PID Loop Oscillates After Valve Replacement
A temperature control loop uses a PID in a PLC driving a modulating valve on a steam line. The loop was tuned two years ago with a 4–20 mA valve actuator. The actuator was replaced with a new unit that has the same 4–20 mA input but a stated 12 s stroke time versus 30 s for the old one, and a deadband of 0.5 % versus 1 %. After replacement the process temperature cycles with a period of about 90 s and an amplitude of 6 °C around setpoint. The PID gains were not changed. The engineer proposes halving the proportional gain and leaving integral and derivative alone.
1. Explain why the loop now oscillates and evaluate the proposed single-parameter change.
Read the worked answer
The new actuator responds about 2.5 times faster and with a smaller deadband, so the effective process gain and phase lag seen by the controller have changed: the loop now has more gain at frequencies where the old tuning had margin, pushing it toward instability, which appears as a sustained 90 s cycle. Halving proportional gain alone may reduce amplitude but does not address the integral term, which is a common source of slow cycling, and it will slow disturbance rejection. A defensible approach is to retune from measured response: perform a step test on the valve output, record the process reaction curve, and derive gains, or run a relay autotune if the PLC supports it. Check derivative action too, since a faster actuator can make derivative noisy.
2. List the records and checks you would complete before and after retuning.
Read the worked answer
Before: confirm the new actuator's input scaling and fail position, verify the 4–20 mA loop calibration, record the old and new stroke times, and note the process load and ambient conditions. Confirm the PID sample time is appropriate for the faster actuator. After: record new gains, capture a trend of setpoint step and load disturbance responses, verify overshoot and settling time against the process requirement, and check valve output for hunting or saturation. File the tuning record with the step-test data so the next change starts from evidence rather than guesswork.