PIDSnap

Tuning flow control loops

Flow is the easiest loop type to work on and the one most consistently tuned badly. Easy, because it settles in seconds — a bump test takes a minute and a change can be judged almost immediately. Badly, because its natural tuning looks wrong to anyone whose instincts came from slower loops.

Derivative action amplifying measurement noise Trend showing noise amplification: a modest amount of noise on the process variable appears on the controller output magnified many times over. high low PV SP Output PIDSnap time
Trend showing noise amplification: a modest amount of noise on the process variable appears on the controller output magnified many times over. The output is noisier than the measurement, not quieter. Every wiggle on the process variable appears larger on the output. The valve is being asked to chase turbulence. On a flow, pressure or speed loop this is almost always derivative action that should not be there.

What makes a flow loop different

A flow loop has almost no lag. Open the valve and the flow changes essentially at once — the dead time is transport delay through a short length of pipe plus the transmitter's own response, typically well under a second to a few seconds.

It also has genuine measurement noise. Turbulence past the flow element is real, physical, and not going away. The signal will never be smooth, and a controller that tries to correct every wiggle will simply wear the valve out.

Those two facts drive everything else. Low lag means high process gain, which wants low controller gain. Fast response means integral action can be fast. Real noise means derivative action is actively harmful.

Structure: PI, and only PI

There is essentially no flow loop that benefits from derivative action. Derivative acts on rate of change, and turbulent noise has a large rate of change even when the flow is steady, so the derivative term multiplies noise and writes it to the valve.

If you inherit a flow loop with rate switched on, take it out before you assess anything else. It is the highest-value single change available on this loop type, and PIDSnap's pre-flight check flags it as critical precisely because it is so consistently wrong.

What usually goes wrong

  • Derivative left switched on, amplifying noise into the valve.
  • Gain carried across from a temperature loop, producing violent instability.
  • Reset set faster than the loop's dead time, giving a cycle no gain change will remove.
  • Equal-percentage trim on a wide-range service, so the process gain varies with load and no single tuning set holds.
  • Heavy transmitter damping added to hide noise, which quietly caps how well the loop can ever perform.

Questions that come up

Why does my flow loop need such a low gain?

Because the process gain is high. With very little lag, a small valve movement produces a large and immediate flow change, so the controller has to be correspondingly gentle. Loop stability depends roughly on the product of process gain and controller gain, and on a flow loop the process is already contributing most of it.

Should a flow loop ever be proportional-only?

Almost never. Without integral action it will sit with a permanent offset that changes with load. The main exception is a flow loop used purely as the inner loop of a cascade where the outer loop provides the integral action, and even then it is unusual.

Related

Last reviewed 2026-08-01.

Run a free PIDSnap check on this loop

Two photos, thirty seconds, no signup. The pre-flight check finds the structural problems that no amount of tuning will fix — before you touch the controller.

Start a free check