Tuning Practice 5 — Flow — control valve in line

The job

What you are looking at

A control valve sits in the line with a flow transmitter downstream of it. Open the valve, more flow. The disturbance is header pressure upstream.

Two things make this loop different from the level loops. It is fast — the valve strokes in a fraction of a second and there is almost no dead time. And the flow measurement carries real turbulent noise, so the trend will never be a clean line.

That fuzz on the trend is real, and it is supposed to be there. A flow measurement is genuinely noisy. The liquid is turbulent, and the transmitter reports what it actually sees, moment to moment. So the flow line will never be a smooth curve the way the level and temperature loops were — it will always look slightly hairy, even when the loop is tuned well and sitting right on setpoint. Nothing is broken, your display is not glitching, and it is not something you can tune away.

Watch what it does to the controller output. Proportional passes that fuzz straight through to the valve. Rate — the derivative term — responds to how fast the measurement is changing, and noise changes fast, so rate multiplies it. Turn rate up on a noisy flow loop and watch the output trace before you decide whether rate belongs here.

The controller is on factory defaults and the acting direction has not been commissioned. All three tuning terms are unlocked on this one, including rate.

This is practice, not a test. Nothing on this page is scored and nothing is submitted anywhere. Judge your tuning by the trend in front of you: how quickly the measurement reaches the setpoint, how far past it goes, whether it settles, and how the loop behaves when the load changes. Change something, watch it, change it again.

How to work through it

Steps

Press Start. The loop begins balanced at setpoint, so nothing will move until you disturb it — move the setpoint, or change the load, and watch which way the measurement goes.
Set the acting direction for this loop.
Tune it until the flow holds setpoint. Move the setpoint and change the Header pressure to test it.
Once it is running well, add some rate (the derivative term) and watch what happens to the controller output trace. Then decide for yourself whether rate belongs on this loop, and set it accordingly.
Before you call it tuned, change the load and confirm the loop brings the measurement back and holds it. A loop that only looks good on a setpoint move is not proven.

All three tuning terms are available on this exercise, including the derivative (rate) term.

Adjustment range available for each tuning term on this exercise
Tuning term	Lowest	Highest
Proportional gain (Kp)	0	2
Integral gain (Ki, 1/s)	0	4
Derivative gain (Kd, s)	0	0.15

These ranges are set for this loop only. Another exercise will have different ones, because a different process needs different numbers.

Simulator

For your own record

Commissioning notes

Nothing here is graded or sent anywhere. Write down what you settled on and why, so you have it to compare against the next exercise.

Which way did this controller have to act, and how did you work that out? What did you leave the tuning set to? Record the numbers themselves, in whichever form the controller was showing them. What did the loop do when you changed the load, and were you satisfied with the recovery?

Your notes stay in this browser, on this computer, and they belong to this exercise alone — each exercise keeps its own set. Open this exercise again on this computer and they will be here waiting for you. They are not sent anywhere and they are not graded, and they will not follow you to another machine, so copy anything you want to keep elsewhere.