Food-Grade Compressed Air and FDA SQF and ISO 22000 Requirements

An SQF auditor pulls a coalescing filter element out of the housing on the main header feeding the packaging hall. It's brown. Oil is running down the pleats. The maintenance tech next to him, guy named Rick, been at the plant eleven years, says he put in three work orders to get a shutdown window for that change. Production said no each time. The CMMS shows the requests. Rick has printouts in his locker because he's been through this before and he knows how the corrective action meeting goes.

The auditor photographs the element, photographs the differential pressure gauge (red zone, pegged, has been for weeks), and writes the finding.

In the corrective action meeting two weeks later, the plant manager asks Rick why the filter wasn't changed. Rick doesn't mention the three denied shutdown requests. He's learned that pointing up the chain in corrective action meetings makes the next shutdown request even harder to get. He says the PM was overdue and it's been corrected. The corrective action form says "element replaced, PM frequency increased to semi-annual." The form gets filed. The underlying problem, that production controls the maintenance schedule and the compressor room has no food safety oversight, doesn't appear anywhere in the corrective action record.

That dynamic is more relevant to compressed air compliance in food manufacturing than any ISO purity class or SQF clause number. The three regulatory frameworks in this article's title (FDA 21 CFR 117, SQF Code Edition 9, ISO 22000 with ISO/TS 22002-1) all require compressed air contacting food to be clean. They differ on how specific they get about what "clean" means and how to prove it. SQF is the most prescriptive. FDA is performance-based and says almost nothing specific about compressed air but holds the manufacturer liable for outcomes. ISO 22000 asks the facility to design its own program through hazard analysis, and the resulting program quality depends almost entirely on whether anyone on the FSMS team understands compressed air engineering, and in most facilities the answer to that is no.

This article spends most of its length on SQF because that's where the prescriptive language lives and that's where the findings get written. FDA gets a section because the enforcement cascade is different and worse in a specific way. ISO 22000 gets a paragraph because the compliance problems it creates with compressed air are mostly PRP-versus-OPRP classification errors, and there isn't much to say about that beyond identifying it.

Section 11.5.4 of Module 11, Edition 9. Compressed air contacting food shall be regularly monitored for purity by qualified personnel. Air handling equipment shall be maintained and verified. Air shall be clean, adequately filtered, present no contamination risk.

Acceptance criteria: the monitoring plan says ISO 8573-1 Class 1.2.1. Ask the facility what the particle limits are for Class 1 and the room goes quiet. ISO 8573-1 costs money. Many facilities reference it in their food safety plan without owning a copy. Class 1 for particles is 20,000 per m3 at 0.1–0.5 microns, 400 at 0.5–1.0, 10 at 1.0–5.0. If those numbers aren't written into the monitoring plan in units that match the lab report format, the pass/fail determination rests entirely on whatever "compliant/non-compliant" column the testing service provider includes on their report template. The facility can't independently verify it. The food safety manager who reviews the report is taking the service provider's word for it.

And the service provider's report might be based on a sample collected at the compressor room wall rather than the point of use, because that's where the test ports are and that's where sampling is fastest, and a Class 1.2.1 result at the compressor room wall can coexist with a Class 4.6.3 reality at the filling nozzle fifteen meters downstream. Between those two points: piping corrosion, PTFE tape fragments from threaded joints, condensation at wall penetrations, quick-disconnect fittings with biofilm in the internal cavity. The purity degrades through the distribution system and the test report reflects conditions at the cleanest point in the loop.

Contact point inventories are where auditors find the second category of problems. The facility's documented list has eight contact points. The auditor walks the floor and counts twenty or more. Blow-off nozzles and air knives were on the list. Pneumatic actuator exhaust vents weren't. Conveying system inlets weren't. A tank agitation port that got added during a line expansion eighteen months ago wasn't. Connections accumulate. Equipment gets installed, lines get modified, maintenance adds a blow-off here and there, and none of it goes through a management-of-change process that includes food safety review.

And SQF requires the risk assessment to cover indirect contact, meaning air released into environments where food is exposed. A pneumatic cylinder exhausting into the packaging room twelve feet from open product counts. Most facilities haven't included actuator exhaust in their risk assessments. Auditors five years ago didn't ask about it. They do now.

Validation. Element 2.4.3. The system has to be demonstrated under operating conditions to achieve the specified purity class. A weekend shutdown validation with half the compressors offline and no production load doesn't represent peak-demand Thursday when system pressure sags, the dryer can't hold its dew point, and the coalescing filter differential pressure is driving captured oil back through the media. Auditors who understand compressed air ask when the validation data was collected relative to production load. The answer usually reveals that the validation was run under conditions chosen for convenience rather than conditions chosen for representativeness.

