A cooling tower controller rarely fails outright. Outright failure would be the easy case, because the alarm fires and somebody responds within the hour.
What happens instead is quieter. A conductivity probe fouls, the reading drifts low, and the controller stops bleeding a system that badly needs it. Everything on the HMI still looks normal.
This is the blind spot in most utility water automation. Plants deploy industrial automation products across the process side with full calibration discipline, then assume the tower controller takes care of itself. The instrument gets trusted precisely because it is automated.
The Failure Mode Nobody Alarms On
Standard controller alarms watch for values outside a configured range. High conductivity, low pH, and no-flow conditions all trigger correctly and get attention.
A drifting probe does not produce out-of-range values. It produces plausible ones, and plausible is what makes it dangerous.
A conductivity probe reading fifteen percent low keeps the controller comfortably inside its band while actual dissolved solids climb past the scaling threshold. Nothing alarms. Scale forms anyway.
By the time the consequence appears as approach temperature loss or a fouled exchanger, the drift has been running for months. The trend data looks reassuringly stable, because the instrument has been consistently wrong.
Why Tower Service Is Brutal on Probes
Cooling water is one of the harshest environments a sensor can occupy. It is warm, aerated, biologically active, and deliberately loaded with treatment chemicals.
Every mechanism that makes a tower work also attacks the instrumentation. Evaporation concentrates dissolved solids. Airborne debris enters through the fill. Biofilm colonizes any wetted surface it can reach.
The relevant point for controls engineers is that cooling tower chemistry does not stop at the heat transfer surface. The same calcium carbonate and silica deposits that insulate exchanger tubes also coat probe electrodes. At the same time, biocide residuals and corrosion inhibitor films alter the surface the sensor depends on to read accurately.
Conductivity Probes
Contacting conductivity sensors measure across a fixed cell constant. Any deposit on the electrodes changes that geometry and invalidates the constant.
Scale is an insulator. A thin carbonate film makes water look less conductive than it is, so the controller bleeds less than it should. The error compounds because reduced bleed accelerates further scaling.
Toroidal sensors resist this better because they do not rely on direct electrode contact. They are not immune, but the degradation is slower and more forgiving.
pH Probes
Glass pH electrodes have a finite service life measured in months rather than years. The reference junction plugs, the membrane ages, and response time lengthens.
A slow electrode still returns a number. It simply returns a stale one, which is useless for closed-loop control.
Check slope and offset at every calibration, not just the final corrected reading. An electrode that calibrates fine but shows a degraded slope is telling you it is near the end of its life.
A Verification Routine That Survives Reality
Weekly handheld comparison is the standard recommendation, and it gets skipped constantly because nobody makes it anyone’s explicit job.
Make it survivable. Assign the check to a named role rather than to a department, and give it a fixed slot in the shift routine instead of leaving it to available time.
Record three values every time: the handheld reading, the controller reading, and the difference between them. The difference is the number that matters, and it is the one most logs omit entirely.
Define a documented action threshold in advance. If deviation exceeds it, the probe gets cleaned and recalibrated before the shift ends, not at the next scheduled service visit.
Alarm Logic That Catches a Dead Instrument
Range alarms are necessary but insufficient. Add logic that watches instrument behavior rather than process value.
Rate of Change Limits
Real cooling water conductivity moves gradually. A jump of several hundred microsiemens within seconds is an electrical fault or a sensor problem, not a process event.
Configure a rate limit and alarm on violation. This catches wiring faults and failing transmitters immediately instead of weeks later.
Frozen Value Detection
A probe reading precisely the same value for hours is not stable. It is stuck.
Alarm on any value that has not moved beyond a small deadband across a defined window. Genuine process noise always produces slight variation, so its complete absence is diagnostic.
Cross-Parameter Sanity Checks
Conductivity and makeup water flow should correlate. If makeup runs continuously while conductivity holds perfectly flat, one of those two instruments is lying to you.
Simple comparison logic in the PLC catches disagreements that no single-parameter alarm will ever detect. Boiler and closed loop systems benefit from the same treatment.
Treat Calibration Records as Process Data
Most calibration logs live on a clipboard in the mechanical room. That is a waste of genuinely useful information.
Deviation history is a leading indicator of instrument failure. A probe needing a two percent correction last quarter and eight percent this quarter is dying, and the trend says so well before the failure does.
Push those values into the historian alongside process data. Then instrument health becomes reviewable remotely, and replacement becomes something you schedule rather than something you react to.
Verify on a Schedule, Not on Suspicion
Utility water automation earns trust the same way process automation does, through verification. The instrument is not the measurement. It is a device producing a number, and that number is only as reliable as the last time somebody proved it.
Build verification into the routine and the automation becomes genuinely dependable. Skip it, and you own an expensive system making confident decisions on bad data. For broader practice guidance on cooling water programs and operator training, the industry association resources published by the Association of Water Technologies are a useful reference.