Sammendrag
Many large- and medium sized industry companies have a number of more or less
similar process-units for processing raw materials or producing finished
products. Three typical examples are
1.Borealis (e.g. polymer reactors for producing plastics raw materials)
2.Norske Skog, (paper machines)
3.Norsk Hydro (e.g. plants for fertilizer production)
Advanced process control has proved successful in many cases, especially in
multivariable processes with strong
coupling between inputs and outputs, and/or processes with strong
non-linearities. An industrial company which has invested, or is about to
invest, in advanced control in one of their units / factories, would more
likely benefit economically if the developed control method could be
efficiently rolled-out on similar units.
The research carried out in this project will focus on robust Model based
Predictive Control (MPC). MPC has a reported payback time of 1-12 months in
multivariable processes; which include refineries, paper machines, oil wells,
etc. The term "robust" reflects the idea that the controller should be
developed to maintain a specified
performance even though the model has a given uncertainty. Three sub-problems
will be further investigated:
1.How to efficiently develop a model, including model uncertainty, for a
process.
2.How to develop a robust MPC, for the process modeled in sub-problem 1.
3.What is required for sub-problems 1-2, so that the control method can be
efficiently rolled-out on other similar process-units.
Case study:
Paper machine 6 (PM6) at Norske Skog, Saugbrugsforeningen, will be studied in
detail. The paper machine will be modeled, and a robust MPC will be designed
for the process. Finally, as Norske Skog has many more or less similar paper
machines, the roll-out of the control method will be discussed.
Vis fullstendig beskrivelse
Vitenskapelig sammendrag
Many large- and medium sized industry companies have a number of more or less
similar process-units for processing raw materials or producing finished
products. Three typical examples are
1.Borealis (e.g. polymer reactors for producing plastics raw materials)
2.Norske Skog, (paper machines)
3.Norsk Hydro (e.g. plants for fertilizer production)
Advanced process control has proved successful in many cases, especially in
multivariable processes with strong
coupling between inputs and outputs, and/or processes with strong
non-linearities. An industrial company which has invested, or is about to
invest, in advanced control in one of their units / factories, would more
likely benefit economically if the developed control method could be
efficiently rolled-out on similar units.
The research carried out in this project will focus on robust Model based
Predictive Control (MPC). MPC has a reported payback time of 1-12 months in
multivariable processes; which include refineries, paper machines, oil wells,
etc. The term "robust" reflects the idea that the controller should be
developed to maintain a specified
performance even though the model has a given uncertainty. Three sub-problems
will be further investigated:
1.How to efficiently develop a model, including model uncertainty, for a
process.
2.How to develop a robust MPC, for the process modeled in sub-problem 1.
3.What is required for sub-problems 1-2, so that the control method can be
efficiently rolled-out on other similar process-units.
Case study:
Paper machine 6 (PM6) at Norske Skog, Saugbrugsforeningen, will be studied in
detail. The paper machine will be modeled, and a robust MPC will be designed
for the process. Finally, as Norske Skog has many more or less similar paper
machines, the roll-out of the control method will be discussed.
Vis fullstendig beskrivelse