Compressed air can pass testing at the compressor house and contaminate the product at the final connection. An oil-free compressor, an effective dryer and a new filter describe equipment; individually, they do not demonstrate air quality during use. Engineering decisions should start where air interacts with the process and work backwards to the intake.
This article addresses process air, air contacting product or primary container surfaces, and instrument air. The network name does not establish criticality. An apparently non-contact actuator may exhaust beside exposed product, while a separate line may serve equipment without affecting quality. Describe the pathway, exposure and failure states before selecting a specification. Include temporary connections and hoses because the installed permanent pipework may represent only part of the actual delivery route.
Turn intended use into a testable specification
[QRM] For each user, record the function, manufacturing stage, contact duration, required flow, pressure and exhaust destination. Distinguish direct contact, indirect contact through surfaces and non-contact use. Also consider shutdown, depressurisation, maintenance and restart: normal flow direction does not exclude reverse flow during these states.
A useful risk assessment connects each contaminant with a consequence. Particles can enter product or interfere with components; water can condense and create unsuitable conditions; hydrocarbons can transfer to surfaces; microorganisms can compromise a sensitive process. Identify the preventive barrier, where it will be verified and the decision required if it fails. A contaminant missing from a supplier brochure has not thereby been excluded from the risk assessment.
Specify the attribute, justified limit, units, reference conditions, sampling location, method and operating state. Separate the quality requirement from the engineering target. A more conservative design target can support reliability and early detection without automatically becoming a batch acceptance criterion. Keep the rationale in approved user requirements, risk assessments and control strategies, and identify who owns each decision when several departments share the utility.
What the sources establish and what the site decides
[REGULATORY REQUIREMENT] Within sterile manufacture, EU GMP Annex 1, the 2022 revision, sections 6.18–6.20, connects gas quality with use and addresses filtration in aseptic processes and prevention of backflow. Establish its applicability to the particular process. It does not prescribe one ISO class for every pharmaceutical compressed air service. Check the actual current monograph where a compendial requirement applies.
[TECHNICAL STANDARD] ISO 8573-1:2010 provides a classification language for particles, water and oil. The ISO catalogue lists it as published with revision planned; a future edition is not thereby effective. Technical classification does not replace GMP justification. [GUIDANCE] ICH Q9(R1), effective in the EU since July 2023, supports decisions proportionate to risk and available knowledge.
[GEP] The architecture, distribution and maintenance considerations below are engineering reasoning to verify against the real installation. [GUIDEGXP RECOMMENDATION] Approve a quality profile for each family of users and document exceptions, avoiding both unexplained uniform specifications and requirements without an accountable owner.
Evaluate the complete treatment architecture
Assess an oil-free compressor together with intake air, lubricants potentially present in auxiliary components, wear, materials and performance under expected conditions. An oil-lubricated machine requires explicit assessment of treatment and failures that could compromise oil removal. The comparison concerns the complete system's ability to maintain quality, rather than the compression chamber technology alone.
Locate the intake away from identified vapour, exhaust and dust sources. Check seasonal variation and changes to neighbouring equipment. Aftercoolers, separators and drains should manage the expected condensate load. A blocked drain can transfer water into subsequent treatment; a continuously open drain can reduce capacity and disturb control. Both conditions require detection and accessible maintenance, including a practical way to confirm that corrective work restored performance.
Compare refrigerant and adsorption dryers against the required moisture condition, minimum network temperature, load, ambient conditions and regeneration method. Include transitions, power failure and high-flow performance. Receivers buffer demand changes, but installation before or after the dryer changes treatment loads, drainage and network behaviour. Available capacity and conforming air remain separate conditions. A standby machine that delivers pressure while bypassing essential treatment has not established continuity of qualified supply.
Particles: define the location and measurement conditions
Particles may originate from the intake environment, wear, desiccant material, corrosion, construction work or terminal connections. Generator testing does not cover contributions introduced downstream. The measured size range, analysed volume and pressure-handling arrangement influence the result and must be consistent with the selected criterion.
A particulate filter does not automatically demonstrate microbial retention, oil vapour removal or suitability of every downstream component. Assess materials, operating conditions, capacity, cleaning compatibility and replacement arrangements. Sampling should prevent dirty adapters from creating misleading failures, while also avoiding unusually favourable preparation that hides the actual condition of use. Keep diagnostic samples distinguishable from samples used to assess compliance.
Moisture content and pressure dew point answer different questions
Water content describes the amount present. Pressure dew point describes the temperature at which water vapour condenses at the stated pressure. A reading after pressure reduction is not directly interchangeable with one in a pressurised line. Record measurement pressure, units, reference conditions and any conversion, together with its technical basis.
Derive the requirement from process sensitivity and the lowest temperature realistically reached in relevant sections. Consider external piping, cold rooms, expansion and shutdown. Monitoring at the dryer outlet indicates treatment behaviour, while measurement on a critical branch checks local conditions. ISO 8573-3:1999 addresses humidity measurement methods; it does not assign the site's GMP limit.
A contaminated or unstabilised probe, or one supplied by a stagnant sample, can produce misleading readings. Define conditioning, sample flow, useful measurement range and response time. A network that is dry during a test may still retain residues from an earlier water ingress event. Assess the event and its consequences before return to service, alongside the recovered reading. Do not assume that restoring dew point also establishes microbiological recovery.
Oil: separate liquid, aerosol and vapour
The analytical strategy must identify the fractions measured. Aerosol testing alone does not establish total oil without additional justification. Hydrocarbons drawn from the surrounding environment can matter even with oil-free compression. Ask the laboratory about measurement range, included substances, collection arrangements, reference conditions and uncertainty suitable for the decision.
