Project thesis topics 2010

The text below is a brief description of the topics. Contact Lars Imsland for further information. Most topics can be continued in a Master if the candidate so wishes.

1. Dynamic positioning systems in arctic environments

One or more assignments is available in connection with a research project on development of technology for dynamic positioning in arctic environments. Possible tasks are simulator development, observer and control design for vessels in ice, observer design for ice behavior, control of unmanned vehicles for use with ice management, supervisory control of arctic marine operations, and other control engineering tasks related to Arctic DP.

Co-advisor: Professor Roger Skjetne, IVT (IMT)

2. Modellering og regulering av boreslamprosess

Automatisering av boreprosessen er høyt prioritert av oljeselskaper. I dag er det en mudingeniør som blander boreslam, tar labprøver og analyserer returstrøm. Oppgaven går ut på å vurdere hvordan dette kan automatiseres. Tre parametere som er spesielt viktig er tetthet, viskositet og rate. Trykket nede i brønnen må holdes lavt nok for å unngå tap av boreslam ut til formasjonen og høyt nok til å unngå innstrømning av hydrokarboner mens man borer. Boreslammet skal transportere ut faste partikler (cuttings). Man kan anta at viskositet og tetthet for boreslam både inn og ut av brønnen måles. Første del av oppgaven går på modellere slamsystemet med brønn og tanker topside med tanke på automatisering. En slik modell må kunne gjenskape de forsinkelser som ligger i systemet. Andre del av oppgaven går ut på å designe en regulator som automatisk justerer boreparametrene basert på en modell og tilbakekobling fra målinger. Regulatoren skal testes med simuleringer. En mulig løsning er MPC.

Medveiledere: John-Morten Godhavn, Statoil Forskningssenter, Rotvoll og Gerhard Nygaard, IRIS, Bergen.

3. Advanced Process Control of an Oxygen-Membrane-based Gas Turbine Power Plant

The gas turbine power plant shown in Figure 1 shall be analyzed in terms of potential start-up and shut-down procedures using feedback control and/or model predictive control methods. Operational as well as material constraints of critical process components need to be considered.

Co-advisor: Post.Doc. Konrad Eichhorn, IVT (EPT)

1. Techniques for efficient covariance propagation in the Extended Kalman Filter based on model reduction

Even though the (recursive) EKF is considered an efficient algorithm for state estimation, it can still be computationally expensive for large/complex systems. The bulk of the complexity stems from the propagation of the state covariance matrix. In this task, the student will consider using techniques from model reduction to make the covariance propagation more efficient, while the full model still is used for state propagation. This task fits a student that is mathematically inclined, and also interested in numerical algorithms.

2. Implementation and comparison of methods for sensitivity integration for gradient calculation in non-linear MPC

Practical implementation of analytical gradient-computation for 'single-shooting' non-linear MPC is usually based on sensitivity integration. The candidate should implement/compare (at least) two different methods for sensitivity integration (BDF and the algorithm of Schlegel et al (2004)), and assess their usefulness and appropriateness for different types of single-shooting NMPC problems.

3. Parallelization in non-linear MPC

Some types of implementation of nonlinear MPC can benefit from parallelization. The candidate should explore strategies for parallelization, and make prototype implementations for assessment and comparison.

4. Implementation of framework for non-linear MPC based on open-source components

The candidate should investigate open-source components that fit into a framework for non-linear MPC. Relevant components might be for modeling, model code-generation, and optimization.