Week in PSE

September 17, 2009

GPOPS – good open source optimal control software!

Filed under: Uncategorized — Tags: , — hdahlol @ 9:25 pm

Finally – I’ve found a good open source software for solution of optimal control problems. It is a Matlab toolbox for the Gaussian pseudo-spectral optimal control (a collocation method). Take a look at it at http://www.gpops.org/. And — wohoo! — it is even well documented!


June 28, 2009

New plants in high-cost countries – not history after all?

Filed under: Uncategorized — hdahlol @ 8:37 am

The French company SNF Floerger is building a new polymer plant in Lousiana, USA. The plant has an estimated cost of $ 362 million. The plant will produce polymers that are used in oil wells to increase recovery.

Not many new plants are built in western countries nowadays. It is nice to see that some belive that some companies will still invest in mature economies, in spite of the financial crisis and the higher cost level in such countries. In spite of the high cost of installing a new production facility in our part of the world, SNF Floerger seems to believe that positive returns can be expected in the special-purpose polymer market.

June 17, 2009

Reading tip!

Filed under: Uncategorized — hdahlol @ 11:22 am

The text book Product and Process Design Principles by Seider, Seader, Lewin and Widagd landed in my shelf at work yesterday. I intended to read a chapter on batch process design and control – but ended up reading almost the entire book in one session! It is a refreshing approach to teaching product and process desgin at once – with emphasis on team work and innovation processes. The book includes a large selection of case studies, including components for the iPhone and design of the ultimate espresso machine. Recommended reading!

PS. I normally adhore text books. This was a very good exception!

June 1, 2009

Simulation with extensive variables to download!

Filed under: Uncategorized — Tags: , , , — hdahlol @ 7:55 pm

Download the file distcolsim.py from the box.net widget! Requires matplotlib to plot results. Enjoy 🙂

May 30, 2009

Modeling startup

Filed under: Uncategorized — hdahlol @ 9:15 am

Often when simulating a process, you need to initialize it by filling up tanks and so on. Take for example a buffer tank with two feed streams (1 and 2) and one effluent stream (3). How do you get the concentration dynamics correct at startup? If you try to model it in the usual form with intensive variables for concentration such as mole fraction, the initial condition is singular. The same goes for concentration in moles per liter;

n/N goes to infinity when N goes to zero
n/V goes to infinity when V goes to zero

So…. how do you fix that? You write your balance equaitons in extensive variables such as the number of moles in the tank. For a two-component system we write the total balance (N) and on component A (nA)

dN/dt = F1 + F2 – F3
dnA/dt = F1_A + F2_A – F3_A

This way there will be no singularity problems!

May 27, 2009

Educational note on observers and RHP zeros

Filed under: Uncategorized — hdahlol @ 2:39 pm

A few weeks ago I created a small example on using estimation of missing measurements in systems with right-half-plane zeros close to right-half-plane poles for a group seminar at the university. If you are trying to use measurements where the bandwidth is severely limited to estimate other measurements that for some reason are not avilable, and then use these estimated values to stabilize a plant – you will fail. The reason is that the fundamental limitation in the input-output map from excitation signal to measurement used for estimation is a fundamental limitation that does not go away. Sometimes this can be easy to forget when working in the time-domain; certain effects are simply more visible in the frequency domain. Remember

Limited bandwidth + limited actuator action = instability!

For details, see: http://www.nt.ntnu.no/users/skoge/book/typo/lqg-rhp-example/
(including Simulink files and a word document describing the whole thing).

May 26, 2009

Proposal to present sent to AIChE

Filed under: Uncategorized — hdahlol @ 12:52 pm

The AIChE annual meeting is in Nashville, November 2009. I’ve sent a proposal to present on a method for using linear combinations of measurements to obtain optimal control of batch processes. I am quite excited about the method as it is theoretically interesting and seems to have potential in practical applications too 🙂 The abstract is at http://aiche.confex.com/aiche/2009/webprogrampreliminary/Paper160628.html.

May 21, 2009

Process systems engineering blog

Filed under: Uncategorized — hdahlol @ 5:48 am

Hello. This is my “professional” blog where I will write things that I notice in my field; process systems engineering. I am a graduate student / research fellow at the Norwegian University of Science and Technology, working on a PhD dissertation in optimal control. My research has focus on optimal operation of chemical processes that are designed to operate under unsteady-state conditions. I’ll try to update this blog at least one time per week. Stay tuned!

« Newer Posts

Blog at WordPress.com.