Page 366 - Design of Simple and Robust Process Plants

9.3 General Operation Optimization of Continuous Processes 353

Maximize
Simulated performance
Optimizer
Prices

Flows

Process
model
Feed flows
Product flows
Energy flows

Fig. 9.1. Process oprimization reflected in a control diagram.

Manual implementation of adapted process conditions is executed based on off-line
optimization results. Manual updating is normally done on a daily basis, or at any
significant change such as processing of another feed as crude in a refinery. The
number of degrees of freedom (DOF)in general is limited in an off-line optimiza-
tion. During the implementation of off-line optimization conditions, the process is
normally not running close to its constraints, unless constraint controllers are
installed. Operators prefer to stay away from constraints to avoid all kinds of inter-
acting actions
Automatic implementation of process conditions is based on continuously run-
ning optimizations, which receive ongoing updated circumstantial information; this
is called CLO. Operational conditions are however only changed in discrete steps.
The discrete implementation steps are the result of: the time it takes to execute an
optimization cycle; and the variation of the circumstantial conditions which often is
based on information which becomes available at discrete times, for example price
sets.
OO as a whole needs to be seen in a hierarchical control structure, as reflected in
Figure 8.1 in Chapter 8. As a basis, there must be DOF available for manipulation,
to achieve a more economic operation.
A CLO cycle must follow a stepped sequence that is enforced through an execu-
tive (Figure 9.2). The sequential steps are:
. data analysis;
. data reconciliation;
. parameter estimation;
. optimization; and
. implementation of new set-points.
Before a step is terminated and the next step is activated, a set of decision criteria
must be met. The individual steps will be described in detail in the following para-
graphs.