Valve position oscillates but the process variable barely moves
This is a diagnostic fingerprint in its own right. A controller output that swings while the process variable stays roughly flat means the loop is expending effort that is not reaching the process — and that narrows the field sharply.
It also means wear. Every one of those output swings is real valve travel, packing friction, and positioner air consumption, bought for no control benefit. Loops in this state are the ones that fail between turnarounds.
Work out whether the process is deaf or the valve is
Two very different situations produce this picture. Either the valve is moving and the process is not responding, or the valve is not actually moving at all despite the signal telling it to.
If you have valve position feedback, this takes ten seconds: compare commanded position against actual position. Without feedback, put the loop in manual and make a series of small output changes in the same direction. If the process only responds every third or fourth change, the valve is not moving on the small ones.
If the valve genuinely is moving
A valve that moves without moving the process is oversized, sitting against a seat, or operating where there is very little pressure drop across it to work with. Check the normal operating position: a valve wandering around a few percent open has almost no authority over the process.
The other possibility is that something downstream is dominating. A parallel path, a recycle line, or a downstream restriction can leave the control valve with far less influence than the P&ID suggests.
What causes it, most likely first
Derivative amplifying measurement noise
most common tuning can fix this
The output is chasing noise, not process movement. Because it is chasing noise, the process quite correctly does not respond.
- How to confirm it
- The output is noisier than the measurement. Activity is fast and irregular rather than a regular cycle.
- What to do
- Set rate to zero and watch the output activity drop.
Stiction — the valve is not following the signal
most common tuning cannot fix this
The output swings because the controller is winding up against a stem that will not move. The process does not respond because nothing has actually happened at the valve.
- How to confirm it
- Compare commanded and actual valve position. Or make small output changes in manual and see how many produce no process response at all.
- What to do
- Check packing friction and the positioner.
Oversized valve near its seat
common tuning cannot fix this
Close to the seat a valve has poor resolution, and small movements do very little. The controller has to move a long way to achieve anything.
- How to confirm it
- Normal valve position below about 20% open.
- What to do
- Reduced trim. Detune in the meantime.
Heavy measurement filtering
less common tuning cannot fix this
The process may be responding perfectly well, but the transmitter damping or controller filter is smoothing the response out of the trend before you see it.
- How to confirm it
- Check transmitter damping and the controller PV filter. If they total more than the loop's own response time, the filter is the problem.
- What to do
- Find out why damping was raised. It is usually masking noise that should have been fixed at source.
Questions that come up
Is this doing any harm if the process is staying on setpoint?
Yes. The process is staying on setpoint despite the loop, not because of it, and you are buying valve wear and air consumption for nothing. A loop in this state also has no capacity left to reject a real disturbance.
Related
Last reviewed 2026-08-01.