The return conductivity reaches its endpoint, yet a difficult surface still carries product residue. The apparent contradiction disappears once the two measurements are separated: the online instrument describes the liquid passing its sensor, while cleaning acceptance concerns specified residues on the equipment within an approved sampling and analytical strategy. A useful monitoring system connects those observations without assuming that they are interchangeable.
This article addresses process monitoring for pharmaceutical clean-in-place cycles: temperature, flow, pressure, conductivity, chemical concentration, pH and total organic carbon where appropriate. It focuses on sensor selection, location, endpoint logic and interpretation. It does not prescribe universal operating values or replace cleaning validation with a collection of online signals.
Define the decision before selecting the sensor
Each measurement should support a stated decision. Flow may confirm delivery through an established route. Temperature may demonstrate that a wash phase remained within its developed operating range. Conductivity may support detergent concentration control or rinse progression when the chemical response is understood. These are related but different uses, and the instrument specification should reflect the intended purpose.
List the residues that matter: previous product, excipients, degradation products where relevant and cleaning agents. Evaluate chemical cleanliness separately from microbiological control and endotoxin concerns where applicable. A low conductivity or TOC result does not establish an acceptable microbiological state. Cleaning, sanitisation, disinfection and sterilisation require distinct objectives even when they share equipment or parts of a cycle.
An endpoint is a rule that determines when a process phase can finish. An acceptance criterion is a justified condition used to judge the cleaning result. The endpoint can contribute to acceptance only through a demonstrated relationship to the actual residues, equipment, method capability and validated process. Otherwise, retain it as process control and preserve the necessary laboratory or other verification evidence.
Regulatory framework and engineering interpretation
For EU human medicinal manufacture, Annex 15 (2015) provides the current cleaning validation framework. The Commission still lists that version; the 2026 revision concept paper is not a new requirement. Equipment cleaning and maintenance are addressed in the applicable GMP framework, including 21 CFR 211.67 for US operations.
The FDA cleaning inspection guide, dated July 1993, is a nonbinding investigator reference with a chemical-residue scope. Read it alongside current applicable requirements and later official positions. Its historical discussion does not create universal residue limits. The monitoring design, matrix and example below are GuideGxP recommendations, not regulatory acceptance tables.
Conductivity: useful response, limited specificity
Conductivity reflects the ability of a solution to carry electrical current. Its response depends on ionic species, concentration and temperature. A detergent may show a useful relationship between conductivity and concentration over a defined range, but that relationship must be established for the actual formulation and conditions. Different chemicals can produce similar readings, and some product residues contribute little conductivity.
Distinguish measured conductivity at the process temperature from a temperature-compensated value. The compensation model should suit the liquid and intended range. A generic setting selected by default can create misleading comparisons during a changing rinse temperature. Retain enough information to understand what the recorded value represents, including relevant configuration changes.
For concentration control, assess fresh and used cleaning solution separately where soil loading, dilution or degradation can change the response. Conductivity does not necessarily measure remaining cleaning activity. Where a laboratory concentration method or another independent check is needed, define its relationship to the online reading and the conditions that trigger investigation. A stable signal is not proof that the chemical composition remains suitable.
TOC and pH: choose the question they can answer
Total organic carbon is a nonspecific measurement of organic carbon under the method’s conditions. It can support a residue strategy when relevant substances are detected with suitable recovery, response and sensitivity. It does not identify which organic compound is present. Inorganic residues, poorly recovered material or compounds that the method does not adequately measure require separate consideration.
Assess background carbon from water, sampling materials, containers and the instrument system. For online measurements, include transport delay, sample conditioning, cleaning of the sample path and possible carryover. For laboratory results, include sample collection and holding conditions. A numerical comparison is meaningful only if units, dilution, method performance and the relationship to the sampled equipment are understood.
Use pH where it provides a reliable decision for the actual chemistry. Low-conductivity water, temperature changes, electrode condition and installation can complicate interpretation. Neither neutral pH nor low TOC automatically proves removal of every cleaning agent or product. Select a specific analytical method when the residue strategy needs chemical specificity that the online signal cannot provide.
