Compressor Intake Valves and Minimum Pressure Valves Explained

Kaeser hard-anodizes the piston bore on the CSD intake valve actuator. Atlas Copco does not do this on the GA. After six years of three-shift cycling you can feel the scoring in the GA bore with a fingernail. The CSD bore is still smooth.

The butterfly disc opens to about 82 degrees on a GA, roughly 80 on a Sullair LS. Not 90. At 90 the disc flutters from vortex shedding, a few hundred hertz, sounds like bearing noise. An Atlas Copco distributor in the Midwestern US replaced bearings on three GA machines in one year before someone traced the noise to aftermarket intake valves with the disc stop machined at 87 degrees instead of 82. Looked right during install. Five degrees off was enough.

No GA under 90 kW has an intake valve position sensor in standard configuration. The Elektronikon reads about 30 millibar of pressure drop across the open valve and infers valve state from that. If carbon on the disc lip holds it three degrees short of its stop, the pressure shift is inside the controller's noise floor. The Elektronikon reports normal. The compressor delivers less air than rated. Catching this requires a discharge flow meter, which does not get installed on compressors in this size class.

The carbon buildup happens on GA machines because of the inlet ducting geometry. Humid air pools on the disc overnight, carries particulate from the filter face onto the seating surface, builds a varnish over months. Sullair LS machines have a worse version because the inlet routes near the oil cooler exhaust. The CSD has less of it. Kaeser did not design against it. The package layout just happens to avoid the worst of the condensation pooling.

Closing Speed

This is the subject that deserves the most attention in any discussion of intake valves, and it is the subject that gets the least. The entire rest of the compressor's thermal behavior during transient operation is downstream of how fast this valve closes.

The actuator gets control air from the compressor's own discharge through a solenoid. At startup from zero pressure, no control air exists. The valve stays open. The machine loads fully until enough pressure builds to actuate the piston, around 1.5 bar on a GA. The CSD needs more because of the heavier return spring, takes a beat longer to gain authority.

When the disc sweeps shut on unload the airend keeps spinning at full speed compressing a shrinking mass of air. Temperature spikes. About 8°C on a clean GA55 at 7.5 bar, under a second. A worn GA actuator with a scored bore, 15 to 20°C for three or four seconds.

Two hundred thousand unload events per year at 25 cycles per hour. The thermal pulse from each one hits the separator element, the oil film on the tank walls, the MPV O-ring. The separator element is a coalescing filter made of borosilicate microfiber bonded with a resin system. The resin has a thermal degradation curve. It is not a cliff, it is a slope, and cumulative thermal dose pushes the element down that slope at a rate determined partly by steady-state discharge temperature and partly by these transient spikes that come with every unload. The spikes are short. There are 200,000 of them per year. The product of a short duration and a very large number of repetitions is not small.

Atlas Copco quotes 4,000 or 8,000 hours for a GA separator element depending on grade. At 16,000 operating hours the actuator has been cycling for four years without piston travel being checked, and the valve closes a second and a half slower than at commissioning. The element that should have lasted 8,000 hours goes at 5,500. Gets written up as defective element or bad oil batch.

Two identical GA75s side by side in the same compressor room, same air, same oil, same filters. One hits 8,000 hours on every separator element. The other goes through elements at 5,500. The maintenance team adjusts the replacement interval downward on the second unit. "Some machines just run hotter."

The temperature difference between the two machines at steady state is maybe 2°C, well within normal tolerance. The difference during the unload transient is 12°C, and it lasts three times as long on the second machine, and it happens 200,000 times a year, and the maintenance system has no way to see it because measuring it requires a fast-response thermocouple at the airend discharge and a data logger. Ten-minute setup. Compressor service vans do not carry this equipment. The measurement is simple. The reason it does not happen is that compressor service is organized around parts replacement intervals, not around diagnostic measurement of transient thermal behavior. The intake valve actuator condition and the separator element life are in different sections of the parts manual and different line items in the service contract. The thermal link between them does not appear in either.

This is worth dwelling on because it is a concrete example of how a cheap, accessible measurement could save thousands of euros per year per machine in separator element cost alone, and the measurement is not done. Not because it is difficult. Not because the equipment is exotic. Because the maintenance structure is not set up to ask the question. The separator element is a consumable. It gets replaced on schedule. If it fails early, the schedule gets shortened. The root cause investigation stops at "this machine eats elements" and does not proceed to "the intake valve actuator is closing 1.3 seconds slower than specification and the cumulative thermal dose from 200,000 degraded transients per year is equivalent to running the machine 15°C hotter than its steady-state temperature indicates."

A maintenance engineer who reads this and has two machines with different element life in the same room should borrow a thermocouple data logger from the plant instrumentation shop, put the sensor at the airend discharge, run both machines through ten unload cycles each, and compare the traces. If one machine shows a broader, higher temperature peak on unload, the actuator on that machine is slow. Rebuilding the actuator, which is a piston seal kit on a GA, half an hour of labor, will bring the thermal transient back in line and the separator element life will follow.

