Air Compressors for Oil and Gas Including Instrument Air Breathing Air and Offshore Applications

Instrument air runs pneumatic control equipment. Breathing air goes into people's respiratory systems. On offshore platforms both come from machinery that has to function inside weight budgets, corrosive marine atmospheres, hazardous area classifications, and on FPSOs, a floor that won't stop moving. The three topics get bundled together because the equipment overlaps, even though the consequences of failure are completely different in each case.

Instrument Air Quality

ISO 8573-1 is how compressed air quality gets specified. Oil and gas instrument air lands on Particulate Class 1, a -40°C pressure dew point (moisture Class 2), and Oil Class 1 at 0.01 mg/m³ residual oil. These numbers show up in nearly every instrument air specification worldwide with minor variation.

The dew point figure is at working pressure. People get this wrong often enough that it is worth stating bluntly: at 7 barg, -40°C pressure dew point means roughly -56°C atmospheric equivalent. Write the spec with the atmospheric number by mistake and the dryer is undersized from day one.

Instrument air piping between the compressor room and the far end of a large oil and gas facility can cover hundreds of meters. Temperature along that run changes depending on sun exposure, insulation, heat tracing, and proximity to hot process equipment. Condensation anywhere in the network puts liquid water into the air stream. Water reaching pneumatic positioner orifices degrades valve response time. Control loops start hunting. In cold climates or during cold nights in desert environments, water in a valve body freezes. Ice expands. Valve bodies crack. A single critical control valve locking up on a facility running hundreds of thousands of barrels per day forces an emergency shutdown.

A single critical control valve locking up on a facility running hundreds of thousands of barrels per day forces an emergency shutdown.

That last point, the shutdown consequence, is why moisture control dominates instrument air engineering discussions. It is also why the following dryer issue consumes so much field troubleshooting time relative to how simple the fix turns out to be.

Twin-tower adsorption dryers switch between towers when the active tower saturates. During the few seconds the switching valve takes to actuate, a slug of wet air passes downstream. Sub-ten-second event. Heatless dryers switch every five to ten minutes. Each switching cycle pushes a small moisture pulse into the network. Over hours the pulses accumulate. Condensate collects at low points near remote instrument tapping points.

Field engineers see clean dew point readings at the dryer outlet and water at the far end of the piping. They check for pipe leaks. They pull apart condensate traps. They swap dew point sensors. They send the dryer to the manufacturer's service rep for evaluation. Weeks pass. The dryer checks out fine because the transient is too short for standard monitoring to register.

Buffer vessel with a drain at the dryer outlet. That fixes it. Not included in packaged dryer configurations.

Why not? Probably because it is so simple that it feels like an insult to include it as a line item on a quotation for a several-hundred-thousand-dollar dryer package. Compressor and dryer packagers compete on features and performance specs, not on a carbon steel pot with a float drain. Nobody's sales presentation has a slide about it. So it gets left off, and the field crew spends three weeks finding the problem, and then someone welds in a buffer vessel and the problem goes away, and the next project's spec still does not include it because the lesson stayed on that site with those people.

Heatless regeneration dryers burn about 15% of their output as purge air. Effective capacity drops to 85% of nameplate. At several thousand Nm³/h total demand, that 15% is expensive in both electricity and compressor oversizing. Blower-heated regeneration dryers cut purge losses to a few percent. They cost more upfront. Heating elements, blowers, extra controls, bigger skid. Past about 1500 Nm³/h, the blower-heated option wins on a five-year total cost basis according to most lifecycle comparisons done by operating companies. Below that, the capital premium is hard to justify.

Oil in Instrument Air

"Oil-free compressor" means no lubricating oil is injected into the compression chamber. It does not mean the compressed air contains zero oil. The machine still has oil-lubricated bearings and a gearbox. Shaft seals separate the lubricated section from the compression path. Seals degrade. Trace oil vapor migrates across the seal clearance into the air stream. How much, and how fast this worsens, depends on the specific machine, operating hours, maintenance history, and seal design. ISO 8573-1 Class 0 was added after the industry spent years arguing about this, and it requires the user to define acceptable limits rather than relying on a fixed table value.

Even with oil-free compressors, breathing air and sensitive process gas applications keep activated carbon filtration downstream. Seal condition is unpredictable enough that eliminating the last line of defense is not considered acceptable practice.

