Application Note

Why 'Follow the Manual' Is No Longer Enough for Instrument Reliability

Posted 2026-11-13 by Noura El-Sayed

The manual tells you what the instrument can do. It doesn't tell you what your plant will do to the instrument.

I've reviewed more than 200 equipment installation and commissioning packages over the past four years. About 18% of first submissions get rejected—not because the hardware failed, but because the paperwork tells me nobody verified whether the installation matches the actual process conditions.

That's the gap. Manuals are written for controlled conditions. Plants aren't controlled. They're negotiated.

I know that sounds harsh. But after four years of walking through installations that passed every checkbox audit yet still drifted, I've stopped trusting checklists. The manual is a starting point. It's not the finish line.

What the book doesn't say about piping stress

Take Coriolis flow meters. The Endress+Hauser flow meter installation guide is thorough—it warns against pipe stress, vibration, entrained gas. It even shows you diagrams of acceptable support spacing. But here's what it doesn't tell you: what "stress" looks like in your specific piping layout, at your specific operating temperature, after your contractor made field modifications to your original design.

We installed a batch of Coriolis meters on a custody-transfer skid last year. The commissioning report said "installation per manufacturer guidelines." Everything looked clean on paper. Six weeks later, three meters were drifting between 0.8% and 1.2% off their reference. Not enough to trigger an alarm. Enough to lose traceability on four batches.

I said, "Check the installation." The vendor heard, "Check the instrument." Result: two weeks of back-and-forth before we found the real problem.

The pipe supports had been field-modified to accommodate a valve package. The modification introduced a fractional misalignment—less than half a degree—that translated into bending stress at the meter flange. The instrument was fine. The installation was "per the manual." The measurement was wrong anyway.

Nothing in the manual says "check for contractor skid modifications that happened after your piping design was frozen." Why would it? The manual assumes the piping design is correct. That assumption never survives contact with an actual construction schedule.

The accuracy illusion spans instrument types

This isn't unique to flow meters. The same pattern shows up anywhere precision matters.

A confocal microscope can resolve features down to roughly 120 nm—according to the spec sheet. But if your sample prep leaves refractive index mismatches, or the immersion oil isn't matched to the objective, you won't see 120 nm resolution. You'll see a very sharp image of the wrong thing. The microscope did its job. The preparation didn't.

A FLIR T865 thermal camera has a stated accuracy of ±1°C or ±1%, whichever is greater. Set the emissivity wrong—say you leave it at 0.95 for a polished aluminum surface that's actually closer to 0.1—and your "accurate" reading is off by tens of degrees. The camera doesn't flag this. It just does what you told it.

Same story with a pipette. The technique guide for how to use an Eppendorf pipette covers aspirating angle, immersion depth, pace. Hold it at 45 degrees instead of vertical and the calibration tolerance means nothing. The error isn't in the instrument. It's in the wrist.

It's tempting to think that a precision instrument delivers precision results automatically. It doesn't. It delivers a precise input. What happens after that is on you.

Annual calibration stopped being a strategy a while ago

Ten years ago, an annual calibration cycle made sense. Production runs were long, processes were stable, and a drifting sensor would usually show up in quality data before it became a traceability problem.

That calculus has changed. Smaller batches. Faster changeovers. More variants. The window between "sensor drifts" and "product ships" has compressed to the point where an annual calibration schedule catches the drift months after it started.

I didn't fully understand this until a pH transmitter drifted 0.3 units over a single weekend. The calibration sticker said it was good for another eight months. The batch records disagreed. We quarantined two lots—roughly $14,000 in material and a week of production time.

The transmitter wasn't faulty. It was within its stated drift specification. The problem was our calibration interval, which assumed a stability profile our process didn't actually have.

This was true in 2015, when most of us still planned around quarterly campaigns and stable recipes. Today, a calibration schedule that doesn't account for your actual process variability is just a piece of paper with a date on it.

"We've always done it this way"

The pushback I hear most often goes something like this: "We've been installing these for 15 years and never had a problem worth mentioning."

Fair point. But I'd push back on it—gently. The absence of a problem isn't evidence of a good process. It's evidence that your process has enough buffer to absorb the errors you haven't found yet. That buffer gets thinner every time you tighten a tolerance or shrink a batch size.

The second objection: "The vendor signed off on it." Vendors check against their manual. They don't check against your process. Those are different standards. Only one of them affects your product.

What I check now that I didn't before

I run three questions on every commissioning package now. They're not in any manual.

One: does the installation match the process conditions, not just the book? A flow meter that meets straight-run requirements on paper can still fail if an upstream valve creates asymmetric flow at your minimum turndown. The manual assumes developed flow. Your process might not deliver it.

Two: how will we know if it drifts? If the answer is "annual calibration," that's not a monitoring plan. That's a hope with a sticker on it. Modern transmitters can often flag their own drift. If you're not using that feature, why did you pay for it?

Three: who checks the human factor? A perfectly installed instrument operated at the wrong angle, with the wrong density setting, by someone who wasn't trained on it—still gives you a wrong answer. And it does it with confidence.

The fundamentals haven't changed. Everything around them has.

None of this means the manuals are wrong. They're not. Endress+Hauser's installation documentation is genuinely good. FLIR's thermal imaging guides are detailed. Eppendorf's pipette technique resources are solid. The problem isn't the documentation quality. The problem is treating the documentation as the entire answer.

Five years ago, specifying a high-accuracy instrument and following the manual was a reasonable recipe for reliable measurement. In most standard applications, it still is. Today, for tighter tolerances and faster cycles, the instrument is one link in a longer chain—installation, commissioning, operation, monitoring, and maintenance all have to hold.

So read the manual. Then walk the line. Look at the supports. Check the settings against the process, not just against the default. Ask the operator how they actually use it—not how they're supposed to.

The manual tells you what the instrument can do. Your job is to make sure the plant lets it.

Share on WhatsApp
Noura El-Sayed

Noura El-Sayed

Noura El-Sayed is an analytical and laboratory instrumentation analyst specializing in HPLC systems and columns, spectrophotometers, pH meters, pipettes, centrifuges, viscometers, and sample filters. She applies ISO 8655 liquid-handling methods and USP <621> chromatography principles while evaluating delivered volume, precision, flow accuracy, resolution, carryover, and detection limits. Her work helps laboratory managers and method developers select instruments, qualify workflows, and control sources of variability from sample preparation through analysis.

Leave a Reply