Coalescence and adsorption address different mechanisms. Do not assign every contaminant-removal function to one filter element. For adsorbents and consumables, assess saturation, temperature, actual loading and storage conditions. Replacement triggered only by differential pressure may miss exhausted adsorption capacity. Procurement should therefore request evidence relevant to the claimed function and the intended operating envelope, rather than an unsupported statement that the entire package produces pharmaceutical air.
Microbiological quality and the final barrier
Particle counting does not measure viability. A microbiological result depends on sampling, recovery, medium, incubation and volume; absence of colonies in a sample does not establish sterility of the whole network. The method should be appropriate for compressed air and sampling pressure, with controls for contamination introduced by the operator or equipment.
For aseptic processes, apply the relevant Annex 1 provisions concerning sterilising filtration at the point of use, sterilisation of subsequent pipework and associated controls. Demonstrate barrier performance under operating conditions. Placing a nominally suitable component far from the process leaves a downstream segment requiring control and does not eliminate terminal contamination.
For other applications, the need for microbiological controls follows the risk and applicable requirements. Document why particular user families are included and how the programme accommodates process changes. Barrier maintenance, integrity testing where required and air monitoring provide complementary evidence. A passing integrity test and a satisfactory air sample answer different questions and should remain separately traceable in the quality record.
A practical decision matrix
| Situation | Engineering decision | Required evidence |
|---|---|---|
| Air contacting a critical surface | Define quality at the actual connection | Representative sample after relevant components |
| Instrument exhausting near product | Assess exhaust pathway and segregation | Risk assessment and installation verification |
| Cold network section | Justify the moisture requirement | Temperature profile and measurement at the correct pressure |
| New final filter | Define its function and protected segment | Performance data, installation and applicable testing |
| Standby compressor | Qualify the takeover mode | Quality and pressure during changeover |
Sample and qualify the complete delivery chain
The plan should cover generation, distribution and points of use selected with a documented rationale. Remote branches, intermittent users, modified segments and simultaneous demand may represent different challenges. Do not assume that one location is worst for every contaminant. A pressure-critical position may not be the most challenging location for microbiology or moisture.
Describe filter and valve states, the presence of hoses, demand during collection and any purge included in the normal operating procedure. Distinguish samples representative of use from investigative or diagnostic samples. Retain instrument identification, calibration status, original data, operating conditions and invalid results with their assessment. Where a laboratory reports a result below its quantification capability, ensure that the reporting limit supports the actual acceptance decision.
Qualification should challenge startup, justified low and high loads, machine changeover, treatment failures and recovery. Test duration and campaign numbers depend on system knowledge and risk, rather than a universal rule. Routine monitoring maintains evidence of control; it does not repair an incomplete initial qualification. Link periodic sampling to continuous operational information so a result can be interpreted against the state of the system when the air was used.
Example: one compressor house, two different needs
A site supplies closed actuators and a drying operation for product-contact components. Pressure is adequate, but the compressor-house sample excludes the hose used for drying. The assessment identifies this hose, the local regulator and repeated connections as possible contamination contributors.
The team defines two user families, retains common requirements where justified and adds specific controls for drying. It verifies quality in the actual configuration, hose handling, filter status and conditions after shutdown. The answer is not necessarily a new compressor. Correcting the terminal interface can be appropriate if testing demonstrates effectiveness and central capacity remains adequate.
The change record also specifies how replacement hoses are approved, where clean assemblies are stored and which intervention triggers verification. Without those operational details, an effective qualification arrangement may gradually be replaced by components that were never evaluated. The engineering solution is complete only when normal work can preserve it.
Respond to a failure without losing evidence
A failed result first requires identification of potentially affected users, periods and production. A sample represents a location and a time; do not automatically extend or restrict its significance to the whole network. Compare configuration, operating sequence, maintenance, dryer information and changeovers. Preserve the observed condition before changing components when compatible with safety and product protection.
Repeat sampling should test a documented hypothesis. If an adapter is suspected, a planned comparison can distinguish equipment contribution from line contamination. A favourable repeat does not erase the original result. Following water ingress, also assess microbiological consequences and residues in infrequently used branches. Following oil contamination, consider affected downstream surfaces alongside the compressor fault.
Return to service includes correction, barrier verification, relevant testing and an authorised decision on use. Identify which evidence is immediately available and which requires laboratory time, and manage affected production through approved procedures. A temporary increase in monitoring should have a defined purpose and an exit criterion. This makes CAPA an evaluation of corrective effectiveness rather than merely a record that a suspected component was replaced.
Common mistakes and approval checklist
Frequent errors include copying an ISO class from another facility, interpreting oil-free as absence of every hydrocarbon, confusing atmospheric and pressure dew points, treating particle counting as microbiological evidence and always sampling without the hose actually used. Another concern is leaving a treatment bypass available without an approved condition of use and a clear mechanism to prevent inadvertent operation.
- Every user has a documented purpose, contact pathway and failure consequence.
- Every attribute has a justified limit, units, location and method.
- Treatment performance includes relevant transitions and ambient conditions.
- Quality is verified after components capable of changing it.
- Standby operation, bypasses, maintenance and return to service have explicit criteria.
- Results, alarms and modifications are reviewed with production information.
Connect these decisions through Critical Utilities Systems. Interfaces with sterile product belong alongside Aseptic Fill-Finish & Barrier Systems; room contamination control remains within Cleanrooms & HVAC Systems.
References and applicability
- EU GMP Annex 1, 2022 revision: sterile manufacture; gas and vacuum provisions.
- ISO 8573-1:2010: technical classification, not a universal GMP specification.
- ISO 8573-3:1999: humidity measurement methodology.
- ISO 8573-7:2003: viable microbiological contamination; establish method suitability.
- ICH Q9(R1), EMA: quality risk management.