SQFI requires food safety knowledge. Compressed air is mechanical engineering. A food science or quality management background gives you Listeria, allergens, sanitation chemistry. It doesn't give you compressed air treatment train design, ISO 8573-4 sampling statistics, or the ability to look at a desiccant dryer regeneration log and spot a pattern of dew point excursions that correlates with downstream particulate spikes. Some auditors have picked this up through experience. These auditors open filter housings, read differential pressure gauges, walk piping from compressor to point of use, check activated carbon replacement records. The facilities they audit develop better programs. Facilities that draw auditors without this background, and there are many, can pass consecutive audits with programs that wouldn't survive ten minutes of engineering scrutiny. GFSI benchmarking evaluates the scheme's written requirements. It doesn't evaluate the auditor population's technical capability for specialized elements.

The unannounced audit changes this dynamic somewhat. SQF mandates unannounced audits for recertification, and the compressor room on a random Wednesday looks nothing like the compressor room the week before an announced audit. Differential pressure gauges in the red. Auto drains that failed closed weeks ago. Point-of-use filter housings removed during changeovers and never reinstalled. The gap between the documentation and the hardware is wider for compressed air than for almost any other SQF element, because compressed air maintenance is what gets deferred first when the maintenance crew is stretched. Sanitation has a pre-op deadline enforced by the production supervisor at 5 AM. The compressed air filter has a CMMS work order that can be rescheduled indefinitely without anyone outside the compressor room knowing.

The maintenance tech knows this. The maintenance tech has been watching the gauge. The maintenance tech put in work orders. Production wouldn't give him the shutdown window, same as the last three times.

SQF scoring makes it commercial. A major drops the score, drops the grade, and the grade goes on the certificate that retailers compare against their supplier thresholds. Getting capital approved for compressed air treatment upgrades is hard in a way that getting capital for a metal detector isn't, because the people approving capital don't know what activated carbon does and the food safety manager often can't explain it because PCQI training doesn't cover it. Capital investment in food plant compressed air tracks regulatory and audit events, not planned upgrade cycles. The carbon tower goes in the quarter after the major. The sterile filters get installed after the 483. The quarterly testing contract gets signed the week the facility's biggest retail customer adds compressed air to their supplier audit protocol.

The testing service provider relationship deserves its own paragraph because the economics work against adequate monitoring. Service providers sell annual testing packages because that's what the market buys. Quarterly testing for direct product contact applications is defensible under any risk-based framework (annual testing means 364 days of unverified performance), but recommending quarterly to a facility budgeted for annual means proposing a four-times cost increase. The provider who proposes quarterly risks losing the contract to the competitor who calls annual "industry standard." So the proposals come in for annual, the facilities sign for annual, and the risk assessment that would justify quarterly exists somewhere in the food safety plan, written to fill a clause rather than to determine an answer. Whether the person who wrote it understood the relationship between sampling frequency and seasonal system performance variation is a question that the answer to, in most facilities, is no.

And most of that annual testing happens at the compressor room wall rather than the point of use. The compressor room is where the test ports are. Sampling there is faster. The results are better, which should raise a question about why the results are better, but it usually doesn't. Between the compressor room and the filling nozzle, compressed air passes through piping that corrodes, through threaded joints that shed PTFE tape fragments, past wall penetrations where condensation forms, through hoses and quick-disconnect fittings with biofilm in the internal cavity. A Class 1.2.1 result at the compressor room wall doesn't tell you what's reaching the product. Testing at the point of use is the only testing that matters for food safety, and it's the testing that costs more and takes longer, so it's the testing that doesn't happen.

Short section. FDA's compressed air requirement is 21 CFR 117.35(a): utilities shall not contaminate food. Performance-based. No purity class, no filter spec, no test frequency. Warning Letters on fda.gov include cases with compressed air contributing to adulteration findings under 402(a)(4).

The PCQI curriculum as taught by most FSPCA lead instructors covers thermal lethality, allergen cross-contact, foreign material. Compressed air barely comes up. The result, recognizable to anyone who's reviewed hazard analyses from mid-size food plants, is compressed air showing up as a single line item under one process step. Control measure: "maintenance of compressed air system." Verification: "annual review of maintenance records." Every other process step where compressed air contacts product or exhausts into the room isn't listed. An FDA investigator walking the floor with that document spots the omissions immediately. And those omissions are the ones that cascade.

