A low-level dissolved oxygen measurement on a glycol-based process stream began showing unstable readings with occasional spikes. Calibration checks remained repeatable, but the online signal did not reflect the expected steady process condition.
This created uncertainty about whether the issue was sensor related or sample related.
Observed symptoms
The reading became unstable after pressure reduction and could rise sharply after interruptions in flow. When flow restarted, trapped gas behaviour in the cell appeared to influence the measured result.
The pattern suggested that the analyser was responding to changing sample presentation rather than to true dissolved oxygen changes alone.
Investigation direction
Review of the sample arrangement focused on pressure reduction, degassing potential, and whether gas pockets could accumulate in the cell. Independent reference checks supported the view that the bulk process oxygen was more stable than the online trace suggested.
This shifted attention away from simple sensor failure.
Likely cause
The main issue was consistent with gas release and trapping in the sample cell, particularly after pressure reduction and temporary flow disturbance. Once gas collected in the wrong location, the reading became artificially high or unstable.
The sample environment was driving the instability.
Practical lesson
Low-level oxygen systems require careful attention to cell orientation, venting, pressure management, and restart behaviour. Even when calibration is healthy, sample presentation can dominate the reading.
The lesson is simple: stable chemistry still needs stable hydraulics.