Manufacturing Process of Screw Air Compressors – Screw Air Compressors Manufacturer

Hanbell in Taichung ships airends to compressor brands across four continents. So does Fusheng. So does Mayekawa's TMC division. Atlas Copco's Airpower operation grinds its own rotors. Kaeser grinds in Coburg. Kobelco grinds in Japan. SRM in Stockholm originated the asymmetric profile mathematics that the whole industry still runs on, licensed it globally for decades, and watched competitors develop proprietary profiles after the key patents expired. Kaeser calls theirs Sigma. Atlas Copco does not publicize the name of theirs. The number of companies that both own a proprietary profile and grind it at production volume on their own equipment is somewhere below twenty.

The number of companies that build the CNC grinding machines those twenty companies use is below five. Holroyd in Rochdale builds most of them. Kapp Niles in Coburg builds the rest. A Holroyd GTG2 costs over a million euros, arrives eighteen months after you order it, needs a room at ±1°C, and sits on a concrete mass foundation that is physically disconnected from the building structure so a truck idling outside the wall does not put a vibration ripple into a rotor profile. Finding operators takes longer than finding the machine. Training one takes years. There is no school for rotor grinding. People learn it by grinding rotors under supervision until their scrap rate drops low enough to be trusted alone.

That paragraph describes why the OEM airend supply model exists, and it is not going away. A compressor packager who wants to grind rotors is committing to a capital project that will not produce saleable parts for two to three years.

01 Rotor Forgings

AISI 4140 or 4340 billets, forged at around 1,200°C. The grain flow advantage of forging over casting for fatigue life is well covered in every materials science textbook and does not need another explanation here.

Die wear does need one, because it creates a problem that compressor manufacturers sometimes do not realize they have until a rotor cracks in the field. Forging dies degrade continuously. By a few thousand impressions, material folds at corners, creating laps. A lap is a surface fold that did not weld shut during deformation. It sits near the surface of the blank, invisible to external inspection, and opens under cyclic load.

Inspection

ASTM E381 Section 9 describes the macro-etch procedure: transverse cross-section, polished, immersed in 50% HCl at 71°C, examined for grain flow continuity. Takes under an hour. The forge shop selling blanks has an incentive to maximize die life. Specifying maximum die life contractually and requiring macro-etch samples per lot is how Atlas Copco and Kaeser manage this. Whether every buyer of every rotor forging worldwide exercises that discipline is a question that answers itself.

Sub-30 kW rotors sometimes use ductile iron per ASTM A536 grade 80-55-06. At those speeds and loads, the bearings will need attention long before the rotor, regardless of material.

02 Grinding

This is where the compressor gets made.

A rough-milled rotor carries profile errors of 30 to 50 microns. The milling cutter's geometry is not the rotor profile. It is a conjugate shape derived through envelope mathematics, and cutter design errors show up on the rotor surface in geometrically non-obvious locations. After milling, the rotor goes to the grinder.

Finish grinding on a CBN wheel brings profile accuracy within about ±6 microns on a well-run line. The gap between 50 microns and 6 microns is the gap between a shape that vaguely resembles a rotor and a surface that seals well enough to make rated specific energy at rated pressure for a decade.

The grinding program is empirical. A process engineer develops it by grinding test rotors, measuring them on CMM equipment from Mahr or Zeiss, adjusting, grinding again. On a 200 mm rotor in 4140 at 58 HRC, a typical finish pass takes about 8 microns per flank. The workpiece turns at roughly 20 RPM. The wheel runs at about 50 m/s. These are not universal constants. They are what one factory settled on for one rotor model after weeks of iteration, and they are not published, not shared at trade shows, and not transferable to a different rotor diameter without starting the iteration over, because the contact arc length between wheel and workpiece changes with diameter and everything downstream of that changes with it: thermal load distribution, coolant film behavior, elastic deflection of the workpiece under wheel pressure.

The CBN wheel gets dressed to its conjugate profile by a CNC diamond roll dresser. The dresser diamond wears at a rate that depends on the wheel bond, dressing depth per pass, and traverse speed. The interaction among those three variables is nonlinear, meaning the process engineer cannot reliably predict dresser life from first principles. Instead, rotor profile deviation data from the CMM gets plotted against the number of rotors ground since the last dresser diamond change, and when the trend approaches the action limit, the diamond gets replaced. Replacing it is a calibration event. The new diamond sits at a slightly different position. The first few rotors after a change get extra inspection.

The wheel itself shrinks as dressing and grinding consume it. A fresh wheel at 400 mm diameter produces a different contact arc at the same programmed depth of cut than the same wheel at 380 mm after weeks of use. The thermal behavior of the grind changes. Some factories adjust parameters as wheel diameter decreases. Others accept wider variation in rotor quality toward the end of a wheel's life and rely on inspection to catch anything that drifts out. The second approach is cheaper in process engineering time and more expensive in scrap.

Spark-out passes close the grinding cycle. The wheel traverses the rotor with zero programmed infeed, cleaning up elastic springback from the cutting passes. The number of spark-out passes was determined by testing on a specific machine. Cutting one pass to save ninety seconds degrades profile accuracy by an amount that may or may not matter depending on how much tolerance margin remains. Factories that are comfortable with thin margins cut passes. Factories that have been burned by warranty claims keep all of them.

A few things about rotor grinding that sit in the category of knowledge accumulated from years of process development and that do not appear in any machine tool brochure or conference paper:

Fixture stiffness dominates profile accuracy at the ends of the rotor more than in the middle. A rotor with a length-to-diameter ratio above about 4.5 will deflect under wheel pressure at the unsupported end even with a tailstock, and the deflection shows up as a profile deviation that is worst at the last 20 mm of the helix. Some factories accept this and open the tolerance at the rotor ends. Others invest in steady rests or modified fixturing to control it. The profile at the rotor ends matters because that is where the discharge port is, and leakage at the discharge end costs more thermodynamically than leakage at the suction end, because the pressure difference is highest there.

