Centrifugal Chiller Surging at Ninety Percent Load
A 250-ton water-cooled centrifugal chiller using R-134a serves a data hall. Design at full load: condenser water 85°F in and 95°F out with a 3°F condenser approach; evaporator chilled water 54°F in and 44°F out with 40°F saturated suction. Today at 90% load the readings are: condenser water 85°F in and 95°F out, saturated condensing temperature 112°F, evaporator saturated suction 42°F, chilled water leaving 46°F. Compressor amps are 96% of rated, the motor is at full load, and the machine surges about every five minutes. Tower fans are at 100% and tower leaving water is 85°F as designed. The purge unit runs continuously with a slight but steady purge count. Two months ago the log shows a condenser approach of 4°F. Refrigerant charge is confirmed correct by the liquid level in the condenser, and no refrigerant has been added. Condenser water pump differential pressure and flow are at design.
1. Calculate the approaches and interpret them. Which side of the machine has the problem, and how do you know?
Read the worked answer
Condenser approach is saturated condensing temperature minus leaving condenser water: 112 − 95 = 17°F against a design of 3°F. That is 14°F of excess approach, which is enormous.
Evaporator approach is leaving chilled water minus saturated suction: 46 − 42 = 4°F, which matches the design value of 44 − 40 = 4°F. The evaporator is behaving normally.
Condenser water flow is confirmed by the range: 95 − 85 = 10°F, exactly the design 10°F at this load. If flow were low, the range would be wider and the pump differential would not be at design. So the problem is not water quantity.
Charge is ruled out two ways: liquid level is correct, and a low charge shows up as a low evaporator approach and poor subcooling, not as a condenser approach that has grown while the evaporator stays clean. The machine is telling you that heat is not moving from refrigerant to water at the rate it should in the condenser bundle. The candidates are fouled or scaled tubes, non-condensables in the shell, or a mechanical obstruction such as a failed tube or a bundle bypass. Because the fault pushes head pressure up, the compressor runs into its surge line at high load - the surge is a symptom, not the disease.
2. How would you distinguish tube fouling from non-condensables, and how much weight should the purge count carry?
Read the worked answer
Non-condensables raise total pressure in the condenser for a given temperature, so the direct test is a pressure–temperature check. Record the shell pressure and the temperature of the saturated refrigerant in the condenser at the same moment, then compare the measured pressure with the saturation pressure for R-134a at that temperature. If the measured pressure is meaningfully higher, there is gas in the shell that should not be there. Repeat at steady conditions, then purge and repeat again after the machine has run a while; a genuine air leak will rebuild the offset, while a once-off offset will clear.
The purge count is supporting evidence at best. A steady, slight count over weeks could be a small leak and a lot of air, or it could be a purge unit pulling a little refrigerant vapor. It does not quantify non-condensables, and it must not be used to skip the pressure–temperature check. Treat a continuous purge as a reason to leak-test the machine, not as a diagnosis.
If purging does not bring the approach down, the water side is the remaining explanation. Pull the condenser heads, inspect tube sheets and tube ends for scale, mud, and biofilm, and brush or rod a sample of tubes. Check the water treatment logs and the last eddy-current inspection, and test for refrigerant in the condenser water in case the fouling is actually a pinhole leak pulling in water.
3. What do you do tonight with the machine surging every five minutes, and what happens if the condition is left alone?
Read the worked answer
Surging is the compressor's flow reversing momentarily because it cannot make head against the condensing pressure. Every surge event loads the thrust bearings, flexes the impeller, and puts a torque transient through the motor and coupling. Repeated surging at five-minute intervals is actively damaging the machine, not just inefficient.
Tonight: reduce load until the machine is comfortably inside its surge line - the operator's lever is chilled water demand, so stage other equipment or shed load. Lower entering condenser water temperature if there is any headroom, since colder tower water directly reduces head and buys surge margin; 85°F is design, not a limit. Log amps, approach, and surge frequency after each change so you can show what worked. Do not add refrigerant and do not raise chilled water setpoint to disguise the problem; both mask the symptom while the condenser stays fouled.
Left alone, the likely outcomes are thrust bearing failure, impeller or seal damage, and eventual motor overload trips with a data hall at risk of losing cooling. The economical path is to arrange condenser cleaning and water treatment correction on a planned basis now, while the machine is still capable of carrying 90% load at reduced efficiency, rather than after a bearing failure at 2 a.m.