The alternative approach, which is common in smaller facilities and budget-conscious projects, uses oil-injected screw compressors followed by coalescing filters and activated carbon adsorbers. This works when everything is maintained on schedule. The activated carbon gets replaced when its capacity drops. The aftercooler keeps the air temperature entering the adsorber within the design window. The coalescing filter elements get swapped before differential pressure climbs too high.

In practice, replacement intervals slip. The aftercooler fouls and nobody notices because there is no alarm on adsorber inlet temperature in most standard packages. The activated carbon saturates faster than expected because it is running hotter than rated. Oil vapor starts getting through. There is no online oil content monitor. Months pass.

Here is what that does downstream, and this part is worth going into at some length because the damage mechanism is slow enough that it falls outside the span of attention of any single maintenance cycle.

Oil film deposits on the diaphragm of a pneumatic control valve actuator. The valve does not stick. It does not fail. What happens is that the diaphragm surface becomes less uniform in friction characteristics. When the control signal calls for a small change in valve position, the actuator has to overcome a slightly higher breakaway resistance on the contaminated portion of the diaphragm. Positioning accuracy drops. Hysteresis increases. The PID controller in the DCS adapts. It increases its output signal amplitude to achieve the same process variable setpoint. The operator sees the valve output trace getting wider on the trend screen, but the process variable is holding steady, so nothing gets flagged. The PID is doing its job. Over six months, sometimes over two years, the PID compensation reaches its output limit. The valve starts visibly sticking at certain travel positions, overshooting at others. At this point the diaphragm and the actuator seals need replacement, and depending on the valve, the body internals may also be damaged.

No inspection interval bridges a causal chain that long. The oil entered the air stream in month one. The valve needed replacement in month eighteen. Those two events do not appear related in any maintenance database because they belong to different equipment tags, different craft disciplines, and different PM schedules.

Specifying oil-free compression for facilities above about five million barrels of oil equivalent annual throughput eliminates this chain at the source. The cost premium over oil-injected is there, but it buys out a class of downstream failures that are expensive, disruptive, and nearly impossible to attribute correctly after the fact.

Breathing Air

Breathing air and instrument air sometimes come from the same compressor. Treatment is completely separate. Instrument air treatment removes moisture and oil. Breathing air treatment removes those plus toxic gases, and it has to contend with whatever the ambient air contains on a given day.

Grade D breathing air under OSHA 29 CFR 1910.134: oxygen 19.5% to 23.5%, CO at 10 ppm max (some operators run a 5 ppm internal limit), CO₂ at 1000 ppm max, low oil, no objectionable odor. EN 12021 in Europe narrows the oxygen band to 21% plus or minus 1%. These numbers are well established and not controversial.

What makes breathing air engineering difficult at oil and gas sites has nothing to do with the standards and everything to do with the intake environment.

A compressor intake on a refinery or gas plant pulls in whatever is in the air. Wind shifts. A pressure safety valve lifts on a process vessel and vents hydrocarbon vapor. Ppm-level concentrations of H₂S or light hydrocarbons enter the intake. For instrument air, this ages rubber components faster. For breathing air, H₂S at low ppm concentrations causes olfactory fatigue followed by respiratory paralysis.

Gas detectors at the compressor intake trigger a shutdown interlock. The detector responds. Electrochemical H₂S sensors have T90 times in the range of thirty seconds to a minute depending on the manufacturer and model. Semiconductor sensors are faster. After the detector responds, the signal goes to the safety logic controller. The controller processes it, sends the output to the compressor control panel, and the compressor shuts down. End to end, about two to three minutes.

In that two to three minutes, the compressor has been running. It has compressed contaminated air and pushed it into receiver tanks and the downstream piping network. If someone on the platform is connected to a breathing air supply hose at that moment, they are breathing stored air that was compressed during the contamination event.

Faster detectors help at the margins. They do not eliminate the problem, because the physical diffusion of gas from a leak source to the detector also takes time, and the compressor runs through all of it. A second H₂S detector and automatic shutoff valve at the use-point end of the breathing air supply, near where workers connect their hoses or masks, creates a second barrier that operates independently of the intake system. The cost for this is low. It shows up in some operator specifications and not others.

Hopcalite catalyst handles CO conversion. Manganese-copper composite oxide. Extremely humidity-sensitive. If the air reaching the catalyst bed carries moisture above a certain threshold, conversion efficiency collapses. The catalytic oxidizer has to sit downstream of the dryer in the treatment sequence. Some packaged breathing air systems put it upstream to make the skid more compact. In the Persian Gulf, Southeast Asia, West Africa, anywhere with high ambient humidity, this layout produces poor CO removal performance. Checking the catalytic oxidizer position on the P&ID relative to the dryer takes about thirty seconds and reveals whether the supplier understood what climate the equipment is going to.

