COLUMBO

COLUMBO: Universal Multivariable Closed-Loop MPC Optimizer

Dynamic matrix control (DMC), Robust model predictive control technology (RMPCT), PredictPro, Connoisseur and other model predictive controllers (MPC) are widely used in chemical, petrochemical, paper, power plant, oil refining and other similar industries. 

The shapes of the model predictive control (MPC) models is the heart and soul of the model predictive control (MPC) system. Most of model predictive control (MPC) systems characterize process dynamics using either transfer function parameters or step response coefficients. Unfortunately, when model predictive control (MPC) models are wrong, it is a very difficult job to fix the models and identify the correct models. With time, due to aging equipment, hardware changes, changes in process, operating and economic conditions and process nonlinearities dynamic models can change significantly causing the model predictive control (MPC) control quality to deteriorate. Model predictive control (MPC) systems are often turned off with subsequent loss of benefits and profits. 

Currently practiced conventional known model predictive control (MPC) technology for its model identification typically involve making small step tests on the setpoints of slave PID control loops in auto mode or, step tests on control valves directly in manual mode. Even in the new and latest DMC3 calibrate method – step tests in manual mode are superimposed on top of the manipulated variables move trajectories from a DMC controller that is active. 

Model predictive control (MPC) models are typically built using small steps (moving SP 1 to 3 %) where typically only one manipulated variable is moved at a time to avoid complications due to correlations. This is recommended by all model predictive control (MPC) vendors as a good guideline to avoid correlation problems and subsequently making identified dynamic models wrong due to those correlations.  

When a measured disturbance changes, the model predictive controller (MPC) tries to compensate it based on the calculated models that were identified using the small steps. The difference now is that a measured disturbance may have changed more than 3% and these larger changes can lead to new and unknown nonlinearities. Also, when model predictive control (MPC) system is running, several manipulated variables are changed simultaneously in order to keep the controlled variables at their targets. All these differences lead to model prediction errors which is often the root cause of poor control in many model predictive control (MPC) systems.  

PiControl Solutions COLUMBO software product provides novel model predictive control (MPC) model identification functionalities:

• identify correct open-loop dynamic predictive control (MPC) models using complete closed or open-loop process data even when model predictive control (MPC) system is running.
• it can simultaneously identify ten (10) model predictive control (MPC) dynamic models even for slave PID controllers in auto or in cascade mode. 
• improves model accuracy and control performance of any model predictive control (MPC) system without conventional step tests.
• allows fixing all known parameters like time to steady state, dead time, time constant or even process gain based on process experience. engineering calculations, vessel dimensions and/or vendor data.