Browse Prior Art Database

MODEL PREDICTIVE CONTROLLER AUTOMATIC ACTIVATION

IP.com Disclosure Number: IPCOM000188195D
Publication Date: 2009-Sep-25
Document File: 2 page(s) / 55K

Publishing Venue

The IP.com Prior Art Database

Abstract

A computer method for automatically activating a model predictive controller from an off-state to an on-state is disclosed. The activation software monitors controller and process statuses. Prior to switching the controller from an off-state to an on-state the activation software loads the controller configuration appropriate for the process operating mode.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 2

MODEL PREDICTIVE CONTROLLER AUTOMATIC ACTIVATION

ABSTRACT

A computer method for automatically activating a model predictive controller from an off-state to an on- state is disclosed. The activation software monitors controller and process statuses. Prior to switching the controller from an off-state to an on-state the activation software loads the controller configuration appropriate for the process operating mode.

DESCRIPTION

The current state of the art, when it comes to activating a model predictive controller (MPC), consists of allocating an operator's time to monitor MPC and perform manual interventions until its execution is stable and satisfactory. Even though "stable and satisfactory" are precise criterion for engineers who commission MPC's, plant operators often lack the necessary expertise and/or time to make this judgment call with accuracy. Not only this condition leads to various "time to activation" between different operators, thus making the activation process highly variable in quality and duration, but also this can lead to a longer-than-necessary time for activating MPC. Indeed if MPC is activated sooner than it should then MPC will become unstable and often will perform actions that may take considerable time to undo, thus reducing operator's availability for other tasks. The above issues are often compounded particularly when the operation modes have significantly changed between the time MPC was switched off and the time of MPC reactivation.

For example in an air separation plant the main actions to be taken by an operator to activate a model predictive controller following a nitrogen liquefier shutdown or startup share some similar tasks and sequence of actions. Consequently prior to MPC activation the plant may be in a different mode thus requiring the operator to monitor different process variables, and activating MPC requires an understanding of both the MPC core calculations as well as detailed process behavior. The latter point makes it difficult for new operators to develop the necessary knowledge to execute this task with a relatively constant outcome. Moreover, this is compounded by the fact that any two MPC activations may not be truly identical because process conditions always vary. Therefore, both operators and plant superintendants would benefit from a technology that would automate a series of tasks requiring a skill set that depends on a large number of variables and significant plant operating experience.

Figure 1 depicts how a software solution for MPC activation would interact at a higher level with controlled variables (CV), manipulat...