09Jan2007

From: Sigurd Skogestad
To: Antonio Araujo

In mode II the loss with the implementation error of 1-4 for xmix_met is 5.33 k$/y (according to Table 13).

Is there something wrong here (in Table 13)? 
I cannot understand why the loss for the implementation error in xmix_met is
independent of what the other controlled variable is (see structures I-X in Table 13 in part I).

(As an example, think of controlling two temperatures in a distillation column. If we select to control two neighborinbg temperatures in the column,
then the effect of implementation error on the loss (composition deviation) is much larger than if they are far apart.)

Reply:

Hello again,

I see your point and agree with you. However, I found nothing wrong with the calculated values in Table 13 in part I. Maybe because the implementation error was chosen so small that the computation didn't catch the trifling change. Anyway, now is to late to change part I for it's been published. Next time, we must increase the implementation error assumption.

Cheers,
Antonio.