A Control Valve Can Pass Its Positioner Calibration and Still Be Controlling Flow Nothing Like Its Design Curve

Every dosing, recycle, and bypass control valve at an ETP or STP gets its positioner calibrated on a routine schedule, a stroke check confirming the valve moves to the commanded position at each test point across its input signal range, typically five points from fully closed to fully open. That check is real and it catches a genuine class of fault, a drifting positioner, a sticking actuator, a feedback linkage that has worn loose. It does not confirm the thing a control loop actually depends on the valve for, because how much flow passes at a given stem position is governed by the valve's installed flow characteristic, and that characteristic can drift substantially from design while the valve continues to stroke to every commanded position exactly as a calibration check expects.

Moving correctly and flowing correctly are different claims

A positioner calibration answers: does the valve stem reach the position the input signal commands, at each test point. It is fundamentally a signal-and-mechanism check, confirming the positioner translates a 4 to 20 mA input into the correct percentage of travel. It says nothing about flow, because flow at a given stem position depends on the valve's trim geometry and the pressure conditions across it, a relationship that can change independently of whether the stem physically reaches the commanded point. Trim wear, erosion from high-velocity or abrasive flow, and cavitation damage from vapor bubbles collapsing against the trim surface, all alter the valve's actual flow-versus-position curve without necessarily preventing the stem from moving freely across its full range. A valve can pass every point on a stroke check while its installed characteristic has drifted far enough from the design curve that the same commanded position now delivers a meaningfully different flow than the control loop assumes.

Why a passing calibration and a misbehaving loop can coexist for months

Trim damage is specifically the kind of fault that hides behind a passing stroke check, because erosion and cavitation attack the flow passage geometry, not the stem's freedom of movement. A plug or seat eroded enough to change the valve's gain across its travel range can still move cleanly from 0 to 100 percent on command, so a calibration technician running the standard five-point check sees nothing wrong, records a pass, and moves on. The symptom that actually shows up is downstream of the valve entirely: a control loop that hunts, overshoots, or responds sluggishly at parts of its range, a pattern that gets diagnosed as a tuning problem because that is the most visible and most commonly adjusted variable, when the valve's own characteristic, not the controller's gain settings, is what has actually changed. Retuning a loop to compensate for a shifted valve characteristic can mask the symptom at one operating point while making it worse at another, because the valve's gain is no longer consistent across its stroke the way the original tuning assumed.

This is the same shape of mistake we described in the piece on flow meter accuracy at low flow: an instrument's calibration or verification check can be entirely correct against its own narrow test and still miss the condition that actually governs real-world performance, because the check was built to confirm one thing and the plant is relying on it to vouch for something broader. There, calibration said nothing about accuracy outside the tested flow range. Here, a stroke check confirms the stem reaches each commanded position and says nothing about whether the flow delivered at that position still matches the curve the control loop was tuned against. It is also a close cousin of the piece on online analyzer sample transport lag: both involve a control loop component that passes its own routine verification while a separate, unmeasured factor, there sample delay, here valve characteristic drift, quietly determines whether the loop actually behaves the way its tuning assumes.

Picture a methanol dosing valve on a denitrification loop, passing its quarterly five-point stroke check without issue for over a year. Over that same period, the valve's trim has eroded under sustained high-velocity flow, shifting its installed characteristic so the first third of its travel now delivers proportionally more flow than the design curve specifies. The dosing loop, tuned against the original characteristic, begins hunting at low dose rates, a pattern the operations team retunes around repeatedly without lasting success, because the controller is being asked to compensate for a valve whose actual gain has changed shape, not for a tuning error. The stroke check keeps passing throughout, because the valve keeps reaching every commanded position exactly as before. Nothing in the routine calibration record ever shows that the relationship between position and flow is no longer what the loop was tuned around.

What actually confirming valve performance requires

The fix is not abandoning the positioner stroke check, which remains the correct way to catch a mechanical or signal-handling fault. It is adding a periodic comparison of the valve's actual flow-versus-position behavior against its baseline or design characteristic, using process data already available on most loops, flow or level response at known valve positions, rather than relying on stroke position alone as proof of flow performance. When a loop that passed its last calibration begins hunting or responding inconsistently across its range, that pattern is itself a diagnostic signal worth checking against the valve before retuning is attempted, since a first instinct to adjust the controller gain treats the symptom rather than the valve-side cause when trim wear is actually responsible.

A simple field test separates the two causes before any retuning is attempted: switch the controller to manual and step the output in small increments. If the hunting or instability stops under manual stepping, the loop's own tuning is the genuine cause and retuning is the right fix. If the same instability persists under manual steps at a given position, the valve itself, not the controller, is where the fault lives, and no amount of retuning will resolve it. Running that test before the first retuning attempt, rather than after several rounds of unsuccessful tuning adjustments, is the difference between catching a worn valve early and spending weeks chasing a tuning fix for a mechanical problem.

Misdiagnosing a worn valve as a tuning problem has a real cost beyond the wasted retuning cycles: a dosing loop that hunts around its target consumes more chemical than a stable one, and the chemical and operating cost of that overdosing accumulates quietly for as long as the real cause goes unidentified. If your plant's control valves are verified by positioner calibration alone, with no periodic check on whether the actual flow characteristic still matches design, that gap is worth closing with Spans' instrumentation team before another round of loop retuning tries to fix a problem that was never in the controller.