Practical Procedure · Safety Systems Control Valve Partial Stroke Testing (PST) for Safety Valves Applies to: Emergency shutdown (ESD) valves and safety instrumented function (SIF) final elements | Equipment needed: PST-capable positioner/logic solver, position feedback monitoring, coordination with process/operations Full Stroke vs. Partial Stroke — Why PST Exists Tap a tab to see the difference between a full functional test and a partial stroke test, and why PST lets you test without shutting down. Full Stroke Test Partial Stroke Test (PST) ESD Valve — Process Running 0% → 100% → 0% process fully interrupted during test ESD Valve — Process Running ...
Get link
Facebook
X
Pinterest
Email
Other Apps
PID
Get link
Facebook
X
Pinterest
Email
Other Apps
-
Part 3, Chapter 1 — PID Control From First Principles
Part 3, Chapter 1 · Industrial Control Systems Mastery
PID Control From First Principles
Three terms, each answering a genuinely different question about the error: how big is it right now, how long has it been building, and how fast is it changing?
📘 PART 3: Control Strategies — Chapter 1 of 7
Live · Simulated Process Response to a Setpoint Step
Every control strategy covered in the rest of this Part — cascade, ratio, feedforward, override — is built on top of the PID controller covered in this chapter. Getting a genuine, correct understanding of what each of the three terms actually does is worth the time it takes.
Proportional (P): Responding to How Big the Error Is Right Now
The proportional term produces an output directly proportional to the CURRENT error — a bigger error produces a bigger corrective output, immediately. A larger Kp produces a more aggressive response, and can genuinely cause visible oscillation (overshooting, dipping back, overshooting again) before eventually settling — Kp alone controls responsiveness, not stability by itself. Critically, even after that oscillation dies out, P alone typically leaves a persistent steady-state offset (sometimes called "droop") for most real processes — the output settles at a value proportional to whatever residual error remains, meaning P alone can rarely drive the error all the way to zero, no matter how the gain is tuned.
Integral (I): Responding to How Long the Error Has Persisted
The integral term accumulates error over TIME, continuing to increase the output as long as any error remains — this is specifically what eliminates the steady-state offset that P alone leaves behind, since the integral keeps growing (and therefore keeps increasing the output) until the error reaches genuinely zero. The practical tradeoff is "integral windup" — if an error persists for a long time before the actuator can actually respond, the accumulated integral value can grow to an extreme, causing a real overshoot once the process finally catches up (covered further in the next chapter).
Derivative (D): Responding to How Fast the Error Is Changing
The derivative term reacts to the RATE of change of error, providing an anticipatory braking effect — if error is rapidly shrinking, the derivative term reduces output slightly ahead of time, helping to reduce overshoot rather than reacting only after the process has already passed the setpoint. The genuine tradeoff is noise sensitivity — the derivative term amplifies rapid, small fluctuations, meaning it's used more cautiously (or omitted entirely) in applications with a naturally noisy measured signal.
Term
Responds To
Genuine Tradeoff
Proportional (P)
Current error magnitude
Leaves a persistent steady-state offset
Integral (I)
Accumulated error over time
Risk of windup and overshoot
Derivative (D)
Rate of change of error
Amplifies signal noise
Watch the Simulation Above Closely
P-only settles noticeably short of the setpoint — a genuine, visible offset. Adding I eliminates that offset entirely, but introduces some overshoot and oscillation before settling. Adding D on top of PI damps that oscillation, reaching a stable final value with less overshoot than PI alone. This is the actual practical behavior every PID tuning decision in the next chapter is built around.
Check Your Understanding
Why does a P-only controller typically leave a persistent steady-state offset, no matter how the proportional gain is tuned?
Answer: The output is only ever proportional to the CURRENT error — once the output reaches whatever value the process needs to hold steady, the error stops shrinking, since a smaller output would mean a smaller error signal, and the two settle together at a nonzero error value.
Why is the derivative term used more cautiously in applications with a naturally noisy measured signal?
Answer: The derivative term reacts to the RATE of change of error — small, rapid noise fluctuations produce large apparent rates of change, which the derivative term amplifies into an erratic, unhelpful control output.
CHAPTER RECAP
Output = Kp × Error + Ki × ∫Error dt + Kd × (dError/dt) — the complete PID equation
Proportional responds to current error but leaves a persistent steady-state offset
Integral eliminates that offset by accumulating error over time, at the risk of windup and overshoot
Derivative anticipates future error trends to reduce overshoot, at the cost of noise sensitivity
Every control strategy in the rest of this Part builds on top of this same PID foundation
Fault Log · Field Report Schenck Weigh Feeder — Zero and Span Drift, Two Different Root Causes Plant: Cement | System: Schenck weigh feeder, belt scale, hopper load cells | Downtime: None Symptom A Schenck weigh feeder started showing both zero drift and span drift at the same time. Two separate accuracy problems showing up together made this a two-part investigation from the start — zero and span errors don't always share the same root cause, even when they appear together. Investigation — Zero Drift Started with the load cell loop and mechanical side of the belt scale: Load cell loop — checked the full loop, cleaned the load cell area. Idlers/rollers — checked the rollers for wear or misalignment. Belt tracking — with the belt empty, noticed it was drifting to one side. Zero was being tared in that empty, drifted position. Once material loaded onto the belt, the belt tracked back closer to center — meaning the zero reference had been tak...
Practical Procedure · Calibration Load Cell Calibration for Weigh Feeders — Dead-Load, Test-Weight, and Why Shunt Calibration Isn't Enough Applies to: Belt weigh feeders, weigh hoppers with load cell-based measurement | Equipment needed: Certified test weights or a dead-load/material reference, shunt calibration resistor (if supported), digital indicator/weigh controller access Before You Start Confirm the feeder/hopper is mechanically free — check idlers, belt tension, and load cell mounting for anything that could add friction or false load before assuming a load cell problem. Check load cell cabling and junction box connections for corrosion or loose terminations, especially in dusty or humid plant environments. Note whether the system uses a single load cell or multiple load cells summed together — multi-cell systems need each cell checked individually, not just the combined output. Method 1 — Shunt Calibration Shunt calibration simulates a known load...
Fault Log · Field Report Zirconia Oxygen Analyzer Reading Maximum O2 — Not the Cell, a Rusted Outlet Plant: Cement | System: Zirconia oxygen analyzer, sample extraction system | Downtime: None Symptom The zirconia oxygen analyzer started reading maximum oxygen — pegged at the top of the range rather than tracking the actual process value. A reading stuck at maximum on a zirconia cell usually means the cell is seeing something close to ambient air instead of process gas, which points straight at a sample delivery problem before you even think about the analyzer itself. Investigation Started at the sample path, from the process side inward: Probe, extractor filter, sampling line, pump — checked the full sample path from the probe through the extraction filter and sampling line, including the sample pump. Performed a leakage test across this section. All found ok. Panel-side tubing and membrane filter — checked all tubing inside the panel along...
Comments
Post a Comment