The Valve in the Control Loop

Why the valve, not the controller, decides how well most loops perform.

11 min

Activities: 0 of 2

The only part that touches the process

A control loop has three parts. The measurement senses the process. The controller compares it with the set point and calculates an output. The final control element — usually a control valve — changes the process. Of the three, only the valve does physical work on the fluid. The transmitter and the controller are electronic and very precise. The valve is mechanical, carries friction and wear, and handles high pressure, erosion and temperature.

This is why studies of process plants repeatedly find that a large share of poorly performing loops have a valve problem at their root. A controller can only be as good as the valve it drives. If the valve does not move when asked, or moves too much, no tuning can fully fix it.

One cycle of a flow loop

  1. 1

    1

    Measure

    The flow transmitter reports 80 cubic metres per hour.

  2. 2

    2

    Compare

    The set point is 100. The controller sees an error of 20.

  3. 3

    3

    Calculate

    The controller raises its output from 45 to 52 per cent.

  4. 4

    4

    Signal

    A 4 to 20 milliamp (mA) signal goes to the valve positioner.

  5. 5

    5

    Move

    The positioner sends air to the actuator; the stem moves; the plug opens.

  6. 6

    6

    Respond

    Flow rises. The transmitter sees the change, and the cycle repeats.

Where the valve can break the loop

Each step after "signal" depends on the valve assembly. If friction is high, the stem does not move until the force builds up, then it jumps too far. If the actuator is slow, the controller sees no response and keeps pushing. If the valve is oversized, a tiny movement changes flow a lot. Each problem looks, on the trend, like a badly tuned controller.

Case: the loop that was "retuned" five times

A steam flow loop on a heater cycled up and down every two minutes. Over a year, it was retuned five times. Each new tuning made it slower but did not stop the cycling. A valve test finally showed that the valve did not respond to changes smaller than 3 per cent, then jumped. The packing was over-tightened. After repacking, the original tuning worked.

Case: the valve that was right

A pressure loop was sluggish. A valve test showed the valve responded well to small steps. The problem was in the controller: its integral time had been set ten times too long. Testing the valve first ruled it out in an hour and pointed to the real cause.

A loop cycles steadily. In manual, the valve does not move for a 1 per cent step, then jumps on the next step. What is the likely root cause?

Order the loop

Put the signal path in order, from measurement to process change.

  1. 1Trim changes the flow
  2. 2Actuator moves the stem
  3. 3Positioner receives the signal
  4. 4Transmitter measures the process
  5. 5Controller calculates its output
1 of 30

Checking your enrolment…