crisis.

In an office building

Compressor damage means long lead time repairs and lost cooling capacity in peak season, when the plant can least spare it. Routine heating, ventilation and air conditioning (HVAC) maintenance rarely tracks surge history or approach temperatures.

In a data center

A compressor fault or lockout removes chiller capacity. As Azure showed, restart logic and chilled water temperature can turn a seconds long power event into hours of lost cooling.

That makes restart testing and night staffing part of compressor health. A chiller that will not restart is as unavailable as one that has failed.

Several faults sound or look like compressor distress. Their pattern over time usually separates them.

• Surge versus a failing bearing: surge is cyclic and tied to load and lift; a bearing fault is steady. Bearings are covered in Part 1.
• Floodback knocking versus a loose mounting rattle: a rattle follows vibration and the base, not suction conditions. Mounting faults are covered in Part 5.
• Motor overheating versus compressor overheating: a motor temperature alarm points at the motor and its supply, a high discharge temperature at the refrigeration cycle. Motors are covered in Part 6.

This is why no single sensor is enough. Sound, current, panel trends and a look at the suction line each rule out a different suspect.

A walk past can prove that a chiller has alarmed or surged, that a suction line is frosting and that a condenser coil is dirty. It cannot prove refrigerant charge or internal compressor wear.

• Directly observable: alarms and surge history on the panel, frost on suction lines, oil sight glass foam and dirty condenser coils.
• Strong indicator: rhythmic surge noise, current swings on the display, and lift and approach temperatures above design.
• Suspected: refrigerant charge, non condensables and internal compressor wear. These need refrigerant side testing by a qualified technician.

That is the limit of any route based check. It finds the surging chiller and the frosted line; refrigerant side testing confirms the cause.

A camera records the panel and the sight glass, and a walkthrough record keeps the history. Trending lift and approaches over time is the core of predictive maintenance.

• Read and photograph the chiller panel: temperatures, approaches, current, guide vane position and alarms.
• Surge count or surge history, where the controller records it.
• Listen for a cyclic whoosh or knock.
• Frost on suction lines and compressor bodies.
• Oil sight glass level, and any foaming.
• Air cooled condenser coils free of dirt and blockage.
• Lift and approaches compared with each chiller's peers.

Coming Soon

1. 1. Does each chiller restart automatically after a voltage sag, and has that been tested?
2. 2. How many surge events has each centrifugal chiller logged this year?
3. 3. How do your condenser approach temperatures compare with design?
4. 4. How many people are on site at night, and how long would a manual restart of every chiller take?
5. 5. At what chilled water temperature can your chillers no longer restart?

What is chiller surge?

Surge is an unstable condition in a centrifugal compressor. At light load and high lift, the impeller cannot push refrigerant into the condenser, so flow reverses, recovers and reverses again in a cycle.

What causes a centrifugal chiller to surge?

Surge is caused by high lift at low load. The main causes of high lift are high condenser water temperature from an underperforming tower, low condenser water flow, fouled or scaled tubes and non condensable gases.

Does surge damage a chiller?

Yes. Repeated surge damages bearings and the impeller, so it should never just be reset and ignored. On oil free magnetic bearing compressors, a hard surge can drop the shaft onto its backup touchdown bearings.

What is chiller lift?

Lift is the condenser saturation temperature minus the evaporator saturation temperature. It is the pressure difference the compressor must overcome, and compressor power rises with it.

What is liquid floodback?

Liquid floodback, or slugging, is liquid refrigerant returning to a screw, scroll or reciprocating compressor. Signs include knocking, frost on the suction line or compressor body, foaming in the oil sight glass and oil dilution. It usually traces back to low suction superheat.

Why does a compressor knock?

Knocking in a positive displacement compressor usually means liquid floodback, with liquid refrigerant reaching the compression parts. A loose mounting can also rattle, but that follows vibration rather than suction conditions.

Why do chillers fail to restart after a power dip?

Restart logic, interlocks or supporting pumps may not bring them back in sequence. In the 2023 Azure Australia East outage, pumps did not get the run signal from the chillers. Once the chilled water loop warmed too far, the last chillers could not restart.

Which chiller readings show compressor health?

Record saturation temperatures or pressures, leaving chilled and condenser water temperatures and approach temperatures. Add oil pressure differential, motor current as a percentage of rated load amps, guide vane position and alarms. Compare each chiller with its peers and its own history.

Can a dirty condenser cause chiller surge?

Yes. Fouled or scaled condenser tubes raise condenser pressure and lift, which pushes a centrifugal compressor toward surge at low load. A surging chiller is often reporting a condenser side problem.

• Motor Overheating: The 10 Degree Rule, Voltage Unbalance and the Cooling Air Nobody Checks
• Cooling and Mechanical: ASHRAE 180 and the Loads Nobody Owns
• Reading the Gauges Nobody Monitors: What Your Facility Already Knows and Never Writes Down
• Acoustic Monitoring in Data Centers: Hearing Equipment Change Before It Fails

• DatacenterDynamics: staffing levels and the Azure Australia East chiller restart failure
• DatacenterDynamics: Microsoft's slow outage recovery in Sydney
• ACHR News: optimizing centrifugal chiller systems