INST-014 · Control-loop fundamentals · Foundation
Recognize disturbances versus commanded changes
By the end, you can: Separate changes in manipulated inputs from load disturbances and reject process-gain estimates when competing events make the observed response causally ambiguous.
Original synthetic instructional content · Technical peer review pending · No site authorization or graded credential
1. Understand the engineering
- A controller test changes a known effective manipulated input. A disturbance enters the process through demand, inlet composition, feed pressure, ambient conditions, or other uncommanded mechanisms.
- A process-variable response after a requested output step is not proof of causality if the actual actuator never moved or an unrelated process event occurred simultaneously.
- A low-residual mathematical fit cannot rescue a contaminated experiment. Record co-occurring events, controller mode, output selection, equipment condition, and measurement quality.
- Frame at least two alternative causal hypotheses and specify what independent evidence or controlled synthetic experiment would discriminate between them without violating process limits.
2. Apply it to a realistic synthetic case
During an educational cooling-water test, PV rises by 7 m³/h after a requested 5-point controller output increase. At the same timestamp, the upstream pump switched to a higher-speed operating mode.
Reveal step-by-step engineering analysis
- The apparent estimate of 7/5 process units per percentage point is contaminated because a second physical input changed at the same time as the controller request.
- Review actual final-element feedback and upstream pump speed, pressure, and event timestamps; either input could contribute to the measured flow change.
- Reject the identification as a single-input causal model until disturbances can be held fixed in synthetic comparison or independently characterized in archived evidence.
3. Defend your decision
Can a clean-looking 63.2% response prove a usable process gain when pump operating mode also changed?
Explain your response first. Identify an observed fact, an uncertainty, and the next authorized check.
Check the model explanation
No. A smooth or apparently first-order response does not separate competing causes. The simultaneous pump-mode change means the response cannot uniquely be attributed to the controller input without additional independent evidence.
Common misconception
A visually attractive curve fit always validates the chosen manipulated input as the cause, even when other physical conditions changed.
4. Safety and independent verification
Never perform uncontrolled plant steps solely to identify a model. Require approved operating conditions, stop criteria, independent protection, and authorization for any real test.
Public reference context
The original scenario does not reproduce licensed standard text. References identify relevant organizations and fields of practice, not independent approval or verification of the lesson.