PiControl Solutions


Request More InFo

Closed-Loop Universal Multivariable Optimizer for Model Predictive Control (MPC)
Artificial Intelligence (AI) based Algorithm Next Generation Model Predictive Control (MPC) Maintenance and Improvement Technology

View Columbo brochure.

Please contact us to get a free demo on COLUMBO - Closed-Loop Universal Multivariable Optimizer for Model Predictive Control (MPC) - Artificial Intelligence (AI) based Algorithm Next Generation Model Predictive Control (MPC) Maintenance and Improvement Technology.

[email protected], Tel: (832) 495 643

Nowadays, many modern industrial plants and manufacturing facilities poses top-notch high-level optimization technology such as Model Predictive Controllers (MPCs). These powerful technology servers to improve plant monetary benefits by often pushing industrial facilities to produce more products or save more on utilities. Typical MPC size can have control matrix between 5 to 20 Manipulated Variables (MVs) and 10 to 50 Controlled Variables (CVs).

At the beginning, many MPCs will produce great results and expected monetary benefits. 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 MPC quality to deteriorate. The root cause of MPC problems and deterioration in quality are most definitely changes in the dynamic models inside the MPC. The shapes of the MPC models are the heart and soul of the MPC system. Unfortunately, when MPC models change, or were wrong to begin with, it is a very difficult job to fix the models and identify the correct models. When MPCs deteriorate, many times oscillations start. Process oscillations are the biggest nightmare for a successful and stable plant performance.

Therefore, MPCs are often turned off with subsequent loss of benefits and profits. This can be a control engineer’s nightmare, and headache for the plant management, since they have spent a lot of money on the MPC project. Thats why continuous maintenance of MPC is important.

Identifying or fixing bad dynamic MPC models is one big challenge. COLUMBO as MPC Maintenance and Improvement Technology helps in confronting these challenges with unique technology unmatched by competitors. It uses data from any MPC operation, such as: DMC (Dynamic Matrix Control) from Aspen tech, RMPCT (Robust Model Predictive Control Technology) from Honeywell, PACE from Yokogawa, Predict Pro from Emerson or any other, and generate and improve correct models for the various MV-CV pairs.

Schedule A Demo

COLUMBO will help you to improve MPC models with either open-loop or completely closed-loop data. It does not need step-tests as required by Aspen tech, Honeywell, Yokogawa and others, and can identify models from a data when the MPC was completely ON and running. By using COLUMBO, upon CV target changes and/or upon a change in a feedforward signal (a measured disturbance), you will see improved control, tighter control and stability. Works entirely in the time domain, easy to use, compact and practical.


Dynamic Model Identification based on Active Model Predictive Controller

Currently practiced conventional known methods of model Identification typically involve making many small plant step-tests on the setpoints of slave PIDs in auto mode (often called closed-loop tests), or making many small step-tests on control valves (final control elements) in manual mode (often called open-loop tests). Even in the new and latest DMC3 Calibrate method – step tests are superimposed on top of the MV move trajectories from a DMC controller that is active (ON). So here, the DMC controller is ON, but step-tests are still needed for the DMC3 Calibrate to work. Moreover, these step-tests are “open-loop” by its definition.

Also, most models are based on making small steps of about 1-3% of the current MV values. If the step changes are too large, they may upset the process and the steps cannot be help for too long, resulting in the ambiguous determination of model parameters due to uncertainty from the data.

During model identification stage, only one MV is moved at a time followed by waiting time to try to see the new steady state. This is recommended by all MPC vendors as a good guideline to avoid correlation problems as if more than one MV (slave PID setpoint) is changed at the same time, then there is ambiguity and uncertainty regarding the impact of the multiple variables and subsequently dynamic models could be wrong due to correlations.

However, when the MPC is ON and running, it moves various MVs (slave PID setpoints) simultaneously to keep the CVs at their targets. Also, while the MPC is ON and running, it moves could be from 1 to 30 % of the current slave setpoints, where the system is going from one stage of process linearity and sensitivity to another. Models based on tiny open-loop steps without simultaneous changes in other MVs could produce models not truly reflecting the real dynamic behavior. These differences could be often masking the principles of linear superposition and assumption of MV-CV linearity (on which many conventional system identification technologies rely) and lead to model prediction errors, which is often the root cause of poor control in many MPC systems.

COLUMBO is the only technology across the globe which can overcome these problems by analyzing completely closed-loop data with MPC ON and running, without the need to provoke any open-loop, closed-loop, or superimposed plant step-tests. The differences between the open-loop model identification stage and the closed-loop stage with MPC ON and running can be significant, especially since there is nonlinearity of varying magnitude in all real-life chemical processes.


Artificial Intelligence (AI) based Dynamic Model Identification

FIR (Finite Impulse Response), ARMAX (Auto Regressive Moving Average with eXogenous inputs) and/or BJ (Box Jenkins) models have been used for several decades and are the main modelling methodology provided by many MPC vendors. Unfortunately, these methods do not work well with complete closed-loop data without any step tests at all. They also do not work well with unmeasured disturbances and noise. These methods offer what they call “noise models” for separating noise from the true process dynamics, but when unmeasured disturbances comprise of a mixture of fast random noise, medium frequency drifts and slow unmeasured disturbances, the noise model concept is not useful.

Schedule A Demo

COLUMBO is equipped with a powerful and new optimizer designed based on the Artificial Intelligence (AI). It is capable to automatically identify one or several unmeasured process disturbances, isolate their pattern, display as a trend, and save in the Excel. Works well admits fast random noise, medium frequency drifts and slow unmeasured disturbances.


Some of the distinguishing and powerful features of COLUMBO are listed below:

  • Reads current models and allows fixing all known model parameters like time to steady state, dead time or time constant or even process gain based on operator experience/knowledge, engineering calculations, vessel dimensions and vendor data. This unique functionality allows incorporation of process and engineering knowledge into the building of accurate dynamic models.
  • Conducts what-if analysis to compare model prediction with real process CV data.
  • Can process oscillatory data, ramped data or any data containing some CV target changes, FF disturbances or conventional open loop steps, though steps are not required.
  • Can identify up to ten MPC dynamic models simultaneously.
  • Can process any sized MPC system – even large MPCs with a 30 x 100 matrix.

All calculations are performed entirely in the time domain, no Laplace or Z (discrete) domain.