Testing frequency for breathing air quality on paper is quarterly full-spectrum lab analysis. Getting sample cylinders from an offshore platform to a certified lab involves logistics that defeat quarterly schedules at most locations. Twice a year is closer to what happens. Between lab reports, daily assurance rests on online CO monitors. These cover one parameter. Hydrocarbon contamination accumulating slowly, catalyst fines shedding into the air stream, trace H₂S getting past a partially degraded carbon bed: the online CO monitor does not see any of it. The system runs with partial visibility between lab tests. Designing with extra source-side redundancy compensates for the gaps in downstream monitoring coverage.

EEBD cylinders charge to 200 or 300 bar. At 300 bar, a 0.5 ppm contaminant at intake conditions concentrates to something in the neighborhood of 150 ppm inside the cylinder, before subtracting what multi-stage cooling and filtration remove. The pressure ratio multiplies intake contamination directly. High-pressure filling systems are far more sensitive to intake quality than low-pressure instrument air, and this is easy to overlook when writing the filling system spec because the spec template focuses on pressure and flow rate.

Offshore Platforms

Weight, space, corrosion, hazardous area classification, helicopter logistics, and on FPSOs, hull motion. All of these hit at once.

Oil-free screw compressors dominate offshore instrument air now. Weight is part of it. The bigger factor is what it takes to do maintenance on an offshore platform. Reciprocating compressors need piston rings, valve plates, and packing replaced at fixed intervals. Offshore, parts come by supply vessel on a schedule that weather can disrupt. Technicians come by helicopter. Each mobilization means bed space allocation on a platform where beds are limited, work permit processing that competes with every other simultaneous maintenance and operations activity, and the downtime itself. Oil-free screw compressors push core component maintenance past twenty to thirty thousand hours. Fewer mobilizations per platform life, and each avoided mobilization saves costs that compound far beyond what the equipment price difference would suggest.

FPSOs move. Oil-free screw compressor rotor bearings are designed and tested with the machine sitting on a flat, stable foundation. An FPSO pitches and rolls continuously in open water, with amplitude and period that vary with sea state. Bearing load distribution shifts with hull angle. Lubricating film formation in the bearings gets disrupted under sustained oscillation. Bearings accumulate fatigue damage faster than the manufacturer's standard life prediction model indicates, because that model does not include dynamic tilt loads.

Some operators have this covered in their procurement specs. They require the compressor manufacturer to provide a bearing life analysis within a defined envelope of tilt angle and oscillation frequency. The manufacturer does not volunteer this.

Their standard test rig does not include a motion platform. The operators who have this clause in the spec got it one of two ways. Either they had bearings fail earlier than expected on a previous FPSO project and investigated until they found the cause, or they worked with a joint venture partner or consultant who had gone through that experience. It is a piece of knowledge that sticks with individuals and gets carried from one project to the next through people, not through published guidelines. When the person who added the clause retires, the next project may or may not include it.

Hazardous area classification: IEC Zone 0, 1, 2 or North American Division 1, 2. Compressor rooms are usually non-hazardous or Zone 2. Space constraints on a platform sometimes push equipment into Zone 1 territory. Ex d or Ex e rated motors and controls cost about twice as much as standard. Pneumatic or hydraulic starters avoid the electric motor classification issue entirely, which is why they appear on offshore compressor equipment lists with some regularity.

Salt spray gets into everything. Galvanized carbon steel piping lasts months, not years, in the North Sea or Southeast Asian waters. Instrument air piping runs in 316L stainless or copper-nickel. Cooler fins get epoxy-coated aluminum or full stainless construction. Compressor intakes need multi-stage coalescing filters to strip salt droplets out of the incoming air before they reach the rotor. Salt crystals inside a screw rotor chamber are abrasive. They also clog dryer adsorbent beds downstream.

There is a material specification gap that keeps showing up on offshore projects, and the reason it keeps showing up has nothing to do with engineering knowledge or budget. The instrument air main header is 316L per the Piping Class datasheet. The small-bore tubing and fittings between the main header and the field instrument, that last meter or so, gets specified by the instrument engineering group from a different specification document. It often ends up as 316 stainless or galvanized carbon steel. On exposed deck areas, the galvanized fitting corrodes from the inside. Airflow carries the corrosion products into the instrument. The instrument blocks. It looks exactly like an air quality problem. It is a materials problem. Running 316L all the way to the instrument connection adds almost nothing to the project cost. The piping engineer and the instrument engineer sit in different departments. Their specs have different scope boundaries. The transition zone between the header and the instrument connection falls in between and gets covered by neither document explicitly. Someone finds a corroded fitting. It gets fixed. The spec gap stays open. Next project, same thing.

