Fixed Schedule or Condition-Based? Lessons from ifm Sensors, Thermocouple Calibrators, and Agilent HPLC Columns

Should you replace a component because the calendar says so, or because the evidence says so? That's the question I get whenever I talk about maintenance and calibration. And for good reason: it affects your total cost, your data quality, and your risk of a big failure.

I'm a quality and brand compliance manager at ifm. I review product documentation before it reaches customers—roughly 200 documents a year. I've rejected my share of ambiguous datasheets. The most frustrating part is not the documents, though. It's watching plants rely on a 'replace it anyway' policy and then claim they have a quality system.

For example, when I read the ifm OGT500 sensor specs, I don't just look at range, output, and protection rating. I look at whether the sensor can tell you what's happening before it fails. That's the key difference between a component that supports condition-based maintenance and one that only supports a guess.

The Two Approaches: A Calendar vs. Evidence

The first approach is time-based replacement. Change the thermocouple every 12 months. Replace the HPLC column every three months. Recalibrate the pressure transmitter every six months. It's simple, auditable, and schedulable.

The second approach is condition-based replacement. You monitor actual performance indicators: sensor signal quality, calibration drift, column backpressure, peak shape, loop output. You replace when a threshold is crossed, not when a date arrives.

It's tempting to think the first one is safer. But that simple rule ignores how different failure modes actually work.

Cost: The Bill You Don't See on the Purchase Order

Let's use total cost thinking, not unit price. The cost of a replacement isn't just the price of the new part. It's labor, downtime, installation error, disposal, and the risk that you replaced something that was still perfectly good.

In the lab version of the question, 'How often to change your columns HPLC Agilent?' is a classic. If you set a conservative date and swap out an Agilent HPLC column that still meets system suitability, you've spent hundreds of dollars plus hours of re-equilibration for zero quality gain. Do that across twenty columns and the total cost is real. The 'cheap' schedule isn't cheap.

Condition-based doesn't mean 'no schedule.' I do not mean that. It means the schedule is driven by data: plate count, tailing factor, backpressure, retention time. When those move outside your method's limits, that's the trigger. The surprise isn't that columns last longer than people think. The surprise is how many quality managers don't track the data in the first place.

Risk: Which Approach Misses More Failures?

Fixed schedules are good at catching slow, predictable wear. They're terrible at catching random events. A photoelectric sensor can get a scratched lens or a bent bracket tomorrow, no matter what the maintenance calendar says.

Condition-based monitoring catches gradual drift. With a thermocouple calibrator, you can verify whether a temperature loop is accurate before it causes an out-of-spec batch. With a real instruments multimeter, you can check a 4-20 mA signal and see whether the transmitter is drifting. That's how you catch the small problem before it becomes a $22,000 redo.

But here's the nuance. For safety functions, condition-based monitoring does not replace mandatory proof testing. If you have an ifm safety sensor in a SIL-rated safety function, the safety manual and standards like IEC 61508/61511 define test intervals. You can't skip the manual proof test just because IO-Link diagnostics look clean. Diagnostic coverage can extend intervals, but it doesn't eliminate them.

People think replacing more often causes higher reliability. Actually, unnecessary replacement is a common cause of failures. Every time you open a connector, recalibrate a loop, or change a column, you introduce a new chance for installation error. The safest number of interventions is the smallest number that keeps the process in control—not the largest number your budget allows.

Data and Tools: You Can't Manage What You Don't Measure

Condition-based replacement depends on measurement quality. This is where a lot of programs fail. They collect data from uncalibrated tools and wonder why the numbers don't make sense.

If you're monitoring temperature loops, a thermocouple calibrator with cold-junction compensation is a must. If you're checking analog outputs, use a proper instruments multimeter with a valid calibration certificate—not a $10 meter from the hardware drawer. The measurement tool is part of the total cost, but it's the part that pays you back with every avoidable shutdown.

The ifm OGT500 sensor specs are a good example of a component designed for condition-based use. It's a diffuse-reflection photoelectric sensor with an M12 connector and IO-Link capability. In practical terms, IO-Link gives you access to signal quality, contamination information, and temperature data. You don't have to wait for the sensor to stop switching; you can see the signal reserve dropping and clean the optics first. That's the difference between maintaining a component and merely replacing it.

People: Why a Good Procedure Can Still Cause Bad Results

A fixed schedule creates a checkbox mindset. 'I replaced it, so now we're safe.' But the most frustrating part is seeing a team replace a good part on Monday because it was due, then miss a bad part on Tuesday because it wasn't due. The calendar is not a proxy for condition.

Condition-based maintenance requires a different kind of discipline: record the readings, understand the limits, and act on evidence. It's harder to delegate. But it usually produces more engaged technicians, because they're making decisions instead of following a list.

It's also worth acknowledging that condition-based isn't always the answer. For low-cost, high-consequence components that fail unpredictably, or for regulatory proof tests, a fixed schedule is still the right call. The goal isn't to be clever. It's to match the maintenance strategy to the failure mode.

What to Do: A Practical Selection Guide

Here's how I decide, and I use this for everything from ifm sensors to Agilent HPLC columns.

• Use condition-based replacement when: the failure mode is gradual, the replacement cost is meaningful, and you can measure the right parameter. Examples: IO-Link photoelectric sensors, pressure transmitters, HPLC columns with daily system suitability checks, thermocouple loops with a calibrator.
• Use fixed-schedule replacement when: the item is cheap, the failure is random or dangerous, or a standard requires it. Examples: safety sensors with defined proof-test intervals, gaskets and o-rings, regulatory calibration of traceable instruments.

For safety sensors, the question isn't 'how often should I replace an ifm safety sensor?' It's 'what does the safety manual require for proof testing and replacement?' Respect that, then use diagnostics to optimize availability.

And if someone asks you 'how often to change your columns HPLC Agilent?' answer with your own question: 'How are the column performance metrics today?' If you don't know, fix that before you worry about the date.

The Bottom Line

I don't care whether you call it predictive, preventive, or condition-based. I care whether you can explain why you changed something. 'Because it was due' isn't a reason—it's a guess with a date label.

The best time to change a component is when the evidence says it's no longer capable of doing its job. If you can measure that, do it. If you can't, a conservative schedule is better than no plan at all. But measure what you can, and let the data argue for itself.