Temperature, flow and pressure describe delivery
Temperature affects cleaning kinetics and solution behaviour, but a supply reading may not represent a remote surface or the return condition. Establish the relevant locations through development studies. Consider heat loss, residence time and sensor response. The phase should be evaluated against the approved conditions at the locations justified by the process, rather than assuming that the heater outlet defines the whole circuit.
Flow measurement can support evidence that a route received cleaning solution. It does not automatically prove local impingement, spray-device rotation or effective cleaning of every branch. Evaluate meter installation requirements, full-pipe conditions, flow direction, entrained gas and the behaviour of two-phase or intermittently flooded return lines. A plausible number from an unsuitable installation can be more misleading than an obvious instrument fault.
Pressure is useful for understanding pumping, restrictions, spray-device supply and abnormal conditions. Interpret it with the route and flow state. High pressure can indicate a restriction rather than strong cleaning action, while low pressure can arise from a changed route or pump condition. Correlate the signals instead of treating each one as an independent proof of effectiveness.
Signal selection matrix
| Signal | Useful process decision | What it does not establish alone |
|---|---|---|
| Conductivity | Detergent control or rinse progression for characterised chemistry | Identity or acceptable amount of every residue |
| TOC | Organic residue response within a suitable method strategy | Specific compound identity or inorganic cleanliness |
| pH | Acid or alkaline behaviour when the measurement is suitable | Universal chemical or microbiological acceptance |
| Temperature | Delivery of the developed thermal cleaning conditions | Equivalent exposure at every unmeasured surface |
| Flow and pressure | Hydraulic delivery and detection of changed operating behaviour | Spray coverage, local soil removal or validated cleanliness |
Use the matrix to decide where measurements complement each other. A final-rinse conductivity endpoint may be informative while a residue-specific laboratory test remains necessary. Conversely, a laboratory result does not erase a process excursion: the deviation still requires assessment of representativeness, untested locations and the validated operating conditions.
Sensor location is part of the method
Document whether a sensor measures supply, bulk vessel contents, a specific branch or the common return. A common return can dilute a poorly rinsed branch with cleaner liquid from another route. The apparent endpoint may therefore depend on which paths are open. Route configuration, return volumes and mixing should be included in development and qualification.
Assess response time from the equipment to the sensor, including pipe volume, sample lines and processing delay. If the PLC stops rinsing immediately when a value crosses a threshold, it may be reacting to liquid that left the equipment earlier. Define the timing relationship explicitly. The same issue affects correlation between online readings and manually collected samples.
Design for verification and maintenance. Provide representative sample points where justified, safe access, hygienic installation and calibration arrangements that preserve the relevant measurement chain. Avoid creating a stagnant branch merely to obtain an easy sample. The sample point itself needs an assessed cleaning procedure and sampling method so its contamination does not masquerade as equipment residue.
Build endpoint logic from evidence
Endpoint development should consider the signal’s starting condition, expected direction, background variability and relationship to residues. Depending on the application, the rule may combine process conditions, a defined stability assessment and minimum qualified exposure. None of these elements has a universal numerical value. Derive them from development, analytical capability, equipment behaviour and the approved cleaning strategy.
Define how the logic handles spikes, oscillation, invalid signals and values already below the threshold at phase entry. A failed sensor that returns a low value must not create an instant pass. Assess whether filtering or averaging can hide meaningful excursions, and retain the data needed to understand the decision. Record the endpoint actually used, including its recipe version and relevant compensation settings.
Establish a maximum phase duration and the approved response when the endpoint is not reached where this is needed for the process. Repeatedly rinsing until the instrument passes can mask deteriorating cleaning or utility performance. Additional rinsing may be part of an approved recovery strategy, but it should not remove the original exception from the record or bypass investigation requirements.
