Making sense of DOCA sensor measurement uncertainty
The DOCA Project is developing an online optical sensor for detecting oil contaminants in high-purity compressed air as liquid droplets, aerosols and vapour. Its measurement uncertainty determines how confidently a facility can identify clean air, trigger action and document compliance with an acceptance limit.
For Australian manufacturers, this matters when the EU-funded DOCA technology is evaluated against local quality systems. Uncertainty needs a clear decision rule, familiar units and evidence gathered under actual sampling and operating conditions.
Why uncertainty changes the decision
A sensor reading is an estimate, not an exact property of the air. Repeatability describes short-term variation; reproducibility captures changes between operators, instruments, sites or days; bias is the systematic difference from a reference value.
Uncertainty matters most when a result lies close to an acceptance criterion. A reading below the limit may still be statistically consistent with a value above it, while an isolated high result may not represent flowing process air. Compliance reporting therefore needs both the number and a rule for near-limit outcomes.
What the DOCA sensor actually measures
Liquid, aerosol and vapour contamination present different sampling challenges. Droplets can be captured by a filter or deposited in a probe, aerosols can enter through a particle pathway, and vapour may interact with line materials before reaching the sensing point. The DOCA design must preserve a representative sample while minimising losses, contamination and time delay.
Optical performance also depends on stable illumination, detector alignment, clean windows and reliable electronics. Controlled optical illumination guidance can inform those design choices, but metrological validation remains essential. A stable electronics baseline does not by itself establish that an oil concentration has been measured accurately.
Building a practical uncertainty budget
An uncertainty budget identifies each significant source of error, assigns a quantity and distribution, and shows how they combine. Likely contributors include the reference standard, calibration fit, flow, pressure, temperature, humidity, optical response, sample-line effects, repeatability and software processing. Irrelevant terms should be excluded.
Combined standard uncertainty, u_c, is obtained by adding variance contributions and taking the square root. Reporting often applies a coverage factor k, commonly about two, to produce expanded uncertainty U. Limited degrees of freedom or a poorly known distribution may require a higher factor or more data.
Separate testing should establish detection and quantitation limits with representative blanks and low-level samples. Detection indicates probable presence; quantitation identifies the lowest concentration reported with acceptable precision. Neither equals the site acceptance limit.
Choosing a defensible compliance rule
A practical approach uses the measured concentration, expanded uncertainty and specified limit in a written decision rule. A guard band can reduce false acceptance near the boundary: for example, a result is accepted only when the upper uncertainty bound remains below the criterion. This must reflect the consequences of error and the applicable quality system, not a universal percentage.
If no external limit applies, a site can define an action level and escalation band based on risk, historical baseline data and process capability. Trending may then reveal drift earlier than repeated pass/fail decisions. Crucially, any guard band or threshold selected for a DOCA installation should be validated with the full system, including sampling hardware and software.
Demonstrating fitness for Australian operations
For pharmaceutical manufacturers around Melbourne and Adelaide, compressed-air assessment sits within a TGA-inspected GMP framework, often aligned with PIC/S principles. A European prototype is not accepted automatically. Validation should connect DOCA performance to the contamination-control plan, sampling locations, maintenance and quality records.
Hospital and laboratory engineering teams in Melbourne and Sydney commonly discuss commissioning, spot checks and trend review, so online DOCA results may complement scheduled verification. In automotive and electronics plants around Brisbane, Adelaide and Melbourne, process air supports pneumatic equipment and precision cleaning, often under supplier-specific acceptance criteria.
Australia’s calibration market also shapes the evidence trail. NATA accreditation under ISO/IEC 17025 is a familiar route to traceable calibration, though not every in-house check needs it. Long distances between sites make reference standards, proficiency testing, field verification and retained records important for separating instrument drift from process change.
Recommendations for defensible reporting
Before using DOCA readings for a compliance decision, the responsible team should:
- State the measurand, location, flow condition and sampling method.
- Report the result with expanded uncertainty, units, coverage factor and decision rule.
- Validate the complete sampling and sensing chain in the intended operating range.
- Check zero response, low-level performance, drift and agreement with a suitable reference method.
- Record deviations, maintenance, calibration history and any assumed conditions.
The final report should distinguish a measured value from a regulatory requirement and identify the source of every acceptance limit. This prevents a technically precise number from being interpreted as proof of compliance when the underlying criterion, sampling plan or uncertainty treatment is incomplete.
Turning uncertainty into operational confidence
The practical value of DOCA is turning uncertainty into a repeatable decision. Teams can define acceptance and action levels, then use project testing and local verification to assess whether the complete measurement system is fit for purpose. Ownership of sampling, calibration and quality review must be clear.
Organisations evaluating DOCA should request its uncertainty budget, validation protocol, detection-limit data and patent information. A monitored trial at a representative site can compare results with existing methods and build a documented case for operational or quality use.