1. Self-optimizing control of batch distillation

Selvoptimaliserende regulering av satsvis destillasjon

Medveileder Håkon Dahl-Olsen

Oppgaven vil gå ut på å sette opp en dynamisk modell av en satsvis destillasjonsprosess og bruke tilgjengelig teori til å designe et selvoptimaliserende reguleringssystem for denne kolonnen. Ytelsen av reguleringssystemet skal vurderes ved å se på avvik i objektfunksjonen. Et av følgende mål kan velges for driften av prosessen:

1.      Maks produkt med gitt minimumsrenhet ved fast driftstid per batch

2.      Minimum tid for gitt renhet og mengde (absolutt eller gjenvinningsgrad)

Naturlige beregningsverktøy vil være gProms og Matlab. Oppgaven krever interesse for optimalisering og numeriske beregninger. Se hjemmeside S. Skogestad for mer informasjon

2. Diabetes control.

Reserved for Christian Ellingsen. To be performed at UC Santa Barbara with Frank Doyle and Dale Seborg as local supervisors.

3. Dynamic Model and Control of Brobekk incineration plant

Oppgaven er reservert Helge Smedsrud. Medveiledere: Johannes Jäschke og Helge Mort (Prediktor) 

Oppgavebeskrivelse:

1.     Videreutvikle eksisterende Simulink-modell til å ta hensyn til trykk.

2.     Bestemme optimale driftsparametere for ulike driftsforhold (sommer/vinter, dag/natt).

3.     Utvikle en optimal reguleringsstruktur for anlegget basert på minst mulig fysiske endringer i forhold til det virkelige anlegget.

4.     Utvikle en ideell reguleringsstruktur for anlegget.

5.     Vurdere, og eventuelt implementere, MPC som tillegg til reguleringsstrukturen i punkt 3. og/eller 4.


4. Use of dynamic degrees of freedom for tighter bottleneck control and maximum throughput

Reserved for Theogene, Uwarwema 

Background:
Maximizing throughput in a network is a common problem in several settings
(Phillips et al.,1976; Ahuja et al., 1993). From network theory, the
max-flow min-cut theorem states that the maximum throughput in a plant
(network) is limited by the ”bottleneck” of the network. In order to
maximize the throughput, the flow through the bottleneck should be at its
maximum flow. In particular, if the actual flow at the bottleneck is not at
its maximum at any given time, then this gives a loss in production which
can never be recovered (sometimes referred to as a ”lost opportunity”).
The back off is an unavoidable “safety factor” because perfect dynamic
control is not possible. Choosing a small back off, improves the profit
(throughput) but increases the risk of not being able to have feasible
operation (satisfying constraints) when a large disturbance occurs. The
necessary back off can generally be reduced by improving the control around
the bottleneck unit. “Improved control” is usually obtained by retuning the
loops to obtain smaller variation.

Project proposal:
To obtain tighter bottleneck control, dynamic degrees of freedom like
hold-up volumes can be used, and hence reduce the back off. In this project
we want to simulate a more realitic process like distillation columns in
series and using the sump volumes in the columns as dynamic degrees of
freedom for tighter bottleneck control. Two apporaches for control is
considered:
1) Adding bias directly to the level controlles upstream the bottleneck
2) Using MPC (like SEPTIC) to manipulate on the level control set points. 
The two approaches should be discussed and also compared.
5. Dynamic simulation for improved operation of the Snøhvit CO2 removal section.

Reserved for Jalal Fahadi [

StatoilHydro has developed a d-spice dynamic process simulation model of the CO2 removal section at Snøhvit. It is known that the the amine absorbtion process is simplified modelled in d-spice, and that the real process responds in a different way than the d-spice model with regards to temperature changes and the amount of MEG present in the amine solution. The d-spice model is used to train operators, develop MPC applications and solve operational challenges.

Proposed project tasksl:

1) Perform model updating and verification of the d-spice model against process data and labdata using eg OPC connection and matlab
2) Develop estimators of unmeasured process parameters like MEG content in the amine using Septic
3) Use of dynamic simulation to solve operational challenges

Medveileder, StatoilHydro: Ingvild Løvik Sperle


6.  Dynamic modelling and control of off-shore process, including water treatment, with emphasis on minimzing emissions

Dynamisk modellering og regulering av off-shore prosessanlegg, inklusive vannbehandling, med vekt på utslipssredulsjoner.

Reservert for Tone S. Pettersen.  I samarbeid med ABB (Espen Storkaas) og muligens StatoilHydsro. Medveileder: Henrik Manum

7. Deriving fast distillation models 
Co-supervisor Andreas Linhart

Distillation is the most important separation technology today. A trend in
controlling distillation columns economically efficient is going toward using
model predictive control (MPC) algorithms, which calculate an optimal input
trajectory to the plant based on repeated simulations of a model of the process
to predict the future behaviour of the plant with respect to disturbances and
control inputs. Since the controllers operate in real-time, very fast process
simulations are needed. In case of distillation, full models are usually too
complex to be simulated sufficiently fast. Therefore, methods to derive reduced
models are of high interest to both industry and research community.
Our approach to derive a reduced model is based on tray aggregation, which means
that the slow dynamics of a number of consecutive stages in the distillation column are
approximated by a large "aggregation tray", and the fast dynamics are approximated
by employing a quasi-steady-state assumption. These reduced models provide a good
gain in computational performance while being sufficiently accurate to be used
in an MPC. While the procedure to derive this models is straightforward, there are
many structural and implementation degrees of freedom which can be used to further
improve the performance of the method. In this work, several important issues
can be addressed. The reduced model contains a large number of algebraic equations,
which can be solved off-line and stored in an appropriate way. The currently used
method of look-up tables should be refined and improved. Furthermore, the
method needs to be extended to models with more than two components. For the reduced
models, optimal dynamic tray positions and sizes have to be chosen. Special attention
should herewith be paid to the interaction of the reduced model with the base-layer
controllers. Finally, the efficient use of the reduced model in an non-linear MPC application
is to be investigated.

8. Andre oppgaver kan være aktuelle, gjerne i samarbeid med industri eller andre institusjoner


 onesejn@stud.ntnu.no] Tuning av basisregulering for offshore-prosesser (medveileder: Henrik Manum, medveilder ABB: Espen Storkaas) 

-- 
************************************************************************
Sigurd Skogestad, Professor and Head of Department  Phone: +47-7359-4154
Department of Chemical Engineering                  Home:  +47-7390-2625
Norwegian Univ. of Science and Technology (NTNU)    Fax:   +47-7359-4080
N-7491 Trondheim, Norway                            Mobile:+47-9137-1669
http://www.nt.ntnu.no/users/skoge                   email: skoge@ntnu.no        
************************************************************************