Coolant temperature stability matters more than coolant volume. A coolant system that delivers 200 liters per minute at a temperature that drifts ±3°C through the day produces less consistent rotors than a system delivering 120 liters per minute at ±0.5°C. The thermal stability of the coolant is part of the dimensional stability of the workpiece during grinding. Factories in climates with large diurnal temperature swings have to invest more in coolant chilling and recirculation systems than factories in thermally stable environments. This is a mundane infrastructure detail that has measurable impact on rotor quality and that does not appear in any discussion of grinding technology.

The time between grinding and inspection matters. A rotor that sits on the shop floor for four hours after grinding equilibrates thermally before measurement. A rotor that goes straight from the grinder to the CMM is still warm from grinding and will measure differently than it will at ambient temperature. The dimensional difference is small, a few microns, and a few microns is the entire tolerance band. Some factories enforce a minimum cool-down period before inspection. Others apply thermal compensation factors in the CMM software. Both work. Neither is better. The point is that someone thought about it and implemented a procedure, and a factory that has not thought about it is measuring warm rotors against cold tolerances and does not know whether its process is centered or biased.

03 Grinding Burn

If heat at the wheel-workpiece contact exceeds coolant extraction capacity, the surface re-austenitizes. On cooling, untempered martensite forms on top of an overtempered substrate. Both layers are dimensionally correct and metallurgically wrong.

Detection

Nital etch per SAE ARP1820: 3 to 5% nitric acid in ethanol on the cleaned surface, look for color anomalies. Barkhausen noise inspection using Stresstech equipment: electromagnetic sensor, no surface contact, faster than etch, requires calibration against reference samples specific to each material and hardness.

The burn happens when a coolant filter clogs, a nozzle shifts, or an operator pushes one extra rotor before dressing the wheel. These are not dramatic process failures. They are Tuesday afternoon. Every-piece inspection catches the result. Statistical sampling misses most of it. The field consequences take months or years to appear: surface cracking, spalling, debris in the oil, secondary damage, vibration increase, airend teardown. Connecting the field failure to the grinding event requires traceability from airend serial number back to grinding machine, date, wheel condition, and inspection record.

04 Heat Treatment, Housing, Assembly

Nitriding at 500 to 580°C for oil-injected rotors. Occurs below austenitizing temperature, so distortion stays minimal, so less finish grinding stock, so shorter grind cycles, so lower burn risk. The downstream benefits propagate through the entire process chain. Carburizing at 850 to 950°C gives deeper case and substantially more distortion, which means more grinding, which means more cost and more thermal risk. Nitriding won.

Oil-free rotor coatings, PTFE-filled epoxy or plasma-sprayed ceramic, applied at 50 to 150 microns and ground to final dimension afterward. Coating debonding is the dominant warranty issue in oil-free machines and traces to surface preparation.

Housing is grey iron, sand-cast, stress-relieved, rough-machined, stabilized for days, finish-machined with both bores in one clamping. Center distance between bores within 10 to 15 microns. Bore finish at Ra 0.8 or better. Rougher bores hold abrasive run-in debris longer.

Bearings by thermal fit per bearing manufacturer specifications. Axial discharge clearance calculated from differential thermal expansion between steel rotor and cast iron housing. Some manufacturers vary clearance by intended operating duty and climate. Oil-injected airends expect controlled run-in wear during initial hours. The 500-hour initial oil filter change captures this debris. Post-run-in spectrometric oil analysis checking iron and chromium is the sharpest assembly quality check available at that stage.

05 Selective Rotor Pairing

After CMM inspection, some manufacturers match male and female rotors by complementary deviation patterns. A male leading flank measuring slightly positive gets paired with a female whose corresponding surface trends negative. The meshing clearance of the pair sits closer to nominal than random pairing. Replacing a single damaged rotor of a matched pair with a standard spare alters the clearance relationship. Whether this matters enough to justify replacing both rotors depends on the application.

06 Profile Geometry

SRM's asymmetric profile provided the foundation. Kaeser's Sigma profile and Atlas Copco's proprietary developments followed. Fewer than thirty coefficients define a profile. A thin female cusp improves sealing and deflects under grinding forces. A tight curvature transition at the lobe tip demands dressing accuracy the wheel may not sustain across its full dress life. Factories that have iterated through multiple profile revisions, feeding back grinding scrap data and decade-long field failure records, carry knowledge about where the boundary sits between what simulates well and what grinds well in production. Profile design and grinding process engineering are not sequential activities separated by a wall between the design office and the shop floor. They are the same activity, or they should be.

07 Oil Separator and Factory Testing

The coalescing separator element determines the oil carryover spec. Bought from filter media suppliers like Mann+Hummel or Donaldson. Variation in fiber packing and bonding quality across lots moves the carryover number between identical machines.

Factory acceptance per ISO 1217 Annex C or E covers FAD, SEC, discharge temperature, oil carryover. Vibration per ISO 20816-3. SEC spread across a batch runs 3 to 6 percent. Tracking the batch mean over weeks reveals upstream process drift. Vibration spectra contain manufacturing signatures: unbalance at 1× shaft speed, profile error at lobe-pass frequency, bearing defects at ball-pass frequencies derivable from bearing geometry via the SKF frequency calculator. A BPFO spike at factory test sends the machine back to assembly. Catching it there costs hours. Missing it costs a field rebuild.

— end —