Process Control: PID, Cascade, Ratio, Override & Feedforward Strategies
A definitive guide to process control strategies — from single-loop PID to advanced cascade, ratio, feedforward and predictive controllers used in real plants.
Process control is the discipline of keeping a process variable (temperature, pressure, flow, level, composition) at a desired setpoint despite disturbances. PID is the foundation, but real plants use a hierarchy of strategies layered on top — cascade for tight control of slow processes, ratio for proportional mixing, feedforward for known disturbances, override for safety, and increasingly model predictive control (MPC) for highly-coupled multivariable processes.
This pillar covers the strategies that show up in every refinery, every chemical plant, every pharma facility, and increasingly in modern packaged equipment.
Continue Learning
Loop tuning and design
Frequently Asked Questions
What is process control?
Process control is the discipline of keeping a process variable (temperature, pressure, flow, level, composition) at a desired setpoint despite disturbances. The foundation is PID feedback control. Real plants layer additional strategies on top: cascade for tight control of slow processes, ratio for proportional mixing, feedforward for measurable disturbances, override for safety, and model predictive control for highly-coupled multivariable processes.
What is cascade control?
Cascade control puts a slow outer loop in series with a fast inner loop. The outer loop controls the primary variable (e.g., temperature) and outputs a setpoint to the inner loop, which controls a secondary variable (e.g., flow) and moves the actuator. The inner loop must be at least 5× faster than the outer loop. Cascade rejects inner-loop disturbances before they affect the primary variable.
What is ratio control?
Ratio control maintains a fixed proportional relationship between two flows. The setpoint of one flow tracks the actual flow of another flow times a configurable ratio. Critical for combustion air-fuel ratio, batch mixing component ratios, and reagent stoichiometry. Combined with PID for actual flow control: PID controls each flow to its calculated setpoint; the ratio block calculates the setpoint.
What is feedforward control?
Feedforward measures a known disturbance and pre-calculates a compensating MV output before the disturbance affects the controlled variable. Combined with PID, feedforward dramatically improves disturbance rejection: PID handles steady-state error and unmodeled disturbances; feedforward handles the modeled disturbance proactively. Requires the disturbance to be measurable and the disturbance effect to be modelable.
What is override control?
Override (or selector) control switches between multiple controllers based on a safety or constraint condition. A primary controller manages normal operation; an override controller takes over when a limit is approached. A LOW SELECT block picks the most-restrictive demand. Common in combustion (temperature override of fuel demand), compressor surge protection, tank overfill protection, and motor overload protection.
What is a Smith predictor?
A Smith predictor is a control structure for processes with long dead time (delay before PV starts responding). It adds a process model in parallel with the real process, predicts what the PV would be without dead time, and feeds that prediction to the PID controller. Allows PID to respond as if there were no dead time. Requires accurate process model; performance degrades with model mismatch.
When should I use Model Predictive Control (MPC)?
Use MPC for multivariable processes (multiple inputs and outputs), constraint-aware control (respect equipment limits explicitly), processes with significant dead time, or anticipation of future setpoint changes. Typical applications: refineries, large chemical plants, pulp and paper, energy. MPC implementations cost $100k-$10M+ and pay back through optimisation gains. For most discrete manufacturing and smaller process plants, well-tuned PID covers needs without MPC complexity.