Practical example: normal conductivity, unexpected product residue
An illustrative multiproduct vessel uses a conductivity endpoint for final rinse. After a product change, a validated residue-specific method detects unacceptable material at a difficult sampling location even though online conductivity was normal. The online result and the laboratory result are not necessarily in conflict: the product may have a weak conductivity response, and the return liquid may not represent the retained local residue.
The investigation reviews product solubility, dirty hold time, cleaning chemistry, local coverage, route configuration and sample recovery. It checks instrument validity and the timing of the return sample without assuming that the sensor is the primary cause. Development trials then compare residue results with process signals at relevant conditions and locations.
If the evidence shows that conductivity remains useful only for detergent rinse control, the strategy retains that role and preserves a separate product-residue assessment. A justified process change may improve local cleaning or modify the recipe. Its effectiveness is demonstrated with the relevant validation evidence. Simply lowering the conductivity threshold would not resolve a residue that the measurement cannot reliably detect.
Calibration, trending and data review
Calibration planning should address range, accuracy, uncertainty, installation and the significance of drift for the intended decision. Frequency depends on risk, performance history, manufacturer information and applicable requirements; it is not a universal calendar interval. When an instrument is found outside tolerance, assess affected cycles and the actual decision margin, rather than closing the event after adjustment alone.
Trend variables that can reveal loss of control: time to endpoint, rinse volume where measured, temperature response, concentration behaviour, flow-pressure relationships and recurring signal faults. Stratify comparisons by equipment route, product or soil family and recipe version. Combining unlike cycles can hide meaningful changes or create misleading alarms.
Separate an operational alert from a validated acceptance boundary. An alert can prompt investigation before a failure occurs, but it should have a defined purpose and response. Avoid moving alert levels merely to reduce the number of notifications. Review whether apparent improvement reflects better cleaning, a changed sensor setting, different incoming water or a modified data-processing rule.
Monitoring design checklist
- State the process decision supported by each measurement and the residues or failure mechanisms it can and cannot detect.
- Identify the sensor location, measured stream, route configuration, transport delay and relevant installation limitations.
- Characterise chemical response, temperature compensation, background and concentration range for the intended use.
- Demonstrate the relationship between endpoint logic and the cleaning strategy without assuming equivalence to laboratory acceptance.
- Define invalid-signal handling, phase timing, maximum duration, alarms and review of additional rinse or recovery actions.
- Verify calibration, data capture, recipe versioning and records needed to reconstruct the decision.
- Connect significant trends or changes in product, detergent, equipment and water to the appropriate impact assessment.
Common mistakes and key takeaways
Warning signs include using the water specification as the equipment residue limit, accepting a common-return result without considering branch dilution, and presenting a low TOC value as compound-specific proof. Other concerns are unexplained temperature compensation, an endpoint evaluated during invalid flow conditions and laboratory samples collected at a different process stage from the online comparison.
Reliable monitoring describes what the process delivered and supports timely decisions. Cleaning acceptance still requires the justified residue and verification strategy. Use conductivity, TOC, temperature, flow and pressure as complementary evidence with known limitations. Maintain the distinction between chemical cleanliness, microbial control and subsequent sterilisation, and revisit the measurement strategy whenever the process assumptions change.
Coordinate final-rinse quality at the equipment interface with Pharmaceutical Water & WFI Systems. Explore the surrounding equipment and project decisions through Pharma Engineering. These links support interface definition; a water-system specification does not automatically establish the acceptance criterion for a cleaned product-contact surface.
Official references
The principal references are EU GMP Annex 15 (2015), 21 CFR 211.67, and the FDA Cleaning Inspection Guide (July 1993). For US validation strategy, also consult the FDA equipment CGMP questions and answers, including the discussion of direct surface sampling when feasible. Official guidance and legal requirements have different status; the engineering recommendations here remain application-specific.
Related decisions
- CIP System Design for Pharmaceutical Equipment: Skids, Flow, Return and Cleaning Architecture
- Cleaning Validation in GMP Manufacturing: Swab, Rinse, Recovery and Acceptance Strategy
Explore all decisions in Cleaning, CIP & SIP Systems.
