Innovation in PID Controller Tuning: A Systematic Approach to Improved Controller Performance Standards Certification Education & Training Publishing Conferences & Exhibits Robert Rice, PhD Vice President, Engineering Control Station, Inc. Outline of Discussion • Introduction to Process Control – Brief history of Process Control – What is process control, and why do we need it – Common Examples of Process Control Systems • Introduction to Process Behavior and the Control Objective – Why understanding the process is fundamental to controlling it – The importance of stating the correct control objective • The PID Controller – What is a PID Controller – The importance of controller ‘tuning’ – Examples of the PID controllers (e.g. PID and the PIDE) • Theory Vs the Real-World • Questions and Answers History of Feedback / PID Control • 300BC – 1200 AD – Float Regulators used in Water Clocks (ON/OFF Control) – Used a float to control the inflow of water through a valve; as the level of water fell the valve opened and replenished the reservoir. This float regulator performed the same function as the ball and cock in a modern flush toilet. • 1700 – 1900 : Industrial Revolution – Centrifugal (Flyball) Governors (P-Only Controller) – This device employed two pivoted rotating flyballs which were flung outward by centrifugal force. As the speed of rotation increased, the flyweights swung further out and up, operating a steam flow throttling valve which slowed the engine down. Thus, a constant speed was achieved automatically. • 1900 – Current : Mass Manufacturing – Pneumatic, Electronic, Model Predictive Controllers – PID Control PID Tuning and Optimization Constraint Optimization – Moving Process Close to Constraint Status Quo– Poor Control PID Tuning – Improved Control Reduced Process Variability A well controlled process has less variability in the measured process variable (PV), so the process can be operated close to the maximum profit constraint. Steps to Successful Controller Design and Tuning 1. 2. 3. 4. 5. 6. Identify the Controller and Specify the Design Level of Operation (DLO) and Control Objective Perform a “Bump Test” and Collect Dynamic Process Data Fit a Model to the Process Data Use Tuning Correlations to Calculate Tunings Based on Model Implement and Test results Document the Tuning Process Good Control is “SIMPLE” afety mpact anagement rofit ongevity quipment What is the worst thing that can happen if this control loop fails? When designing your control objective, the safety considerations are paramount to all others. Where does this control loop fit in the overall process diagram. Where do the process disturbances come from? When this process changes, who gets impacted by it? If management desires a certain type of performance, it is your job to match that objective or explain in a logical manner why that type of control is not possible. What are the primary economics factors associated with this loop? How often is this loop the ‘culprit’ when diagnosing process performance? Keep the control strategy simple, the more complicated the strategy, the more likely it will fail. Process equipment is expensive, its components are expensive, and as such, should be included when formulating your control objective. Example Process Reflux Drum – Level Control Example What is/are the primary Control Objective(s)? Data Should Show “Cause and Effect” A bump test must generate a response that clearly dominates the random (noisy) PV behavior Here the PV moves about 4 times the noise band, a good value PV = 4% NB = 1% NB brackets the PV noise (3) when is CO constant. Here NB = 1% CO bump CO constant Copyright © 2007 by Control Station, Inc. All Rights Reserved Copyright © 2007 by Control Station, Inc. All Rights Reserved 8 Good Bumps Tests Open loop tests require the controller output to be stepped Closed loop tests require a sharp controller output change sharp CO movement sharp CO movement Bad Bump Tests AVIOD Disturbance Driven Data Slow Ramp CO Changes • Bump test data must contain reasonably pure CO to PV information so the model will accurately describe the cause-and-effect relationship 10 Types of Process Behavior • Self-Regulating – If all inputs & outputs are held constant, the process will seek a steady-state – Ex: Heat Exchanger • Non Self-Regulating – Process will only reach a steady-state at its ‘balancing’ point – Ex: Surge Tank First Order Models for Modeling Self-Regulating Non Self-Regulating “All models are wrong, some are useful” George Box 12 First Order Plus Deadtime Self-Regulating Model 13 First Order Plus Deadtime Non Self-Regulating Model 14 Tunings Only As Good as the Model • By Hand Approach Sufficient for Simplest of Controllers • Software Modeling Much More Robust – Handle Open/Close Loop – Noisy / Non-Steady State Conditions SIMPLE 15 PI Tuning Correlations (IMC) Dependent PI, Self-Regulating Process 1 CO KC E E TI PID Tuning Correlations (IMC) Dependent PID, Self-Regulating Process 1 dPV CO KC E E TD T dt I PID Tuning Correlations (IMC) Dependent PI, Non Self-Regulating Process 1 CO KC E E TI Closed Loop Time Constant Selection Rules of Thumb 19 Expected PI Controller Response Self-Regulating Processes Conservative Moderate Aggressive Copyright © 2007 by Control Station, Inc. All Rights Reserved. • Set point tracking (servo) response as tc changes Challenges of PI Control Interaction Self-Regulating Processes Kc*2 2 Kc Base Case Performance Kc Kc Kc/2 Kc/2 t ITi/2 /2 tTiI Copyright © 2007 by Control Station, Inc. All Rights Reserved. 22Ti tI Challenges of PI Control Interaction Non Self-Regulating Processes 2*Kc Kc Kc / 2 Ti/2 Ti 2Ti PI vs PID Set Point Tracking Response Heat Exchanger – Aggressive Tuning SP PI PID Copyright © 2007 by Control Station, Inc. All Rights Reserved. • PID shows decreased oscillations compared to PI performance • PID has somewhat: – – – Shorter Rise Time Faster Settling Time Smaller Overshoot Example Process: Heat Exchanger • Process Variable (PV) • Set Point (SP) • Controller Output (CO) • Disturbances (D) D CO PV SP Processes Have Time Varying Behavior Heat Exchanger Shows Nonlinear Behavior equal SP steps PV response varies with a fixed-tuning PI controller Copyright © 2007 by Control Station, Inc. All Rights Reserved • Processes often exhibit changing (or nonlinear) behavior as operating level changes • As a result, “best” tuning can change if the set point moves the PV across a range of operation Controller’s Robust Stability • What does it mean for a controller to be Robustly Stable? – Controller Robustness measures the Ability to Tolerate Variations in Process Behavior (e.g., Nonlinearity) • Visual Robust Stability Plot Actual Process Gain Increases – Plots Plant-Model Mismatch in Gain vs. Plant-Model Mismatch in Dead Time – Stable and Unstable Regions shown on Plot Moderately Tuned Aggressively Tuned UNSTABLE UNSTABLE STABLE Actual Deadtime Increases STABLE Summary • First Order Models provide Important Information – How Far?; How Fast?; With How Much Delay? – Fit by Hand or Use Software • Systematic Approach to Tune PID Controllers – Internal Model Control (IMC) Tuning – Uses the FOPDT Model in the Tuning Correlation – Specifying the Single Adjustable Tuning Parameter, tc – Decrease tc for a Faster, More Aggressive Response – Increase tc to Increase Robustness • Understanding Robust Stability – Processes Change over time and with Operating Level – Controller Performance can degrade over time – Select Tunings which balance performance with robust stability Questions? • Thank you for attending! • Contact Information: Bob Rice, PhD Vice President, Engineering +1-860-872-2920, ext. 1601 +1-860-420-7158 (m) bob.rice@controlstation.com