Page 389 - Design of Simple and Robust Process Plants

376 Chapter 9 Operation Optimization
the hold-up and integrates over time. The parameter step is operable if decider 1 and
decider 2 are true. The optimization step is operable if deciders 1, 2, and 3 are true.
The status of the different steps must be presented at the operator interface that
requires the following information:
. A step flow diagram with indication of the activated steps.
. The last and recent released information from each step.
. Cumulative information of each step.
. An overview of the recent downloaded setpoints from the optimization in
relation to the actual set-points and its measurement.
. An overview of the hierarchy of the control regarding the set-point download-
ing. The signal transmission from the model-based controller to the basic
control layer and from the optimization layer to the model-based layer and
the basic control layer. The overview needs to include the position of the con-
trol loops (closed or open).

Summary
Closed loop optimization is not a trivial effort, they are differentiated in several
methods: self-optimizing control or sub-optimization, empirical optimization, and
fundamental model based control. These methods are summarized as:

. Self-optimizing control, a unit optimization technique applied as a subopti-
mization is discussed with, as an example, a divided wall column. This is a
promising, but quite new technique.
. Self-optimizing control is simple feedback system where the controlled vari-
able has been selected based on steady-state analysis to obtain a system
where the optimal value is:
± only weakly dependent on disturbances;
± sensitive to changes in independent variables; and
± easy to control accurately.
. Empirical optimization for processes with a few units and limited DOFs is
presented for process where fundamental models are not available. The
development of empirical models based on an experimental design applied
to the existing process can only be successful if the process is adequately
stable. Implementation might be off-line, but with quality loops closed. This
technique has not found wide application although some parts of a funda-
mental model as applied for optimization may en-capsule an empirical
model.
. Fundamental model-based optimization is the most widely used approach, as
it brings process understanding that definitely leads to process improve-
ments next to an optimized operation. Crucial to the development of the opti-
mization models are: the reactor and unit operation models all written for an
equation-based simulator, an accurate description of the constraints, its accu-
racy and robustness.