From skoge Wed Jun  3 11:23:41 1992
To: gundersen%fs.decnet@bg.hbg.hydro.no, hovd,morud,lundstrm,mathisen,sorensen,wittgens,wolff,zhao,flatby,mejdell,jacobsen,
    lien, morari@macpost.caltech.edu, naess%fs.decnet@bg.hbg.hydro.no,
    saelid%fs.decnet@bg.hbg.hydro.no, thz
Subject: ESCAPE-1 


Short report from ESCAPE-1, Helsingor, DK, May 24+28, 1992.
###########################################################

European Symp. on Computer-Aided Process Engineering (ESCAPE).

This was a pretty good meeting technically, and the people I talked to
agreed that it was better than the previous meeting in Barcelona.

The organization and location was also very good, although the
registration fee was too high. 

The proceedings are available as a special issue of Computers and
Chemical Engineering, and some of the papers will be selected to appear
in extended versions in another issue. 

There were a large group of people from Norway (Norsk Hydro and NTH),
and it was noticeable that there were almost none from Sweden (except
for Urban Gren who came to chair one session).

The ESCAPE-meeting are supposed to be once every year. This year there
is an abnormality with the ESCAPE-2 meeting in France in October.
This meeting has no control sessions, and was "forced" upon the CAPE working 
party by the French (we have no papers there). The next regular meeting
is in Graz, Austria 5-7 July 1993, with Nov. 1 1992 as the deadline for
submitting abstracts. 

The ESCAPE-1 meeting had some good plenary talks. One was by Roger
Benson, ICI who accused the chemical engineering community or rather the
process industry for not applying new technology, and predicted the 90's
to be the decade of change. Steinar Saelid from Norsk Hydro also gave a
plenary talk on their operator support system for the Glomfjord
Fertilizer plant. Other plenaries were by M. Michelsen, DTH, W. Seider,
Pennsylvania, L. Biegler, CMU, and J. Perkins, Imperial College.
Unfortunately, M. Kubicek did not come as was announced.

Here are a few comments/notes of particular interest for our group:

1. Warred Seider presented a biochemical example which by the way was
heavily criticized by the bio experts. He claimed that one may operate a
bioreactor at an operating point which is both unstable and has change
in the sign of the steady-state gain. We should do a linear
controllability analysis of his example as an exercise (Zi-Quin or
Zhao).

2. H.A.Kooijman (worked with Ross Taylor, Clarkson) presented a good
paper on noneq-distillation which may be important for mixtures other
than binary (for binary one may as well use Murphee efficiency or
overall tray efficiency, Wolfgang Marquardt promised to send proof).

3. John Perkins gave a good overview of integrated design. Something is
wrong in the way we design plants, when the first thing that is done on
a new plant is to redesign the control system.

4. I talked to some UMIST people. (Vikas Dhole and Richard Camm from
Linnhoffs group). The said they were interested in a
cooperation and indicated that they might invite me to there consortium
meeting in October.

5. Ian Cameron gave a good lecture on the importance of model structures.
There is a literature review of structural things on p. 183 in the
Supplementary volume (Knut, Erik) although there does not seem to be
much new compared to Manfreds paper (part I, 1980). 
Cameron talked about the evaporator example that may be good for Erik Wolff to look at (we have the book).

6. I talked to Angelo Lucia, and told him that Elling Jacobsen had new results
that seem to be a counterexample for the uniqueness for the Qd-Qb (duties)
specification. Lucia did not believe this, and said he had a new proof
for the binary case with much fewer assumptions than before, that proved
this. 

7. My talk went reasonable good, except that I only got 18 min, and I
had to skip quite a lot of frequency-plots etc. I said
essentially everything 	 wanted, but unfortunately the audience may have
got the impression that this was not the case. 

8. In the talk I tried to stress the different levels in the control
hierarchy.  Actually, a good general way of looking at it is that the
higher levels sends down commands (setpoints), and receives feedback
from the level below about possible problems (e.g., inputs reaching
constraints). 

9. Colin McGreqvy from Leeds gave a good talk on control of cooled
tubular reactors (John M.!).

There were a number of other interesting things, but I expect people
from Kristian Lien or Terje Hertzbergs group to mention these. 

Sigurd Skogestad.