Q5Control System
Question
Q.5. Compare PI, PD and PID controllers.
Answer
PI control eliminates steady-state error but slows response and can add overshoot; PD control speeds up response and improves damping but does not remove steady-state error; PID combines both, using integral action to kill steady-state error and derivative action to improve transient damping.
PI (Proportional-Integral) Controller
Transfer function Gc(s) = Kp + Ki/s. The integral term accumulates the error over time and drives steady-state error to zero for step inputs (increases the system type by 1), improving steady-state accuracy significantly. However, the added pole at the origin reduces phase margin and stability margin, tends to slow the transient response, and can introduce overshoot or oscillatory behavior if Ki is not tuned carefully; it is best applied to systems with unacceptable steady-state error where speed is secondary.
PD (Proportional-Derivative) Controller
Transfer function Gc(s) = Kp + Kd·s. The derivative term responds to the rate of change of error, providing anticipatory (predictive) correction that increases damping, reduces overshoot, and speeds up the transient response (reduces rise/settling time) by adding phase lead near the crossover frequency. However, a PD controller does not affect the system type and therefore cannot eliminate non-zero steady-state error caused by step or higher-order inputs; it also amplifies high-frequency measurement noise since pure differentiation has high gain at high frequency.
PID (Proportional-Integral-Derivative) Controller
Transfer function Gc(s) = Kp + Ki/s + Kd·s. PID combines the strengths of both: the integral term drives steady-state error to zero (excellent accuracy), while the derivative term compensates for the phase lag and sluggishness introduced by the integral action, restoring good transient response, adequate phase margin and reduced overshoot. This makes PID the most versatile and widely used industrial controller, but it requires careful three-parameter tuning (e.g., Ziegler-Nichols method) since improper tuning can lead to instability, and derivative action still requires noise filtering in practical implementation.
- Steady-state error: PI → eliminated (Type increases); PD → unchanged; PID → eliminated.
- Transient response/speed: PI → slower, more overshoot; PD → faster, better damped; PID → fast and accurate if well-tuned.
- Stability margin: PI → reduces phase margin; PD → increases phase margin; PID → net effect depends on tuning, generally balanced.
- Noise sensitivity: PI → low; PD → high (amplifies noise); PID → moderate, needs derivative filtering.
In practice, the choice among these three controller structures depends on which specification dominates the application: PI is preferred where steady-state accuracy is critical and moderate speed is acceptable (e.g., temperature or flow regulation), PD is preferred where fast, well-damped tracking matters more than eliminating a small residual offset (e.g., certain servo positioning tasks with an inherently Type-1 or higher plant), and PID is the default choice for general-purpose industrial process control where both accuracy and transient quality matter, provided the extra tuning effort and noise-filtering requirement are acceptable.