ISO 22000's compressed air issues boil down to PRP-versus-OPRP misclassification. Direct compressed air contact with post-lethality RTE product needs OPRP-level control. Facilities classify it as a routine PRP, assign it to maintenance, and the program has no monitoring, no corrective action procedures, no verification. Certification body auditors are catching this more than they used to. Clauses 7.1.6 and 8.8 cause problems too (outsourced testing where facilities don't verify the provider's methodology, monitoring conflated with verification on the same form) but these are documentation gaps rather than the engineering and organizational failures that drive the SQF and FDA findings.

ISO 8573-4 specifies minimum sampling volumes for particle measurement at each purity class. At Class 1, 95% confidence requires a sample volume larger than what most field kits collect. The number on the report looks precise. The sample behind it doesn't support the precision.

Annual testing is the market default because that's what facilities buy and that's what service providers sell. The risk assessment that would justify a different frequency exists as a document written to fill a clause.

Piping gets more space here than everything else because it's the contamination source that is hardest to fix and most likely to be the root cause of a compressed air contamination event that trace-back can't explain.

Most food plants are running compressed air through piping that was installed when the building was built or during a capital project that predates the current food safety team. Maybe it's carbon steel. Maybe it's aluminum. Maybe it's a mix, with carbon steel in the ceiling runs and stainless drops near the equipment, or maybe it's all carbon steel because the original contractor priced it that way and nobody specified otherwise. Carbon steel corrodes internally. The corrosion generates iron oxide particulate that travels downstream, loads against filter elements, shortens filter life, and reaches the point of use when filter maintenance lapses. Replacing those runs with stainless or aluminum is a capital project requiring engineering, a piping contractor, and production downtime on the lines affected. It competes for budget with every other capital project.

In plants that swapped oil-lubricated compressors for oil-free units without replacing the distribution piping, the pipe walls retain an oil film from years of lubricated operation. The compressor is oil-free. The piping isn't. An oil vapor test at the point of use will find hydrocarbons that the oil-free compressor didn't produce, and the facility won't be able to explain where they came from without understanding the history of the piping.

Compressor intake location matters and isn't in the hazard analysis. Seasonal humidity swings of the kind you see in the Gulf Coast or upper Midwest can push the dryer past its rated capacity curve, raising outlet dew point above spec, producing condensation in the piping that feeds microbial growth and corrosion and carries particulate to the point of use. If the validation was in April and the annual test is in March, the worst performance window hasn't been evaluated.

Condensate drains compound the seasonal problem. Timer drains discharge on a fixed cycle regardless of condensate volume, and the cycle that keeps up in January can't keep up in August when the moisture load has tripled. Zero-loss drains with float sensing respond to actual condensate levels but fail silently in the closed position. When one fails during peak humidity, the filter housing fills with water within hours, the coalescing element gets overwhelmed with liquid it wasn't designed to handle in that volume, and liquid water enters the distribution piping. Nothing on any panel or display or alarm indicates the failure. It persists until somebody physically inspects the drain, which happens at the next PM, which may be weeks away if maintenance is stretched thin, and maintenance is always stretched thin.

That's the sentence, and it's the entire explanation for why compressed air programs fail. The food safety team writes requirements they can't verify. Maintenance runs equipment they don't know is food-safety-critical. Out-of-spec results get "retest" as the corrective action because the tech filing the form doesn't know why the result matters.

Building the program means walking every line and documenting every contact point, which consistently turns up 30 to 50% more points than the food safety team had on record. It means writing purity targets with numerical limits rather than class numbers alone. It means testing at the point of use rather than the compressor room wall. It means frequencies from a risk assessment that answered a question rather than filled a section of the plan.

For a mid-size plant this is a capital project on top of recurring monitoring cost on top of organizational change on top of a food safety team that's already overextended. Compressed air loses the budget fight year after year because it doesn't make anyone sick in a way that gets traced back to the air system. When it finally does become visible, whether through a 483 or an SQF major or a customer audit finding, the cost of the retroactive fix runs double the proactive investment because everything has to be expedited. Equipment procurement on rush. Overtime installation. A consultant brought in to rewrite the hazard analysis before the re-audit deadline. The food safety manager who couldn't get $150k approved last year is now explaining why the same project costs $300k and has to be done in six weeks.

The facilities that avoid this cycle, and there aren't many, share one characteristic. Someone in the organization, usually a maintenance engineer or a quality manager with an engineering background, decided at some point that compressed air was going to be their problem. Not because the job description said so. Not because any of the three frameworks require the role. Because the person looked at the compressor room and looked at the food safety plan and realized that the gap between the two was going to produce a finding eventually, and they'd rather build the program on their own timeline than build it on an auditor's corrective action deadline. That person is the compressed air program. When they leave, the program starts to decay within months.