A simplified predictive control algorithm for disturbance rejection

This is from a series of articles reprinted from the journal ISA Transactions.  All ISA Transactions articles are free to ISA members, or can be purchased from Elsevier Press.

Fig. 1. Estimation of the effects of unmeasured disturbance.

Abstract: Model predictive control (MPC) offers several advantages for control of chemical processes. However, the standard MPC may do a poor job in suppressing the effects of certain disturbances. This shortcoming is mainly due to the assumption that disturbances remain constant over the prediction horizon. In this paper, a simple disturbance predictor (SDP) is developed to provide predictions of the unmodeled deterministic disturbances for a simplified MPC algorithm. The prediction is developed by curve fitting of the past information. A tuning parameter is employed to handle a variety of disturbance dynamics and a procedure is presented to find an optimum value of the tuning parameter online. A comparison is made with the commonly used disturbance prediction on three example problems. The results show that an improved regulatory performance and zero offset can be achieved under both regular and ramp output disturbances by using the proposed disturbance predictor.

 Free Bonus: To read the full article on a predictive control algorithm for disturbance rejection, click here.

ISA membership entitles you to free access to all ISA Transactions articles plus a wealth of technical content, industry information, free webinars, training opportunities, program discounts, certification and licensure and professional networking.

Join ISA … learn, advance, succeed!


2006 Elsevier Science Ltd. All rights reserved.

Tuning Control Loops Course: What is a “Bump Test”?

This article was submitted by Don Lovell, an automation consultant assisting customers in defining their automation road map to meet their business objectives.
Don Lovell

Don Lovell

A customer had a poor performing loop in the brewery. The problem would really become an issue once in awhile depending on the operating conditions in the plant. The problem would be passed from person to person, shift to shift with no real answer.

In discussing the problem with a controls engineer, the engineer asked if an “open loop” bump test had been performed. The answer was no. Question: What is a bump test? A bump test was performed and initial tuning parameters were calculated and installed. The loop began to settle down.

An “open loop” bump test is performed by placing the loop in manual, making a step change in the manual output and observing or trending the result of the process variable. With the resultant process curve displayed, the process gain, dead time and process time constant can be determined. With these three pieces of information the technician can determine the initial tuning parameters by selecting among hundreds of tuning strategies. Strategies encompass the type of process, process characteristics and tuning objectives. Common or classical strategies include: Ziegler-Nichols, Cohen-Coon and Lambda.

The initial problem solving method is discussed in ISA, Troubleshooting: A Technicians Guide, 2nd Edition.

From page 65:

Consultation, also known as the “third head” technique, means that you and the equipment operator use a third person, perhaps someone from engineering or an outside consultant, with advanced knowledge about the system or the principles for troubleshooting the problem. This person may not solve the problem but may ask questions that make the cause apparent or spark fresh ideas for you. This process allows you to stand back during the discussions, which sometimes can help you distinguish the trees from forest.”


2012 AIChE Spring Meeting & 8th Global Congress on Process Safety

In case you missed it, the 2012 AIChE Spring Meeting & 8th Global Congress on Process Safety is being held this coming week (April 1-4, 2012) in Houston, TX.  There is some great content being presented:

Event Keynote:

MICHAEL J. DOLAN Senior Vice President, ExxonMobil Corporation “Process Safety and Corporate Responsibility”

Other Process Safety Content:

Look Ma No Faceplates: Benefits of In Situ Modification of Set Point and Output on Operator Displays

This presentation illustrates the need and solution for operators viewing and making changes to set points and outputs in context. It shows the penalty and risks of operators shifting focus to faceplates and performing other window management tasks when trying to operate an industrial process.  In-situ modification improves the speed of operator response, reduces entry errors and operator loading.  For the industrial process, this means less spurious trips due to operator entry and more reliable and efficient plant operation.  The presentation will cover how to incorporate items normally displayed in the faceplate directly into the operator display.  The illustrated examples will be running ProSys’ Interface Dynamics display library, but concepts in presentation are applicable to any industrial HMI and can be achieved in almost any control system.

Fire and Explosion Hazards

This session invites papers on recent developments in evaluating fire and explosion hazards in chemicals processing, storage and transportation. Example areas of new interest include flammable liquid transport and storage, the distinction between fire point and flash point in the evaluation of intermediate bulk containers (IBCs), and methods for predicting or modeling the behavior of non-homogeneous mixtures.

The Connection of Peak Alarm Rates to Plant Incidents and What You Can Do to Minimize

Even after several years of trying, many plants still struggle with controlling alarm floods.  Static rationalization can reduce your average number of alarms, but without controlling the alarm floods, there is no help for the operator when he needs it the most.  This session will cover the justification for alarm management from the safety and environmental as well as economic perspective.

Hazard Identification and Risk Assessment

Effective process management programs utilize numerous assessment techniques to identify, and evaluate hazards associated with the processes employed to produce the products customers desire. When the evaluations show the associated risk of a given process hazard to be above tolerance levels recommendations/actions are generated to help manage these risks. The sheer number and magnitude of the recommendations typically overwhelm the resources of most organizations. We are interested in receiving abstracts for papers that discuss: • Systems that track and mange the recommendations through their life cycle. • Methods of integration of recommendation priorities generated by a variety of assessment techniques to generate one master list. • The process of how recommendations are prioritized and selected for funding

Eliminating Difficult Start Ups with State Based Controls

Some processes are just difficult to start up. Every year, significant production is lost due to start up delays, product quality issues after startup, mistakes or gaps in operator expertise and other reasons. In fact, the U.S. Chemical Safety Board has estimated that the total loss due to operator error is $10B per year. In addition, because 70% of process incidents occur during start-up and shutdowns, safety can also be positively impacted by automating the process start-up. Process start-ups can be completely automated and places the operator in supervisory mode approving the continuation of the process at specific intervals. The automation of start-up allows every start up to be exactly the same, no matter which operator is at the controls. The automation eliminates potential errors and overshoots which preserves the integrity of the startup and eliminates most recycle, purges, safety trips and lost production.

Using state based controls has cut average start-up time by a substantial amount, some as much as 50% or more. This is accomplished by streamlining the process and using the logic tools for automation. Automated start-up improves profitability and enhances flexibility of the process. Project ROI estimates show that the project typically pays for itself in about 3 months.


Good luck to all ISA members who are presenting there! Make sure you try to find each other there…

Pin It on Pinterest