The Ingersoll Rand Diaphragm Approach

Certain R-series models use a diaphragm actuator. No bore to score. Diaphragms tear if oil-laden blowdown air gets into the housing through a leaking check valve. A torn diaphragm kills the valve completely and immediately. A scored piston bore kills it gradually over years while the controller reports normal. The diaphragm design is the better choice for facilities where the maintenance culture is reactive, because the failure is binary and obvious. The piston design is better where maintenance is predictive, because the gradual degradation provides warning. Ingersoll Rand does not frame the choice this way.

Modulation

At half capacity, compression ratio from 8:1 to above 11:1. Shell Corena S4 R46 is formulated for about 100°C, at 120°C it oxidizes four times faster. Kaeser forces unload below 48%, Atlas Copco in the low 40s. VSD keeps compression ratio flat by reducing rotor speed. Modulating control should not be specified where demand drops below 60%. It continues to sell because the purchase price is lower and the person signing the purchase order is not the person paying the electricity bill.

Blowdown

After the intake valve closes, trapped pressure vents through a fixed orifice back to the airend inlet. Oil-laden air goes backward through the airend, some reaches the inlet filter, oily film on the clean side. Normal.

Atlas Copco has a service bulletin on some GA models against drilling out this orifice. The bulletin exists because technicians were doing it after seeing elevated unloaded power. The elevated power was caused by something else. The enlarged orifice floods the inlet ducting with oil on every unload and fixes nothing.

Minimum Pressure Valve

Oil moves on pressure differential from separator tank to airend injection point. No pump. The MPV holds the tank closed until about 4 bar gauge on a GA.

The spring targets cold startup with high-viscosity oil. Mobil Rarus SHC 1026 at 5°C is viscous enough that the 4 bar threshold is needed to push it through the cooler core. During warm running the threshold is conservative by a wide margin.

At 1,500 meters elevation, atmospheric pressure is 0.84 bar, absolute pressure at the cracking point drops, and oil circuit driving force at startup is less than the designer intended. Compressors get relocated from coastal plants to mine sites and the MPV spring stays where the factory set it. Cold start at elevation with high-viscosity oil: seized airend.

The O-ring sits at the hottest point in the separator tank. Under loaded operation at 7.5 bar and 25°C ambient, the O-ring environment is 85 to 100°C. NBR is rated around 100°C continuous. The GA and CSD service kits ship with NBR. FKM handles 200°C and costs two or three euros more per ring. The service kits should contain FKM. They do not. Field technicians who know this order FKM separately, install it without recording the change, and the next technician who services the machine finds NBR in the parts system and may put NBR back in.

The poppet reciprocates in a bore. Millions of cycles develop clearance. Angled seating, uneven O-ring loading. New O-ring in a worn bore lasts three to six months. On a GA the MPV is a cartridge, worn bore means new cartridge. On CompAir and Sullair models where the valve integrates into the separator tank head casting, worn bore means new casting. Service technicians on those machines swap O-rings every six months because the casting cost is prohibitive. Each swap seals worse than the last.

Bore wear correlates with cycle count, not hours. The Elektronikon and Sigma Control both log load/unload events. The cycle count data exists in the controller. OEM maintenance schedules index MPV inspection to hours.

Backflow

Leaking MPV seat lets downstream air bleed back into the separator tank during unload. System pressure drops. Compressor reloads early. Energy goes up. Looks like a piping leak.

Ball valve downstream of MPV, closed during unload. Five minutes. Pressure holds: MPV leaking. Pressure drops: piping. Atlas Copco's GA service documentation includes a version of this test. Most technicians go straight to a leak survey.

Both valves can leak simultaneously. Minor intake leak feeds air into the airend during unload while minor MPV seat leak bleeds system pressure backward. Slightly elevated tank pressure, slightly elevated power, slightly increased cycle frequency. No alarm. The dual condition persists until someone runs the ball valve test and interprets the result with both valves in mind.

Maintenance

MPV at every separator element change. Tank is already open. Pull the cartridge. FKM O-ring. Spring free length against the service manual. Poppet lateral play in the bore. Seat surface under good light. Thirty to forty-five minutes.

The intake valve air filter on a GA75 is under 30 euros, the assembly is above 400. Replace the filter at or before 25 millibar differential. Annual actuator check: pressurize the control air line with the machine locked out, watch the piston stroke.

On machines with copper control air tubing inside the canopy, blow through the tubing. Condensate and scale build up inside over years. The valve starts closing late. A technician will check the solenoid, check the actuator seals, check the wiring. Two hours. Eventually someone suggests blowing through the tube and a slug of dirty water comes out. Copper tubing does not appear on PM task lists. It is not a wear part. It does not have a replacement interval. It gradually obstructs.