C-P Systems
What is a Proportional-Integral-Derivative (PID) Control?| Process Engineering Glossary
What is a Proportional-Integral-Derivative (PID) Control?
In piping engineering and process engineering, PID control — Proportional-Integral-Derivative control — is the standard feedback control algorithm used to regulate a process variable at its desired setpoint by continuously adjusting a manipulated variable. It accounts for more than 90 percent of all feedback control loops operating in process plants worldwide, from simple flow and pressure loops to complex temperature and composition controllers on reactors and distillation columns.
The PID controller computes a control output u(t) as the weighted sum of three terms based on the error signal e(t) = SP − PV, where SP is the setpoint and PV is the measured process variable:
u(t) = K_c × [e(t) + (1/τ_I) × ∫e(t)dt + τ_D × de(t)/dt]
Or equivalently using individual gains:
u(t) = K_P × e(t) + K_I × ∫e(t)dt + K_D × de(t)/dt
Where K_c is the controller proportional gain, τ_I is the integral time (reset time, minutes per repeat), τ_D is the derivative time (rate time, minutes), and K_P = K_c, K_I = K_c/τ_I, K_D = K_c × τ_D in the ISA standard form. The controller output u(t) drives the final control element — typically a control valve, a variable-speed pump, or a heater power supply — to minimise the error between the setpoint and the measurement.
Applications of PID Control
Fired Heater Outlet Temperature Control
Fired heaters (process furnaces) use PID control to maintain the coil outlet temperature at the process set point by manipulating the fuel gas control valve to the burners. The process time constant of a fired heater is typically 5 to 30 minutes (the thermal mass of the refractory and the process fluid in the coil), with dead time of 0.5 to 3 minutes from the thermocouple location to the combustion zone. PID settings are conservative: K_c of 0.5 to 2.0, τ_I of 5 to 15 minutes, and often τ_D of 1 to 5 minutes to reduce peak temperature excursions during feed rate changes.
Distillation Column Pressure Control
Column pressure is typically controlled by manipulating the condenser cooling medium flow (increasing cooling lowers pressure by condensing more overhead vapour). The process gain for pressure-to-cooling-flow is usually high (small changes in cooling flow produce relatively large pressure changes) and the dynamics are fast (minutes for the column vapour space to respond to cooling changes). PI control with moderate K_c and short τ_I provides tight pressure control. Derivative is not used because pressure measurements in vapour service often carry high-frequency noise from vapour bubbling.
Benefits of PID Control
Universal applicability. PID control handles self-regulating and some integrating processes, covers single-input single-output loops of any industrial type, and provides adequate performance for the vast majority of process plant control requirements without more complex algorithms.
Physical interpretability. Each of the three PID modes has a clear physical meaning: proportional responds to current error, integral eliminates past accumulated error, derivative anticipates future error trend. This interpretability allows engineers to diagnose loop performance issues — too much oscillation (K_c too high), persistent offset (τ_I too long), noise-amplified chattering (τ_D too high) — directly from the process variable trend.
Mature tuning methodology. Decades of industrial experience with step test methods, Ziegler-Nichols, Lambda tuning, and auto-tuners embedded in modern DCS platforms give engineers multiple well-validated paths to adequate controller tuning without requiring detailed mathematical process models.
Limitations to Consider
Dead time limits performance. For processes with long dead time relative to the process time constant (θ/τ > 0.5), PID control cannot achieve tight setpoint regulation regardless of tuning. The controller must wait through the dead time before any response is apparent, and by then the process may have moved substantially. Smith Predictor and model-predictive controllers handle large dead time more effectively than standard PID.
Nonlinear processes require gain scheduling or adaptive tuning. When K_p varies significantly across the operating range (pH control, variable-throughput reactors, distillation columns with wide composition swings), a single fixed PID tuning set cannot provide optimal performance throughout. Gain scheduling, feedforward compensation, or model-predictive control is needed for these applications.
Interactions between loops require decoupling. In multivariable processes where multiple PID loops interact — changing flow in one loop changes the process variable of another — individually tuned PID controllers may fight each other, causing sustained oscillations or sluggish response that no amount of single-loop retuning can resolve. Decoupling networks, relative gain array analysis, or MPC are required for well-interacting multivariable processes.
PID Control FAQ
What is PID control and how does each term contribute to loop performance?
PID control is a feedback algorithm that adjusts a manipulated variable to keep a process variable at its setpoint, computing u(t) = K_c × [e + (1/τ_I) × ∫e dt + τ_D × de/dt] where e = SP − PV. Process engineering relies on it for virtually all closed-loop control applications. Proportional action provides immediate response proportional to current error but leaves a permanent offset. Integral action accumulates error over time to eliminate steady-state offset — the dominant reason most loops use PI rather than P-only. Derivative action responds to the rate of error change, anticipating future deviation and reducing overshoot in slow processes with long time constants, but amplifies measurement noise and is omitted in most flow and turbulent measurement loops.
How does a PID controller drive a control valve and how are the tuning parameters determined?
The PID controller output signal (typically 4-20 mA or a digital equivalent) drives the control valve positioner, which positions the valve stem to match the controller output — low signal closes the valve, high signal opens it for a direct-acting valve configuration (or the reverse for air-to-close). Tuning parameters K_c, τ_I, and τ_D are determined from a bump test that measures the process gain K_p, process time constant τ, and dead time θ from the step response. Lambda tuning gives K_c = τ/(K_p × (λ + θ)) where λ is the desired closed-loop time constant. The instrumentation signal chain from the process measurement transmitter through the DCS to the valve positioner introduces time delays that add to θ and must be accounted for in the effective dead time used for tuning calculations.
How does PID control interact with feedforward, cascade, and advanced control strategies?
Standard PID is purely reactive — it responds to error after a disturbance has already affected the process variable. Feedforward control adds a proactive element that measures an incoming disturbance (feed flow change, ambient temperature variation, steam header pressure change) and applies a compensating adjustment to the manipulated variable before the disturbance reaches the process, reducing the error that the PID must then correct. Cascade control nests two PID loops: the outer (primary) PID manipulates the setpoint of an inner (secondary) PID that controls the immediate manipulated variable (typically valve flow), with the inner loop rejecting fast disturbances before they reach the primary loop. Both strategies improve on the basic PID loop by reducing the effective dead time and disturbance magnitude that the feedback PID must handle, enabling tighter control of the primary process variable than would be achievable with feedback alone.
About C-P Systems
SETTING THE STANDARD FOR CHEMICAL ENGINEERING FIRMS EVERYWHERE
Through unmatched professionalism, knowledge and experience, we set the industry bar for chemical engineering firms. With decades of chemical plant engineering and piping design experience, our team of licensed engineers can handle any project scope.