How to Run a Positioner Autocal Safely on a Live Process

Back to Blog

Gather What You Need and Who You Need

You need the positioner's configuration access, either its local interface or a handheld communicator, and you need to know the valve's fail action and its role in the process before anything else. A conversation with the board operator matters more than any tool here: the routine will drive the valve to both ends of travel, and the person running the process has to agree that is acceptable, either because the process can ride through it or because flow will be carried around the valve on a bypass during the work.

Have the loop drawing or a current valve lineup checklist if a bypass changeover is involved, and know where the air supply isolation for the actuator is. The mechanics of the calibration itself, feedback linkage checks, what the routine learns, and the manual zero and span fallback, are covered in the guide to calibrating a valve positioner; this page is about doing that work safely while the plant is running.

Confirm the Process Can Tolerate a Full Stroke

Ask the question in plain terms: what happens downstream if this valve goes fully closed for a minute, and what happens if it goes fully open? For some services the honest answer is nothing much, a level rides up a little, a flow dips and recovers. For others, full closed means a compressor surges, a pump deadheads, or a low-flow interlock trips the unit, and full open means overpressure or a flooded vessel. The autocal does not care which case you are in, so you have to.

If the answer is that the process cannot tolerate either extreme, you have three honest options: carry the flow on a manual bypass while the control valve is isolated, defer the calibration to a planned outage, or accept a limited manual calibration if the positioner supports one. What you must not do is launch the routine hoping it finishes before anything drifts too far. A smart valve positioner mid-autocal ignores the control signal entirely, so the controller cannot rescue the process while the routine runs.

Line Up the Bypass and Isolate the Valve

Where a block-and-bypass arrangement exists, use it. Put the controller in manual, have an operator take the process on the bypass globe by opening it gradually while the upstream and downstream block valves around the control valve are closed, and confirm the process variable is being held steadily by hand before you touch the positioner. The control valve is now yours: it can stroke through its full travel without the process feeling it.

With the valve isolated, treat the stroke itself as the remaining hazard. The valve will move hard and without warning when the routine commands it, so keep fingers away from the stem, the feedback linkage, and any exposed pinch point between the yoke and the travel indicator, and warn anyone working nearby. Venting actuators are loud, and on large actuators the exhaust blast is significant. If the valve has a handwheel or a mechanical travel stop, confirm it is backed off to neutral first, because a restricted stroke will either fail the autocal or teach the positioner false endpoints.

Run the Autocal and Watch the Whole Stroke

Launch the routine and stay with the valve, not the display. Watching the physical stroke tells you things the positioner cannot: a stem that hesitates mid-travel, a linkage that flexes, an actuator that labors near the seat. The routine will drive to one stop, pause, drive to the other, and may make several smaller moves while it learns the actuator's response. Let it finish; interrupting an autocal can leave the positioner with half-learned endpoints that are worse than what you started with.

If the routine aborts, resist the urge to simply run it again. A failed autocal usually has a physical cause, low or unstable air supply, feedback out of range, or the valve never reaching a firm stop, and rerunning it without fixing that wastes the isolation window you negotiated. Diagnose, correct, then rerun once.

Return the Valve to Service and Verify

Before handing the valve back, verify the calibration the same way you would in a control valve loop check: command a few known percentages from the control system and confirm the physical travel and the position feedback both agree. Then reverse the bypass changeover slowly, setting the control valve near the position that matches current flow in manual, opening the blocks, closing the bypass gradually while the operator watches the process variable, and only then returning the controller to auto.

Tell the control room the work is complete and what changed. If the positioner learned new endpoints, the valve may sit at a slightly different position for the same signal than it did yesterday, and an operator who knows that will not chase a phantom process change. Log the as-left travel readings so the next technician has a baseline.

Common Mistakes

The classic failure is running the routine on a live valve because it worked fine last time on a different valve. Every service is its own case, and the only person who can clear a full stroke is the one operating the process. A close second is leaving the controller in auto during the work: the controller winds up its output while the positioner ignores it, and the moment the autocal releases the valve, the wound-up output slams it to an extreme.

The quieter mistakes surface later. A handwheel left a few turns engaged teaches the positioner a short stroke that shows up as a valve that never fully closes. Skipping the post-calibration verification leaves a reversed or offset feedback undiscovered until an operator acts on it. And failing to log the work means the next investigation starts from nothing.

1. Operations agreed the valve may stroke fully, or the bypass is carrying flow
2. Controller in manual, control room notified
3. Handwheel and travel stops backed off to neutral
4. Air supply healthy and stable
5. Hands clear of stem and linkage, area warned
6. Verify travel and feedback before returning to auto

Six lines on a card cover most of what goes wrong in practice; the checklist is short because the failures are repetitive.

Frequently Asked Questions