Page 459 - Design of Simple and Robust Process Plants

446 Chapter 11 Operation Optimization
layer; and Optimization layer. The basic control layer should function totally inde-
pendently of higher layers.
Design of the basic control layer should be very robust. In case of failure of a
higher layer, operation should still be able to run the process with minimal atten-
tion.
Robustness of basic control layer can be achieved by selection of the optimal pair-
ing of CVs and MVs by application of static and dynamic interaction analysis. The
selection of alternative CVs has to be evaluated for a self-optimizing control struc-
ture in case any interaction with the selected pairings is too large and would require
a model-based controller..
The design of the model-based control layer should be based on a robust BPCL,
and preferably designed on a first-principle dynamic simulation.
Model-based controllers are also designed as constraint controllers in order to
maximize benefits from process capacity capabilities and fulfil quality requirements.
These types of controllers are specifically important in combination with operation
optimization.

11.8
Operation Optimization

The applications of operation optimization are growing. Specifically, off-line applica-
tions have found wide use in the field of production scheduling for batch-type opera-
tions as well as for continuous operations to determine static optimal operation con-
ditions. Optimization of continuous process is based on first-principle models.
Closed loop static optimizations, based on nonlinear programming techniques, for
continuous processes are increasingly applied, but have still not reached their full
potential. Dynamic optimization is applied initially to optimize transient operations.
The benefits of closed-loop static operation optimization are large in the order of
percentages of the operational cost and capacity. The basic requirements are con-
straint controllers to push the process against its constraints, and robust and vali-
dated process models.
The methodology for closed-loop as well as open-loop process optimizations
needs to be followed strictly to achieve a successful project. Too many projects have
suffered from a recognized methodology not being followed.
The installation of a performance or profit meter is an important intermediate
step for optimization. The meter gives valuable actual information on process opera-
tion in money terms. It is in use to compare actual performance with recommended
performance from off-line optimization models. The performance meter is used as
a stimulation for operation, to operate closer to its target (optimized) values.
Another application is the validation of closed-loop optimization models which is
quite important as any error in the model results in a lost opportunity.
The gap between the dynamics of the control of the process and the static closed-
loop optimization, which normally runs on an hourly basis must be bridged. The
self-optimizing control structure operating at basic control level was introduced.