Process Control and Information Handling

Researchers: Charlotta Johnsson (area leader)

Start-date: 2010-08

End-date: 2011-10.

Results: Today, the project has its own financial sources. It received financial support through the PIC-opic project (Process Industrial Centre, supported by SSF, start 2012-01) and LISA (Line Information Systems Architecture, supported by VINNOVA, start 2011-10).  

Acknowledgement: Thanks to LCCC for 14 months of financial support greately facilitating the startup of this project. 


Objective: To show how data from control systems used at production plats in the process industry can be transformed into more elaborated information, and to demonstrate/examine how this information can be used to control the plant from a plant-wide perspective.

Description: At production plants in the process industries there are many variables that are being monitored and controlled. A common number is to have between 1000-10 000 variables and about 100-1000 control loop at each site. Today, all variables are saved in an historical database. A dilemma is that these variables and control loops do not have their focus on the overall plant performance but rather on a local part of the plant (e.g. a control loops controlling the level in a tank or the temperature in a tank).

When designing local control loops, the common procedure is to 1) start by selecting the control-parameters and the sensors i.e., select what variable should be controlled and select how it should be measured. 2) When this is done, the manipulated-variables and the actuators are selected, 3) thereafter the control loop can be constructed and the search for an optimal set-point can start.

The task in this project is to examin if a similar procedure can be applied for ”plant-wide control-loops”.

1)   At the plant wide level it is difficult to directly measure a variable that indicates the plant performance, instead this is a variable that depends upon one or many measureable-variables. It is therefore a variable that is calculated rather than measured.  I.e. by using the data from the control systems and historical databases (originally intended for local control) more elaborated information can be obtained. For an average site with about 1000-10 000 variables and about 100-1000 control loop, it would be reasonable to calculate about 10 performance indicators. The work of defining performance indicators will partly be an answer to the industry-wide problem of having ”poor visibility into plant operations” and to start utilizing ”the hidden resource that data is known to be”. This part of the work is coordinated with the ISO 22400 standard (Key performance Indicators for manufacturing Operations Management)

2)   The second step consists of understanding what variables that should be manipulated in order to make a change in a performance indicator in a desired direction.  At the plant-wide level, the same variable often influences many performance indicators. It is therefore of importance to have an understanding of the intricate web of variables and performance indicators, sometimes refered to as a Vollmann decompositioning. With a tool that displays the variable-performance-web, the manual work of performing the plant-wide control could be facilitated and  executives could get some help in ”capturing and understanding information rapidly in order to make sound business decisions”.

3)   The third and last step consists of creating the control loops, selecting control structure and finding a good set-points. This is sometimes refered to as Closed-loop enterprise control (ref). Other related industrial/academic domains are; Enterprice Manufacturing Intelligence, real-time enterprise control, operations management. Controlling the plant or the site is seen as a craftmanship where the controller (i.e., the plant manager) has a gut feeling for controlling the plant. It would be of interest trying to set up dynamical models for the Performance indicators without taking all the details of the production into account. There could be synergies with the LCCC-area: Modeling support for Design and verification (Topic C:  Real-time simulation of  Physical systems)

More details: A more detailed description of the project is also available.