The control loop
Every feedback loop has the same parts: a sensor/transmitter measures the process variable (PV); a controller compares it with the setpoint (SP) and calculates an output (OP); a final control element, usually a valve, acts on the process. The aim is to keep PV at SP despite disturbances.
Before tuning anything, check the basics: is the transmitter range right, does the valve stroke fully and smoothly, and is the controller's action (direct or reverse) correct? Most "tuning problems" are actually instrument or valve problems.
PID control
OP = Kc · [ e + (1/Ti)·∫e dt + Td · de/dt ] e = SP − PV (reverse acting)
- Proportional (P): output proportional to error. Alone it leaves an offset. Gain Kc = 100 / Proportional Band (%).
- Integral (I): keeps moving the output while any error exists, which removes the offset. Too much integral causes slow oscillation and overshoot. Watch for integral windup when the valve is saturated.
- Derivative (D): reacts to the rate of change. It helps slow loops such as temperature, but amplifies noise, so it is rarely used on flow or level.
| Loop type | Typical behaviour | Usual controller |
|---|---|---|
| Flow | Fast, noisy | PI, low gain, fast integral |
| Pressure (liquid) | Fast | PI |
| Pressure (gas) | Moderate, self-regulating | PI, higher gain |
| Level | Integrating (no self-regulation) | P or PI; averaging control for surge tanks |
| Temperature | Slow, dead time | PID |
Tuning approaches
- Bump test first: in manual, step the output by 5–10 % and record the process gain, dead time and time constant. These three numbers drive every tuning method.
- Lambda (IMC) tuning: choose the desired closed-loop speed (λ), typically 1–3× the process time constant. It gives smooth, non-oscillatory response, and is widely used in the process industries.
- Ziegler–Nichols: a classic method, but aggressive (quarter-amplitude decay), so usually too oscillatory for process plants without detuning.
- Change one parameter at a time, record before and after, and test with both setpoint changes and disturbances.
Common control strategies
- Cascade: an outer (master) loop sets the setpoint of a faster inner (slave) loop, e.g. reactor temperature → cooling water flow. Tune the inner loop first.
- Ratio: keep one flow proportional to another, e.g. fuel/air or blending.
- Feedforward: act on a measured disturbance before it affects the PV, e.g. steam flow feedforward in three-element drum level control.
- Split range: one controller output drives two valves over different parts of its range, e.g. heating 0–50 %, cooling 50–100 %.
- Override / selector: high or low selectors protect constraints, e.g. compressor surge and motor current limits.
DCS, PLC and SCADA
| System | Strength | Typical use |
|---|---|---|
| DCS | Integrated continuous control, one database, operator graphics, alarm handling | Refineries, petrochemical, power plant unit control |
| PLC | Fast, robust sequential and discrete logic | Package units, compressors, water treatment, machinery |
| SCADA | Supervision over wide areas and slow communication links | Pipelines, water networks, substations, wellheads |
The boundaries have blurred: modern PLCs run PID well and DCSs do sequences. What matters is integration, lifecycle support and the operators' needs.
Safety instrumented systems (SIS)
The SIS is an independent protection layer that brings the process to a safe state when the control system and operators can't. It is designed to IEC 61511 (IEC 61508 for the equipment):
- A hazard analysis (HAZOP, LOPA) assigns each safety instrumented function a SIL (Safety Integrity Level 1–4). Most process functions are SIL 1–2.
- The SIL target sets the required reliability. It drives the architecture (1oo1, 1oo2, 2oo3 voting), the device selection (SIL-capable, proven in use) and the proof test interval.
- Keep SIS sensors and valves separate from control, manage bypasses strictly, and record every proof test.
Alarm management
ISA-18.2 / IEC 62682 define the alarm lifecycle. Every alarm should be unique, need an operator action, and be prioritised. As a benchmark, an operator should see on average no more than about one alarm per 10 minutes in steady operation. Floods, chattering alarms and standing alarms all hide the alarm that matters.
Reading tag numbers (ISA-5.1)
| First letter (variable) | Succeeding letters (function) |
|---|---|
| F: Flow | T: Transmitter |
| L: Level | I: Indicator |
| P: Pressure | C: Controller |
| T: Temperature | V: Valve |
| A: Analysis | E: Primary element |
| S: Speed · W: Weight | S: Switch · A: Alarm (H/L) |
So FIC-101 is a flow indicating controller in loop 101; LT-205 a level transmitter; PSV-310 a pressure safety valve; TE-404 a temperature element.