Loss of instrument air offshore means every pneumatic valve on the platform loses motive power. Full emergency shutdown. Redundancy is universal: 2x100% or 3x50%, with automatic standby start on lead machine trip.

Standby compressors sit. Sometimes for weeks, sometimes for months. Control system batteries drain slowly. Starter valve seals lose elasticity from static compression. Lubricating oil oxidizes or absorbs moisture. Motor insulation absorbs ambient humidity. The machine does not look any different from the outside. Running it on a rotation schedule, at least a day or two per month on lead duty, catches these degradation modes. Receiver tank sizing has to cover total system demand during the startup interval, which is longer than the manufacturer's stated startup time once signal delays, soft-start acceleration, and unloader valve response are included. Most operators add around 50% margin on the datasheet figure for tank sizing.

Rotation gets skipped. It gets skipped because nothing bad happens when it gets skipped, at least not immediately. The maintenance planner has a screen full of overdue work orders with more immediate consequences. Standby rotation gets deferred. After a few months the answer to "will this machine start on demand" is "probably." Some operators assign rotation the same priority class as safety-critical device testing, which protects it from schedule pressure. Whether this happens depends on who the maintenance lead is and what that person considers important.

Breathing air backup offshore is an independent high-pressure cylinder bank. Capacity is maximum personnel on board multiplied by supply duration, fifteen to thirty minutes typically. Cylinders charged at commissioning and left at pressure for years develop seal problems. Nitrile and fluoroelastomer seals under sustained high-pressure static loading take a permanent compression set. The seal material deforms and does not recover its original shape when pressure is released. Cylinder valve seals leak. Pressure regulator seat seals do not seal properly against the downstream side. Either outcome compromises the emergency supply.

Gas quality testing of these cylinders has regulatory backing, audit checklists, third-party lab reports. Seal condition does not. No regulation specifies a cycle or a method for seal inspection or replacement on emergency breathing air cylinder banks. The gas quality gets tested because auditors ask for the certificate. The seals get tested if someone on the maintenance team knows about elastomer compression set and has the authority to put it on the schedule, and then defends it when the schedule gets squeezed, which it always does during turnarounds and during periods of high concurrent work activity. If that person leaves the platform, seal testing may stop being scheduled.

Where These Applications Intersect

Instrument air, breathing air, and offshore constraints converge on the same skid, the same deck space allocation, the same procurement package. ISO 8573-1 for instrument air quality. EN 12021 and OSHA Grade D for breathing air. ATEX and IECEx for hazardous area equipment. Minimum weight and footprint.

The failure modes in this article do not show up during factory acceptance testing and do not appear on commissioning punch lists. Factory tests verify performance at rated conditions on a test stand in a controlled environment. The dryer switching transient matters only after thousands of switching cycles have deposited enough moisture at remote points. The catalytic oxidizer humidity sensitivity matters only in tropical or Gulf climates, and the FAT happens in the manufacturer's shop in Germany or the United States. The intake detector response lag matters only during a coincidence of gas release and personnel being connected to the breathing air network. Standby compressor degradation matters only when the lead machine trips after months of the standby sitting idle.

These are boundary conditions that exist outside the scope of what project delivery teams check. The project team commissions the plant, verifies performance, closes the punch list, hands over to operations, and moves on. Operations inherits whatever was built. If the design engineer who wrote the specification had read incident investigation reports from similar facilities, the spec may contain clauses that address these conditions. If not, the conditions exist in the system, unmitigated, waiting for the right combination of circumstances.

Procurement specifications that address obscure but consequential conditions, FPSO bearing motion envelopes, breathing air use-point isolation, dryer outlet buffer vessels, cylinder bank seal testing schedules, tend to trace back to individuals who encountered the failure and added the clause afterward. The knowledge lives with the person. Specifications carry the clause but not the reasoning or the history that led to it. When the person moves on, the clause sometimes survives in the template and sometimes gets deleted by someone who does not recognize its purpose during a spec rationalization exercise. The lesson re-enters the organization's experience the hard way, through recurrence.